{ "best_global_step": null, "best_metric": null, "best_model_checkpoint": null, "epoch": 100.0, "eval_steps": 20000, "global_step": 309400, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.03232062055591468, "grad_norm": 63.80897521972656, "learning_rate": 9.900000000000002e-06, "loss": 5.0757, "step": 100 }, { "epoch": 0.06464124111182935, "grad_norm": 51.32004165649414, "learning_rate": 1.9900000000000003e-05, "loss": 4.3807, "step": 200 }, { "epoch": 0.09696186166774402, "grad_norm": 9.607514381408691, "learning_rate": 2.9900000000000002e-05, "loss": 4.1432, "step": 300 }, { "epoch": 0.1292824822236587, "grad_norm": 7.153324127197266, "learning_rate": 3.99e-05, "loss": 4.0461, "step": 400 }, { "epoch": 0.16160310277957338, "grad_norm": 3.7312490940093994, "learning_rate": 4.99e-05, "loss": 3.9927, "step": 500 }, { "epoch": 0.19392372333548805, "grad_norm": 2.4463391304016113, "learning_rate": 5.9900000000000006e-05, "loss": 3.9632, "step": 600 }, { "epoch": 0.22624434389140272, "grad_norm": 3.8260347843170166, "learning_rate": 6.99e-05, "loss": 3.9379, "step": 700 }, { "epoch": 0.2585649644473174, "grad_norm": 5.175188064575195, "learning_rate": 7.99e-05, "loss": 3.9001, "step": 800 }, { "epoch": 0.2908855850032321, "grad_norm": 13.827736854553223, "learning_rate": 8.989999999999999e-05, "loss": 3.8392, "step": 900 }, { "epoch": 0.32320620555914675, "grad_norm": 2.378403425216675, "learning_rate": 9.99e-05, "loss": 3.8368, "step": 1000 }, { "epoch": 0.3555268261150614, "grad_norm": 4.115757942199707, "learning_rate": 0.0001099, "loss": 3.8114, "step": 1100 }, { "epoch": 0.3878474466709761, "grad_norm": 2.430039882659912, "learning_rate": 0.00011990000000000001, "loss": 3.8009, "step": 1200 }, { "epoch": 0.42016806722689076, "grad_norm": 3.2554678916931152, "learning_rate": 0.00012989999999999999, "loss": 3.7745, "step": 1300 }, { "epoch": 0.45248868778280543, "grad_norm": 1.9953078031539917, "learning_rate": 0.0001399, "loss": 3.7577, "step": 1400 }, { "epoch": 0.4848093083387201, "grad_norm": 2.4289982318878174, "learning_rate": 0.0001499, "loss": 3.72, "step": 1500 }, { "epoch": 0.5171299288946348, "grad_norm": 2.354421377182007, "learning_rate": 0.00015989999999999998, "loss": 3.7104, "step": 1600 }, { "epoch": 0.5494505494505495, "grad_norm": 2.225425958633423, "learning_rate": 0.0001699, "loss": 3.7334, "step": 1700 }, { "epoch": 0.5817711700064642, "grad_norm": 7.562653541564941, "learning_rate": 0.0001799, "loss": 3.6986, "step": 1800 }, { "epoch": 0.6140917905623788, "grad_norm": 2.0109822750091553, "learning_rate": 0.0001899, "loss": 3.7064, "step": 1900 }, { "epoch": 0.6464124111182935, "grad_norm": 2.309058904647827, "learning_rate": 0.0001999, "loss": 3.6754, "step": 2000 }, { "epoch": 0.6787330316742082, "grad_norm": 2.2192893028259277, "learning_rate": 0.0002099, "loss": 3.668, "step": 2100 }, { "epoch": 0.7110536522301228, "grad_norm": 2.3891031742095947, "learning_rate": 0.0002199, "loss": 3.6532, "step": 2200 }, { "epoch": 0.7433742727860375, "grad_norm": 2.036247491836548, "learning_rate": 0.0002299, "loss": 3.6497, "step": 2300 }, { "epoch": 0.7756948933419522, "grad_norm": 2.8763175010681152, "learning_rate": 0.0002399, "loss": 3.6293, "step": 2400 }, { "epoch": 0.8080155138978669, "grad_norm": 2.384962558746338, "learning_rate": 0.0002499, "loss": 3.6139, "step": 2500 }, { "epoch": 0.8403361344537815, "grad_norm": 2.3743388652801514, "learning_rate": 0.00025990000000000003, "loss": 3.6356, "step": 2600 }, { "epoch": 0.8726567550096962, "grad_norm": 2.2165491580963135, "learning_rate": 0.0002699, "loss": 3.5882, "step": 2700 }, { "epoch": 0.9049773755656109, "grad_norm": 2.483201503753662, "learning_rate": 0.0002799, "loss": 3.5897, "step": 2800 }, { "epoch": 0.9372979961215255, "grad_norm": 2.3332512378692627, "learning_rate": 0.0002899, "loss": 3.5775, "step": 2900 }, { "epoch": 0.9696186166774402, "grad_norm": 2.368666887283325, "learning_rate": 0.0002999, "loss": 3.5657, "step": 3000 }, { "epoch": 1.0019392372333549, "grad_norm": 1.8092663288116455, "learning_rate": 0.0003099, "loss": 3.578, "step": 3100 }, { "epoch": 1.0342598577892697, "grad_norm": 1.7232047319412231, "learning_rate": 0.0003199, "loss": 3.512, "step": 3200 }, { "epoch": 1.0665804783451842, "grad_norm": 1.9123388528823853, "learning_rate": 0.00032990000000000005, "loss": 3.5244, "step": 3300 }, { "epoch": 1.098901098901099, "grad_norm": 1.1431645154953003, "learning_rate": 0.00033989999999999997, "loss": 3.5241, "step": 3400 }, { "epoch": 1.1312217194570136, "grad_norm": 1.532568335533142, "learning_rate": 0.0003499, "loss": 3.4954, "step": 3500 }, { "epoch": 1.1635423400129283, "grad_norm": 1.2241241931915283, "learning_rate": 0.0003599, "loss": 3.5262, "step": 3600 }, { "epoch": 1.195862960568843, "grad_norm": 1.7371819019317627, "learning_rate": 0.0003699, "loss": 3.5145, "step": 3700 }, { "epoch": 1.2281835811247577, "grad_norm": 1.4901371002197266, "learning_rate": 0.0003799, "loss": 3.5058, "step": 3800 }, { "epoch": 1.2605042016806722, "grad_norm": 1.6141966581344604, "learning_rate": 0.00038990000000000004, "loss": 3.5136, "step": 3900 }, { "epoch": 1.292824822236587, "grad_norm": 1.7414507865905762, "learning_rate": 0.00039989999999999996, "loss": 3.4896, "step": 4000 }, { "epoch": 1.3251454427925016, "grad_norm": 2.876239538192749, "learning_rate": 0.0004099, "loss": 3.479, "step": 4100 }, { "epoch": 1.3574660633484164, "grad_norm": 1.535861849784851, "learning_rate": 0.0004199, "loss": 3.4806, "step": 4200 }, { "epoch": 1.389786683904331, "grad_norm": 1.4972548484802246, "learning_rate": 0.0004299, "loss": 3.4871, "step": 4300 }, { "epoch": 1.4221073044602457, "grad_norm": 2.6041390895843506, "learning_rate": 0.0004399, "loss": 3.4717, "step": 4400 }, { "epoch": 1.4544279250161603, "grad_norm": 1.3674488067626953, "learning_rate": 0.00044990000000000004, "loss": 3.4634, "step": 4500 }, { "epoch": 1.486748545572075, "grad_norm": 1.598819375038147, "learning_rate": 0.0004599, "loss": 3.4396, "step": 4600 }, { "epoch": 1.5190691661279896, "grad_norm": 1.1317894458770752, "learning_rate": 0.0004699, "loss": 3.4456, "step": 4700 }, { "epoch": 1.5513897866839044, "grad_norm": 1.3716052770614624, "learning_rate": 0.0004799, "loss": 3.455, "step": 4800 }, { "epoch": 1.5837104072398192, "grad_norm": 1.099937915802002, "learning_rate": 0.0004899, "loss": 3.4444, "step": 4900 }, { "epoch": 1.6160310277957337, "grad_norm": 1.3480145931243896, "learning_rate": 0.0004999000000000001, "loss": 3.4497, "step": 5000 }, { "epoch": 1.6483516483516483, "grad_norm": 1.1580358743667603, "learning_rate": 0.0005099, "loss": 3.4217, "step": 5100 }, { "epoch": 1.680672268907563, "grad_norm": 1.1731674671173096, "learning_rate": 0.0005199, "loss": 3.4201, "step": 5200 }, { "epoch": 1.7129928894634778, "grad_norm": 1.1458629369735718, "learning_rate": 0.0005299, "loss": 3.4209, "step": 5300 }, { "epoch": 1.7453135100193924, "grad_norm": 1.2101383209228516, "learning_rate": 0.0005399000000000001, "loss": 3.4244, "step": 5400 }, { "epoch": 1.777634130575307, "grad_norm": 1.1774795055389404, "learning_rate": 0.0005499000000000001, "loss": 3.4151, "step": 5500 }, { "epoch": 1.8099547511312217, "grad_norm": 1.2488919496536255, "learning_rate": 0.0005599, "loss": 3.4146, "step": 5600 }, { "epoch": 1.8422753716871365, "grad_norm": 1.0765422582626343, "learning_rate": 0.0005698999999999999, "loss": 3.4276, "step": 5700 }, { "epoch": 1.874595992243051, "grad_norm": 1.0370737314224243, "learning_rate": 0.0005799, "loss": 3.3888, "step": 5800 }, { "epoch": 1.9069166127989656, "grad_norm": 1.0570347309112549, "learning_rate": 0.0005899, "loss": 3.3894, "step": 5900 }, { "epoch": 1.9392372333548804, "grad_norm": 1.089784860610962, "learning_rate": 0.0005999, "loss": 3.3823, "step": 6000 }, { "epoch": 1.9715578539107952, "grad_norm": 1.1070977449417114, "learning_rate": 0.0006099, "loss": 3.3782, "step": 6100 }, { "epoch": 2.0038784744667097, "grad_norm": 0.9683001041412354, "learning_rate": 0.0006199, "loss": 3.3913, "step": 6200 }, { "epoch": 2.0361990950226243, "grad_norm": 0.8402898907661438, "learning_rate": 0.0006299000000000001, "loss": 3.3484, "step": 6300 }, { "epoch": 2.0685197155785393, "grad_norm": 0.8574873208999634, "learning_rate": 0.0006399, "loss": 3.3462, "step": 6400 }, { "epoch": 2.100840336134454, "grad_norm": 0.733184278011322, "learning_rate": 0.0006499, "loss": 3.3413, "step": 6500 }, { "epoch": 2.1331609566903684, "grad_norm": 1.1950469017028809, "learning_rate": 0.0006599, "loss": 3.348, "step": 6600 }, { "epoch": 2.165481577246283, "grad_norm": 0.8092224597930908, "learning_rate": 0.0006699000000000001, "loss": 3.349, "step": 6700 }, { "epoch": 2.197802197802198, "grad_norm": 1.2157089710235596, "learning_rate": 0.0006799, "loss": 3.334, "step": 6800 }, { "epoch": 2.2301228183581125, "grad_norm": 0.9567965865135193, "learning_rate": 0.0006899, "loss": 3.3301, "step": 6900 }, { "epoch": 2.262443438914027, "grad_norm": 0.7585746049880981, "learning_rate": 0.0006998999999999999, "loss": 3.3377, "step": 7000 }, { "epoch": 2.2947640594699417, "grad_norm": 1.1528912782669067, "learning_rate": 0.0007099, "loss": 3.3225, "step": 7100 }, { "epoch": 2.3270846800258567, "grad_norm": 1.0700244903564453, "learning_rate": 0.0007199, "loss": 3.3434, "step": 7200 }, { "epoch": 2.3594053005817712, "grad_norm": 1.1488701105117798, "learning_rate": 0.0007299, "loss": 3.3283, "step": 7300 }, { "epoch": 2.391725921137686, "grad_norm": 1.4043810367584229, "learning_rate": 0.0007399, "loss": 3.3344, "step": 7400 }, { "epoch": 2.4240465416936003, "grad_norm": 1.1909414529800415, "learning_rate": 0.0007499000000000001, "loss": 3.3465, "step": 7500 }, { "epoch": 2.4563671622495153, "grad_norm": 0.8690209984779358, "learning_rate": 0.0007599, "loss": 3.3478, "step": 7600 }, { "epoch": 2.48868778280543, "grad_norm": 0.767558217048645, "learning_rate": 0.0007699, "loss": 3.3331, "step": 7700 }, { "epoch": 2.5210084033613445, "grad_norm": 0.9267312288284302, "learning_rate": 0.0007799, "loss": 3.3127, "step": 7800 }, { "epoch": 2.553329023917259, "grad_norm": 1.0323102474212646, "learning_rate": 0.0007899000000000001, "loss": 3.3026, "step": 7900 }, { "epoch": 2.585649644473174, "grad_norm": 0.7380838990211487, "learning_rate": 0.0007999000000000001, "loss": 3.3376, "step": 8000 }, { "epoch": 2.6179702650290886, "grad_norm": 0.9032792448997498, "learning_rate": 0.0008099, "loss": 3.3435, "step": 8100 }, { "epoch": 2.650290885585003, "grad_norm": 0.9353654980659485, "learning_rate": 0.0008198999999999999, "loss": 3.3104, "step": 8200 }, { "epoch": 2.682611506140918, "grad_norm": 0.8816394209861755, "learning_rate": 0.0008299, "loss": 3.335, "step": 8300 }, { "epoch": 2.7149321266968327, "grad_norm": 0.8619208931922913, "learning_rate": 0.0008399, "loss": 3.3452, "step": 8400 }, { "epoch": 2.7472527472527473, "grad_norm": 0.9694051146507263, "learning_rate": 0.0008499, "loss": 3.3382, "step": 8500 }, { "epoch": 2.779573367808662, "grad_norm": 1.0988004207611084, "learning_rate": 0.0008599, "loss": 3.329, "step": 8600 }, { "epoch": 2.8118939883645764, "grad_norm": 1.3006855249404907, "learning_rate": 0.0008699000000000001, "loss": 3.3152, "step": 8700 }, { "epoch": 2.8442146089204914, "grad_norm": 0.9456707239151001, "learning_rate": 0.0008799000000000001, "loss": 3.3295, "step": 8800 }, { "epoch": 2.876535229476406, "grad_norm": 1.0091285705566406, "learning_rate": 0.0008899, "loss": 3.3332, "step": 8900 }, { "epoch": 2.9088558500323205, "grad_norm": 1.110554814338684, "learning_rate": 0.0008999, "loss": 3.3078, "step": 9000 }, { "epoch": 2.9411764705882355, "grad_norm": 1.009962558746338, "learning_rate": 0.0009099, "loss": 3.3246, "step": 9100 }, { "epoch": 2.97349709114415, "grad_norm": 0.8776062726974487, "learning_rate": 0.0009199000000000001, "loss": 3.3259, "step": 9200 }, { "epoch": 3.0058177117000646, "grad_norm": 1.2100518941879272, "learning_rate": 0.0009299, "loss": 3.3019, "step": 9300 }, { "epoch": 3.038138332255979, "grad_norm": 1.0079325437545776, "learning_rate": 0.0009399, "loss": 3.1978, "step": 9400 }, { "epoch": 3.070458952811894, "grad_norm": 1.0548369884490967, "learning_rate": 0.0009498999999999999, "loss": 3.2165, "step": 9500 }, { "epoch": 3.1027795733678087, "grad_norm": 0.9353946447372437, "learning_rate": 0.0009599, "loss": 3.2126, "step": 9600 }, { "epoch": 3.1351001939237233, "grad_norm": 1.093759536743164, "learning_rate": 0.0009699, "loss": 3.2195, "step": 9700 }, { "epoch": 3.167420814479638, "grad_norm": 1.34163236618042, "learning_rate": 0.0009799, "loss": 3.2211, "step": 9800 }, { "epoch": 3.199741435035553, "grad_norm": 0.8957286477088928, "learning_rate": 0.0009899, "loss": 3.1988, "step": 9900 }, { "epoch": 3.2320620555914674, "grad_norm": 0.7971066236495972, "learning_rate": 0.0009999, "loss": 3.2276, "step": 10000 }, { "epoch": 3.264382676147382, "grad_norm": 0.6855546832084656, "learning_rate": 0.001, "loss": 3.2354, "step": 10100 }, { "epoch": 3.2967032967032965, "grad_norm": 1.0885473489761353, "learning_rate": 0.001, "loss": 3.217, "step": 10200 }, { "epoch": 3.3290239172592115, "grad_norm": 1.4477795362472534, "learning_rate": 0.001, "loss": 3.2184, "step": 10300 }, { "epoch": 3.361344537815126, "grad_norm": 0.9638878703117371, "learning_rate": 0.001, "loss": 3.2117, "step": 10400 }, { "epoch": 3.3936651583710407, "grad_norm": 0.9306259751319885, "learning_rate": 0.001, "loss": 3.2184, "step": 10500 }, { "epoch": 3.425985778926955, "grad_norm": 0.8912835717201233, "learning_rate": 0.001, "loss": 3.2402, "step": 10600 }, { "epoch": 3.45830639948287, "grad_norm": 1.0636497735977173, "learning_rate": 0.001, "loss": 3.2319, "step": 10700 }, { "epoch": 3.490627020038785, "grad_norm": 0.9546043276786804, "learning_rate": 0.001, "loss": 3.2295, "step": 10800 }, { "epoch": 3.5229476405946993, "grad_norm": 0.7783652544021606, "learning_rate": 0.001, "loss": 3.2105, "step": 10900 }, { "epoch": 3.555268261150614, "grad_norm": 0.9355916976928711, "learning_rate": 0.001, "loss": 3.2095, "step": 11000 }, { "epoch": 3.587588881706529, "grad_norm": 0.7190990447998047, "learning_rate": 0.001, "loss": 3.2194, "step": 11100 }, { "epoch": 3.6199095022624435, "grad_norm": 1.257772445678711, "learning_rate": 0.001, "loss": 3.227, "step": 11200 }, { "epoch": 3.652230122818358, "grad_norm": 0.7877817749977112, "learning_rate": 0.001, "loss": 3.24, "step": 11300 }, { "epoch": 3.684550743374273, "grad_norm": 1.0141783952713013, "learning_rate": 0.001, "loss": 3.2138, "step": 11400 }, { "epoch": 3.7168713639301876, "grad_norm": 0.9734593033790588, "learning_rate": 0.001, "loss": 3.1886, "step": 11500 }, { "epoch": 3.749191984486102, "grad_norm": 0.7290641665458679, "learning_rate": 0.001, "loss": 3.2082, "step": 11600 }, { "epoch": 3.7815126050420167, "grad_norm": 0.8883140683174133, "learning_rate": 0.001, "loss": 3.2206, "step": 11700 }, { "epoch": 3.8138332255979313, "grad_norm": 0.9745876789093018, "learning_rate": 0.001, "loss": 3.2189, "step": 11800 }, { "epoch": 3.8461538461538463, "grad_norm": 0.8779596090316772, "learning_rate": 0.001, "loss": 3.1992, "step": 11900 }, { "epoch": 3.878474466709761, "grad_norm": 0.8627077341079712, "learning_rate": 0.001, "loss": 3.2177, "step": 12000 }, { "epoch": 3.9107950872656754, "grad_norm": 0.7932115793228149, "learning_rate": 0.001, "loss": 3.2022, "step": 12100 }, { "epoch": 3.9431157078215904, "grad_norm": 0.9975898861885071, "learning_rate": 0.001, "loss": 3.1953, "step": 12200 }, { "epoch": 3.975436328377505, "grad_norm": 1.192795753479004, "learning_rate": 0.001, "loss": 3.2168, "step": 12300 }, { "epoch": 4.0077569489334195, "grad_norm": 1.080036997795105, "learning_rate": 0.001, "loss": 3.1953, "step": 12400 }, { "epoch": 4.040077569489334, "grad_norm": 0.8876031041145325, "learning_rate": 0.001, "loss": 3.0628, "step": 12500 }, { "epoch": 4.072398190045249, "grad_norm": 0.9834277629852295, "learning_rate": 0.001, "loss": 3.0713, "step": 12600 }, { "epoch": 4.104718810601163, "grad_norm": 0.9485188126564026, "learning_rate": 0.001, "loss": 3.0813, "step": 12700 }, { "epoch": 4.137039431157079, "grad_norm": 1.029014229774475, "learning_rate": 0.001, "loss": 3.0846, "step": 12800 }, { "epoch": 4.169360051712993, "grad_norm": 0.7999043464660645, "learning_rate": 0.001, "loss": 3.0871, "step": 12900 }, { "epoch": 4.201680672268908, "grad_norm": 1.1836215257644653, "learning_rate": 0.001, "loss": 3.0808, "step": 13000 }, { "epoch": 4.234001292824822, "grad_norm": 0.7550233602523804, "learning_rate": 0.001, "loss": 3.0935, "step": 13100 }, { "epoch": 4.266321913380737, "grad_norm": 0.9442141056060791, "learning_rate": 0.001, "loss": 3.0744, "step": 13200 }, { "epoch": 4.298642533936651, "grad_norm": 0.864423930644989, "learning_rate": 0.001, "loss": 3.0986, "step": 13300 }, { "epoch": 4.330963154492566, "grad_norm": 0.780209481716156, "learning_rate": 0.001, "loss": 3.0885, "step": 13400 }, { "epoch": 4.3632837750484805, "grad_norm": 0.9403650164604187, "learning_rate": 0.001, "loss": 3.1163, "step": 13500 }, { "epoch": 4.395604395604396, "grad_norm": 0.7610721588134766, "learning_rate": 0.001, "loss": 3.0917, "step": 13600 }, { "epoch": 4.4279250161603105, "grad_norm": 1.266564130783081, "learning_rate": 0.001, "loss": 3.108, "step": 13700 }, { "epoch": 4.460245636716225, "grad_norm": 1.096002221107483, "learning_rate": 0.001, "loss": 3.1087, "step": 13800 }, { "epoch": 4.49256625727214, "grad_norm": 0.7693293690681458, "learning_rate": 0.001, "loss": 3.1152, "step": 13900 }, { "epoch": 4.524886877828054, "grad_norm": 0.8701288104057312, "learning_rate": 0.001, "loss": 3.1059, "step": 14000 }, { "epoch": 4.557207498383969, "grad_norm": 0.995335578918457, "learning_rate": 0.001, "loss": 3.0896, "step": 14100 }, { "epoch": 4.589528118939883, "grad_norm": 1.0865423679351807, "learning_rate": 0.001, "loss": 3.1025, "step": 14200 }, { "epoch": 4.621848739495798, "grad_norm": 1.1124467849731445, "learning_rate": 0.001, "loss": 3.1061, "step": 14300 }, { "epoch": 4.654169360051713, "grad_norm": 1.4063979387283325, "learning_rate": 0.001, "loss": 3.097, "step": 14400 }, { "epoch": 4.686489980607628, "grad_norm": 0.9458435773849487, "learning_rate": 0.001, "loss": 3.1045, "step": 14500 }, { "epoch": 4.7188106011635425, "grad_norm": 0.6742769479751587, "learning_rate": 0.001, "loss": 3.1088, "step": 14600 }, { "epoch": 4.751131221719457, "grad_norm": 0.8667743802070618, "learning_rate": 0.001, "loss": 3.1002, "step": 14700 }, { "epoch": 4.783451842275372, "grad_norm": 0.9243529438972473, "learning_rate": 0.001, "loss": 3.1099, "step": 14800 }, { "epoch": 4.815772462831286, "grad_norm": 1.021361231803894, "learning_rate": 0.001, "loss": 3.1036, "step": 14900 }, { "epoch": 4.848093083387201, "grad_norm": 0.7967251539230347, "learning_rate": 0.001, "loss": 3.0663, "step": 15000 }, { "epoch": 4.880413703943116, "grad_norm": 0.8347923159599304, "learning_rate": 0.001, "loss": 3.0948, "step": 15100 }, { "epoch": 4.912734324499031, "grad_norm": 0.8334933519363403, "learning_rate": 0.001, "loss": 3.0957, "step": 15200 }, { "epoch": 4.945054945054945, "grad_norm": 0.9289120435714722, "learning_rate": 0.001, "loss": 3.1009, "step": 15300 }, { "epoch": 4.97737556561086, "grad_norm": 1.081323504447937, "learning_rate": 0.001, "loss": 3.0972, "step": 15400 }, { "epoch": 5.009696186166774, "grad_norm": 0.886936366558075, "learning_rate": 0.001, "loss": 3.0692, "step": 15500 }, { "epoch": 5.042016806722689, "grad_norm": 0.8752295970916748, "learning_rate": 0.001, "loss": 2.9501, "step": 15600 }, { "epoch": 5.0743374272786035, "grad_norm": 0.7193301916122437, "learning_rate": 0.001, "loss": 2.9623, "step": 15700 }, { "epoch": 5.106658047834518, "grad_norm": 1.147542953491211, "learning_rate": 0.001, "loss": 2.9765, "step": 15800 }, { "epoch": 5.1389786683904335, "grad_norm": 0.8573718070983887, "learning_rate": 0.001, "loss": 2.9609, "step": 15900 }, { "epoch": 5.171299288946348, "grad_norm": 0.9090355634689331, "learning_rate": 0.001, "loss": 2.9924, "step": 16000 }, { "epoch": 5.203619909502263, "grad_norm": 0.8234449625015259, "learning_rate": 0.001, "loss": 2.9723, "step": 16100 }, { "epoch": 5.235940530058177, "grad_norm": 0.712264358997345, "learning_rate": 0.001, "loss": 2.9773, "step": 16200 }, { "epoch": 5.268261150614092, "grad_norm": 1.019116997718811, "learning_rate": 0.001, "loss": 2.9837, "step": 16300 }, { "epoch": 5.300581771170006, "grad_norm": 1.2517943382263184, "learning_rate": 0.001, "loss": 2.9781, "step": 16400 }, { "epoch": 5.332902391725921, "grad_norm": 0.9867748618125916, "learning_rate": 0.001, "loss": 3.0102, "step": 16500 }, { "epoch": 5.365223012281835, "grad_norm": 0.8714147806167603, "learning_rate": 0.001, "loss": 3.0023, "step": 16600 }, { "epoch": 5.397543632837751, "grad_norm": 0.852895975112915, "learning_rate": 0.001, "loss": 2.9872, "step": 16700 }, { "epoch": 5.429864253393665, "grad_norm": 1.0148080587387085, "learning_rate": 0.001, "loss": 3.0073, "step": 16800 }, { "epoch": 5.46218487394958, "grad_norm": 1.0061284303665161, "learning_rate": 0.001, "loss": 3.0088, "step": 16900 }, { "epoch": 5.4945054945054945, "grad_norm": 1.031447410583496, "learning_rate": 0.001, "loss": 2.9962, "step": 17000 }, { "epoch": 5.526826115061409, "grad_norm": 0.8130742311477661, "learning_rate": 0.001, "loss": 3.0175, "step": 17100 }, { "epoch": 5.559146735617324, "grad_norm": 0.78955078125, "learning_rate": 0.001, "loss": 3.0058, "step": 17200 }, { "epoch": 5.591467356173238, "grad_norm": 0.7605058550834656, "learning_rate": 0.001, "loss": 2.9887, "step": 17300 }, { "epoch": 5.623787976729153, "grad_norm": 0.8350357413291931, "learning_rate": 0.001, "loss": 3.0166, "step": 17400 }, { "epoch": 5.656108597285068, "grad_norm": 0.7349731922149658, "learning_rate": 0.001, "loss": 2.9868, "step": 17500 }, { "epoch": 5.688429217840983, "grad_norm": 0.8610934615135193, "learning_rate": 0.001, "loss": 3.0184, "step": 17600 }, { "epoch": 5.720749838396897, "grad_norm": 0.9035360813140869, "learning_rate": 0.001, "loss": 3.0072, "step": 17700 }, { "epoch": 5.753070458952812, "grad_norm": 0.8169543147087097, "learning_rate": 0.001, "loss": 3.0184, "step": 17800 }, { "epoch": 5.785391079508726, "grad_norm": 0.932416558265686, "learning_rate": 0.001, "loss": 3.0175, "step": 17900 }, { "epoch": 5.817711700064641, "grad_norm": 0.83762526512146, "learning_rate": 0.001, "loss": 2.9901, "step": 18000 }, { "epoch": 5.850032320620556, "grad_norm": 0.7610704898834229, "learning_rate": 0.001, "loss": 3.0045, "step": 18100 }, { "epoch": 5.882352941176471, "grad_norm": 0.7906739711761475, "learning_rate": 0.001, "loss": 3.0014, "step": 18200 }, { "epoch": 5.914673561732386, "grad_norm": 1.1893870830535889, "learning_rate": 0.001, "loss": 3.0193, "step": 18300 }, { "epoch": 5.9469941822883, "grad_norm": 1.0120941400527954, "learning_rate": 0.001, "loss": 3.0242, "step": 18400 }, { "epoch": 5.979314802844215, "grad_norm": 0.8534390926361084, "learning_rate": 0.001, "loss": 3.008, "step": 18500 }, { "epoch": 6.011635423400129, "grad_norm": 0.9205257296562195, "learning_rate": 0.001, "loss": 2.9515, "step": 18600 }, { "epoch": 6.043956043956044, "grad_norm": 1.1282649040222168, "learning_rate": 0.001, "loss": 2.8832, "step": 18700 }, { "epoch": 6.076276664511958, "grad_norm": 1.0840119123458862, "learning_rate": 0.001, "loss": 2.865, "step": 18800 }, { "epoch": 6.108597285067873, "grad_norm": 1.0464580059051514, "learning_rate": 0.001, "loss": 2.8871, "step": 18900 }, { "epoch": 6.140917905623788, "grad_norm": 0.8120374083518982, "learning_rate": 0.001, "loss": 2.8726, "step": 19000 }, { "epoch": 6.173238526179703, "grad_norm": 0.8664466738700867, "learning_rate": 0.001, "loss": 2.899, "step": 19100 }, { "epoch": 6.2055591467356175, "grad_norm": 0.9399696588516235, "learning_rate": 0.001, "loss": 2.8829, "step": 19200 }, { "epoch": 6.237879767291532, "grad_norm": 1.3675711154937744, "learning_rate": 0.001, "loss": 2.9043, "step": 19300 }, { "epoch": 6.270200387847447, "grad_norm": 1.2555503845214844, "learning_rate": 0.001, "loss": 2.8953, "step": 19400 }, { "epoch": 6.302521008403361, "grad_norm": 0.8409241437911987, "learning_rate": 0.001, "loss": 2.9037, "step": 19500 }, { "epoch": 6.334841628959276, "grad_norm": 0.7829875946044922, "learning_rate": 0.001, "loss": 2.9159, "step": 19600 }, { "epoch": 6.36716224951519, "grad_norm": 0.9084135293960571, "learning_rate": 0.001, "loss": 2.911, "step": 19700 }, { "epoch": 6.399482870071106, "grad_norm": 1.1866672039031982, "learning_rate": 0.001, "loss": 2.8979, "step": 19800 }, { "epoch": 6.43180349062702, "grad_norm": 0.7364369034767151, "learning_rate": 0.001, "loss": 2.9264, "step": 19900 }, { "epoch": 6.464124111182935, "grad_norm": 0.8288730382919312, "learning_rate": 0.001, "loss": 2.9046, "step": 20000 }, { "epoch": 6.496444731738849, "grad_norm": 1.1130245923995972, "learning_rate": 0.001, "loss": 2.9216, "step": 20100 }, { "epoch": 6.528765352294764, "grad_norm": 1.074400782585144, "learning_rate": 0.001, "loss": 2.9239, "step": 20200 }, { "epoch": 6.5610859728506785, "grad_norm": 0.8148232102394104, "learning_rate": 0.001, "loss": 2.8964, "step": 20300 }, { "epoch": 6.593406593406593, "grad_norm": 1.0962389707565308, "learning_rate": 0.001, "loss": 2.9386, "step": 20400 }, { "epoch": 6.625727213962508, "grad_norm": 0.788106381893158, "learning_rate": 0.001, "loss": 2.9134, "step": 20500 }, { "epoch": 6.658047834518423, "grad_norm": 0.8781216144561768, "learning_rate": 0.001, "loss": 2.9286, "step": 20600 }, { "epoch": 6.690368455074338, "grad_norm": 0.8809503316879272, "learning_rate": 0.001, "loss": 2.9277, "step": 20700 }, { "epoch": 6.722689075630252, "grad_norm": 0.8170621991157532, "learning_rate": 0.001, "loss": 2.9222, "step": 20800 }, { "epoch": 6.755009696186167, "grad_norm": 1.054281234741211, "learning_rate": 0.001, "loss": 2.9309, "step": 20900 }, { "epoch": 6.787330316742081, "grad_norm": 0.8404674530029297, "learning_rate": 0.001, "loss": 2.9194, "step": 21000 }, { "epoch": 6.819650937297996, "grad_norm": 0.8002380132675171, "learning_rate": 0.001, "loss": 2.9142, "step": 21100 }, { "epoch": 6.85197155785391, "grad_norm": 1.09268057346344, "learning_rate": 0.001, "loss": 2.9427, "step": 21200 }, { "epoch": 6.884292178409826, "grad_norm": 1.0005022287368774, "learning_rate": 0.001, "loss": 2.9272, "step": 21300 }, { "epoch": 6.91661279896574, "grad_norm": 0.8553439974784851, "learning_rate": 0.001, "loss": 2.9235, "step": 21400 }, { "epoch": 6.948933419521655, "grad_norm": 0.8628080487251282, "learning_rate": 0.001, "loss": 2.9177, "step": 21500 }, { "epoch": 6.98125404007757, "grad_norm": 1.0014255046844482, "learning_rate": 0.001, "loss": 2.9363, "step": 21600 }, { "epoch": 7.013574660633484, "grad_norm": 1.1169780492782593, "learning_rate": 0.001, "loss": 2.8747, "step": 21700 }, { "epoch": 7.045895281189399, "grad_norm": 1.1986509561538696, "learning_rate": 0.001, "loss": 2.7947, "step": 21800 }, { "epoch": 7.078215901745313, "grad_norm": 0.9472404718399048, "learning_rate": 0.001, "loss": 2.781, "step": 21900 }, { "epoch": 7.110536522301228, "grad_norm": 0.8761770725250244, "learning_rate": 0.001, "loss": 2.8102, "step": 22000 }, { "epoch": 7.142857142857143, "grad_norm": 1.10127854347229, "learning_rate": 0.001, "loss": 2.7989, "step": 22100 }, { "epoch": 7.175177763413058, "grad_norm": 1.1972665786743164, "learning_rate": 0.001, "loss": 2.8007, "step": 22200 }, { "epoch": 7.207498383968972, "grad_norm": 0.9232390522956848, "learning_rate": 0.001, "loss": 2.8016, "step": 22300 }, { "epoch": 7.239819004524887, "grad_norm": 1.0866069793701172, "learning_rate": 0.001, "loss": 2.8312, "step": 22400 }, { "epoch": 7.2721396250808015, "grad_norm": 1.0556726455688477, "learning_rate": 0.001, "loss": 2.8322, "step": 22500 }, { "epoch": 7.304460245636716, "grad_norm": 1.025449514389038, "learning_rate": 0.001, "loss": 2.8113, "step": 22600 }, { "epoch": 7.336780866192631, "grad_norm": 0.8753816485404968, "learning_rate": 0.001, "loss": 2.8309, "step": 22700 }, { "epoch": 7.369101486748546, "grad_norm": 1.1139119863510132, "learning_rate": 0.001, "loss": 2.8213, "step": 22800 }, { "epoch": 7.401422107304461, "grad_norm": 1.1850379705429077, "learning_rate": 0.001, "loss": 2.8138, "step": 22900 }, { "epoch": 7.433742727860375, "grad_norm": 0.8396257162094116, "learning_rate": 0.001, "loss": 2.8561, "step": 23000 }, { "epoch": 7.46606334841629, "grad_norm": 1.0254182815551758, "learning_rate": 0.001, "loss": 2.845, "step": 23100 }, { "epoch": 7.498383968972204, "grad_norm": 1.0984256267547607, "learning_rate": 0.001, "loss": 2.8638, "step": 23200 }, { "epoch": 7.530704589528119, "grad_norm": 1.1533817052841187, "learning_rate": 0.001, "loss": 2.8465, "step": 23300 }, { "epoch": 7.563025210084033, "grad_norm": 0.9635702967643738, "learning_rate": 0.001, "loss": 2.8595, "step": 23400 }, { "epoch": 7.595345830639948, "grad_norm": 0.8350183367729187, "learning_rate": 0.001, "loss": 2.8639, "step": 23500 }, { "epoch": 7.6276664511958625, "grad_norm": 1.0109552145004272, "learning_rate": 0.001, "loss": 2.8395, "step": 23600 }, { "epoch": 7.659987071751778, "grad_norm": 1.049338698387146, "learning_rate": 0.001, "loss": 2.8587, "step": 23700 }, { "epoch": 7.6923076923076925, "grad_norm": 0.985700786113739, "learning_rate": 0.001, "loss": 2.8637, "step": 23800 }, { "epoch": 7.724628312863607, "grad_norm": 0.9394757151603699, "learning_rate": 0.001, "loss": 2.8453, "step": 23900 }, { "epoch": 7.756948933419522, "grad_norm": 1.203756332397461, "learning_rate": 0.001, "loss": 2.8625, "step": 24000 }, { "epoch": 7.789269553975436, "grad_norm": 0.89085853099823, "learning_rate": 0.001, "loss": 2.8558, "step": 24100 }, { "epoch": 7.821590174531351, "grad_norm": 1.2652662992477417, "learning_rate": 0.001, "loss": 2.8569, "step": 24200 }, { "epoch": 7.853910795087265, "grad_norm": 1.0638023614883423, "learning_rate": 0.001, "loss": 2.8699, "step": 24300 }, { "epoch": 7.886231415643181, "grad_norm": 1.0257869958877563, "learning_rate": 0.001, "loss": 2.8554, "step": 24400 }, { "epoch": 7.918552036199095, "grad_norm": 1.2246901988983154, "learning_rate": 0.001, "loss": 2.8755, "step": 24500 }, { "epoch": 7.95087265675501, "grad_norm": 0.8839750289916992, "learning_rate": 0.001, "loss": 2.845, "step": 24600 }, { "epoch": 7.983193277310924, "grad_norm": 1.1866658926010132, "learning_rate": 0.001, "loss": 2.858, "step": 24700 }, { "epoch": 8.015513897866839, "grad_norm": 1.3818105459213257, "learning_rate": 0.001, "loss": 2.7831, "step": 24800 }, { "epoch": 8.047834518422754, "grad_norm": 1.0419820547103882, "learning_rate": 0.001, "loss": 2.7238, "step": 24900 }, { "epoch": 8.080155138978668, "grad_norm": 1.126366376876831, "learning_rate": 0.001, "loss": 2.7257, "step": 25000 }, { "epoch": 8.112475759534583, "grad_norm": 1.0105080604553223, "learning_rate": 0.001, "loss": 2.7445, "step": 25100 }, { "epoch": 8.144796380090497, "grad_norm": 1.1063216924667358, "learning_rate": 0.001, "loss": 2.7421, "step": 25200 }, { "epoch": 8.177117000646412, "grad_norm": 1.4662680625915527, "learning_rate": 0.001, "loss": 2.7432, "step": 25300 }, { "epoch": 8.209437621202326, "grad_norm": 1.0692142248153687, "learning_rate": 0.001, "loss": 2.7346, "step": 25400 }, { "epoch": 8.241758241758241, "grad_norm": 1.5519376993179321, "learning_rate": 0.001, "loss": 2.7417, "step": 25500 }, { "epoch": 8.274078862314157, "grad_norm": 1.1988738775253296, "learning_rate": 0.001, "loss": 2.7734, "step": 25600 }, { "epoch": 8.306399482870072, "grad_norm": 1.592111587524414, "learning_rate": 0.001, "loss": 2.7705, "step": 25700 }, { "epoch": 8.338720103425986, "grad_norm": 1.6889156103134155, "learning_rate": 0.001, "loss": 2.7662, "step": 25800 }, { "epoch": 8.371040723981901, "grad_norm": 1.0734881162643433, "learning_rate": 0.001, "loss": 2.7336, "step": 25900 }, { "epoch": 8.403361344537815, "grad_norm": 1.3303179740905762, "learning_rate": 0.001, "loss": 2.7566, "step": 26000 }, { "epoch": 8.43568196509373, "grad_norm": 1.2512389421463013, "learning_rate": 0.001, "loss": 2.7635, "step": 26100 }, { "epoch": 8.468002585649645, "grad_norm": 1.4159555435180664, "learning_rate": 0.001, "loss": 2.7741, "step": 26200 }, { "epoch": 8.50032320620556, "grad_norm": 1.1148388385772705, "learning_rate": 0.001, "loss": 2.7955, "step": 26300 }, { "epoch": 8.532643826761474, "grad_norm": 0.9710886478424072, "learning_rate": 0.001, "loss": 2.7931, "step": 26400 }, { "epoch": 8.564964447317388, "grad_norm": 1.1238210201263428, "learning_rate": 0.001, "loss": 2.7663, "step": 26500 }, { "epoch": 8.597285067873303, "grad_norm": 1.0266234874725342, "learning_rate": 0.001, "loss": 2.7859, "step": 26600 }, { "epoch": 8.629605688429217, "grad_norm": 1.1067254543304443, "learning_rate": 0.001, "loss": 2.7709, "step": 26700 }, { "epoch": 8.661926308985132, "grad_norm": 1.1781024932861328, "learning_rate": 0.001, "loss": 2.7905, "step": 26800 }, { "epoch": 8.694246929541046, "grad_norm": 1.460496187210083, "learning_rate": 0.001, "loss": 2.7954, "step": 26900 }, { "epoch": 8.726567550096961, "grad_norm": 1.214035987854004, "learning_rate": 0.001, "loss": 2.798, "step": 27000 }, { "epoch": 8.758888170652877, "grad_norm": 1.1744524240493774, "learning_rate": 0.001, "loss": 2.773, "step": 27100 }, { "epoch": 8.791208791208792, "grad_norm": 1.1752805709838867, "learning_rate": 0.001, "loss": 2.7875, "step": 27200 }, { "epoch": 8.823529411764707, "grad_norm": 1.3713490962982178, "learning_rate": 0.001, "loss": 2.7948, "step": 27300 }, { "epoch": 8.855850032320621, "grad_norm": 1.4796494245529175, "learning_rate": 0.001, "loss": 2.7963, "step": 27400 }, { "epoch": 8.888170652876536, "grad_norm": 1.0539920330047607, "learning_rate": 0.001, "loss": 2.8062, "step": 27500 }, { "epoch": 8.92049127343245, "grad_norm": 1.4270648956298828, "learning_rate": 0.001, "loss": 2.8028, "step": 27600 }, { "epoch": 8.952811893988365, "grad_norm": 1.0808746814727783, "learning_rate": 0.001, "loss": 2.8078, "step": 27700 }, { "epoch": 8.98513251454428, "grad_norm": 1.1713131666183472, "learning_rate": 0.001, "loss": 2.8011, "step": 27800 }, { "epoch": 9.017453135100194, "grad_norm": 1.2255527973175049, "learning_rate": 0.001, "loss": 2.7072, "step": 27900 }, { "epoch": 9.049773755656108, "grad_norm": 1.5293316841125488, "learning_rate": 0.001, "loss": 2.6307, "step": 28000 }, { "epoch": 9.082094376212023, "grad_norm": 1.493971586227417, "learning_rate": 0.001, "loss": 2.6732, "step": 28100 }, { "epoch": 9.114414996767938, "grad_norm": 1.0354042053222656, "learning_rate": 0.001, "loss": 2.6383, "step": 28200 }, { "epoch": 9.146735617323852, "grad_norm": 1.18374502658844, "learning_rate": 0.001, "loss": 2.6923, "step": 28300 }, { "epoch": 9.179056237879767, "grad_norm": 0.874068021774292, "learning_rate": 0.001, "loss": 2.6817, "step": 28400 }, { "epoch": 9.211376858435681, "grad_norm": 1.5086418390274048, "learning_rate": 0.001, "loss": 2.6822, "step": 28500 }, { "epoch": 9.243697478991596, "grad_norm": 1.9380152225494385, "learning_rate": 0.001, "loss": 2.6832, "step": 28600 }, { "epoch": 9.276018099547512, "grad_norm": 0.9644330739974976, "learning_rate": 0.001, "loss": 2.7148, "step": 28700 }, { "epoch": 9.308338720103427, "grad_norm": 1.3587816953659058, "learning_rate": 0.001, "loss": 2.6998, "step": 28800 }, { "epoch": 9.340659340659341, "grad_norm": 1.1259444952011108, "learning_rate": 0.001, "loss": 2.6989, "step": 28900 }, { "epoch": 9.372979961215256, "grad_norm": 1.3353569507598877, "learning_rate": 0.001, "loss": 2.7075, "step": 29000 }, { "epoch": 9.40530058177117, "grad_norm": 1.4237273931503296, "learning_rate": 0.001, "loss": 2.6998, "step": 29100 }, { "epoch": 9.437621202327085, "grad_norm": 1.4491281509399414, "learning_rate": 0.001, "loss": 2.7173, "step": 29200 }, { "epoch": 9.469941822883, "grad_norm": 1.1129282712936401, "learning_rate": 0.001, "loss": 2.7142, "step": 29300 }, { "epoch": 9.502262443438914, "grad_norm": 1.2722301483154297, "learning_rate": 0.001, "loss": 2.6983, "step": 29400 }, { "epoch": 9.534583063994829, "grad_norm": 1.1919955015182495, "learning_rate": 0.001, "loss": 2.7242, "step": 29500 }, { "epoch": 9.566903684550743, "grad_norm": 0.8851849436759949, "learning_rate": 0.001, "loss": 2.7172, "step": 29600 }, { "epoch": 9.599224305106658, "grad_norm": 1.1396182775497437, "learning_rate": 0.001, "loss": 2.7365, "step": 29700 }, { "epoch": 9.631544925662572, "grad_norm": 1.0789813995361328, "learning_rate": 0.001, "loss": 2.7288, "step": 29800 }, { "epoch": 9.663865546218487, "grad_norm": 1.1112487316131592, "learning_rate": 0.001, "loss": 2.7228, "step": 29900 }, { "epoch": 9.696186166774401, "grad_norm": 1.112724781036377, "learning_rate": 0.001, "loss": 2.7157, "step": 30000 }, { "epoch": 9.728506787330316, "grad_norm": 1.1656497716903687, "learning_rate": 0.001, "loss": 2.724, "step": 30100 }, { "epoch": 9.760827407886232, "grad_norm": 1.4802846908569336, "learning_rate": 0.001, "loss": 2.727, "step": 30200 }, { "epoch": 9.793148028442147, "grad_norm": 0.9352574348449707, "learning_rate": 0.001, "loss": 2.7296, "step": 30300 }, { "epoch": 9.825468648998061, "grad_norm": 1.2167083024978638, "learning_rate": 0.001, "loss": 2.7345, "step": 30400 }, { "epoch": 9.857789269553976, "grad_norm": 1.1363461017608643, "learning_rate": 0.001, "loss": 2.7302, "step": 30500 }, { "epoch": 9.89010989010989, "grad_norm": 1.0233099460601807, "learning_rate": 0.001, "loss": 2.7356, "step": 30600 }, { "epoch": 9.922430510665805, "grad_norm": 1.6240216493606567, "learning_rate": 0.001, "loss": 2.7486, "step": 30700 }, { "epoch": 9.95475113122172, "grad_norm": 1.293664813041687, "learning_rate": 0.001, "loss": 2.7391, "step": 30800 }, { "epoch": 9.987071751777634, "grad_norm": 1.1840318441390991, "learning_rate": 0.001, "loss": 2.746, "step": 30900 }, { "epoch": 10.019392372333549, "grad_norm": 1.403177261352539, "learning_rate": 0.001, "loss": 2.653, "step": 31000 }, { "epoch": 10.051712992889463, "grad_norm": 0.9433863759040833, "learning_rate": 0.001, "loss": 2.598, "step": 31100 }, { "epoch": 10.084033613445378, "grad_norm": 1.0219208002090454, "learning_rate": 0.001, "loss": 2.6092, "step": 31200 }, { "epoch": 10.116354234001292, "grad_norm": 0.9918056130409241, "learning_rate": 0.001, "loss": 2.6183, "step": 31300 }, { "epoch": 10.148674854557207, "grad_norm": 1.1675612926483154, "learning_rate": 0.001, "loss": 2.6016, "step": 31400 }, { "epoch": 10.180995475113122, "grad_norm": 0.7957185506820679, "learning_rate": 0.001, "loss": 2.6274, "step": 31500 }, { "epoch": 10.213316095669036, "grad_norm": 1.0217318534851074, "learning_rate": 0.001, "loss": 2.6152, "step": 31600 }, { "epoch": 10.24563671622495, "grad_norm": 1.0155982971191406, "learning_rate": 0.001, "loss": 2.6283, "step": 31700 }, { "epoch": 10.277957336780867, "grad_norm": 0.8074186444282532, "learning_rate": 0.001, "loss": 2.634, "step": 31800 }, { "epoch": 10.310277957336782, "grad_norm": 0.8662866353988647, "learning_rate": 0.001, "loss": 2.6434, "step": 31900 }, { "epoch": 10.342598577892696, "grad_norm": 1.2984246015548706, "learning_rate": 0.001, "loss": 2.6334, "step": 32000 }, { "epoch": 10.37491919844861, "grad_norm": 1.0286608934402466, "learning_rate": 0.001, "loss": 2.6406, "step": 32100 }, { "epoch": 10.407239819004525, "grad_norm": 1.104810118675232, "learning_rate": 0.001, "loss": 2.6734, "step": 32200 }, { "epoch": 10.43956043956044, "grad_norm": 0.9450679421424866, "learning_rate": 0.001, "loss": 2.6495, "step": 32300 }, { "epoch": 10.471881060116354, "grad_norm": 1.0712677240371704, "learning_rate": 0.001, "loss": 2.6659, "step": 32400 }, { "epoch": 10.504201680672269, "grad_norm": 0.8774548768997192, "learning_rate": 0.001, "loss": 2.6579, "step": 32500 }, { "epoch": 10.536522301228183, "grad_norm": 0.9978793859481812, "learning_rate": 0.001, "loss": 2.6579, "step": 32600 }, { "epoch": 10.568842921784098, "grad_norm": 1.344751000404358, "learning_rate": 0.001, "loss": 2.662, "step": 32700 }, { "epoch": 10.601163542340013, "grad_norm": 1.010345697402954, "learning_rate": 0.001, "loss": 2.6454, "step": 32800 }, { "epoch": 10.633484162895927, "grad_norm": 0.9622514247894287, "learning_rate": 0.001, "loss": 2.671, "step": 32900 }, { "epoch": 10.665804783451842, "grad_norm": 0.9483785033226013, "learning_rate": 0.001, "loss": 2.6475, "step": 33000 }, { "epoch": 10.698125404007756, "grad_norm": 1.1246763467788696, "learning_rate": 0.001, "loss": 2.6651, "step": 33100 }, { "epoch": 10.73044602456367, "grad_norm": 1.0108826160430908, "learning_rate": 0.001, "loss": 2.6842, "step": 33200 }, { "epoch": 10.762766645119587, "grad_norm": 0.9328263401985168, "learning_rate": 0.001, "loss": 2.6932, "step": 33300 }, { "epoch": 10.795087265675502, "grad_norm": 0.9086120128631592, "learning_rate": 0.001, "loss": 2.6769, "step": 33400 }, { "epoch": 10.827407886231416, "grad_norm": 0.953272819519043, "learning_rate": 0.001, "loss": 2.6935, "step": 33500 }, { "epoch": 10.85972850678733, "grad_norm": 1.2190536260604858, "learning_rate": 0.001, "loss": 2.6817, "step": 33600 }, { "epoch": 10.892049127343245, "grad_norm": 0.8844929337501526, "learning_rate": 0.001, "loss": 2.6891, "step": 33700 }, { "epoch": 10.92436974789916, "grad_norm": 0.9220634698867798, "learning_rate": 0.001, "loss": 2.6707, "step": 33800 }, { "epoch": 10.956690368455074, "grad_norm": 0.8071881532669067, "learning_rate": 0.001, "loss": 2.6822, "step": 33900 }, { "epoch": 10.989010989010989, "grad_norm": 1.1711838245391846, "learning_rate": 0.001, "loss": 2.6829, "step": 34000 }, { "epoch": 11.021331609566904, "grad_norm": 1.0321942567825317, "learning_rate": 0.001, "loss": 2.6256, "step": 34100 }, { "epoch": 11.053652230122818, "grad_norm": 1.0227347612380981, "learning_rate": 0.001, "loss": 2.5575, "step": 34200 }, { "epoch": 11.085972850678733, "grad_norm": 1.0060416460037231, "learning_rate": 0.001, "loss": 2.5526, "step": 34300 }, { "epoch": 11.118293471234647, "grad_norm": 1.0068382024765015, "learning_rate": 0.001, "loss": 2.5789, "step": 34400 }, { "epoch": 11.150614091790562, "grad_norm": 1.3233875036239624, "learning_rate": 0.001, "loss": 2.563, "step": 34500 }, { "epoch": 11.182934712346476, "grad_norm": 0.952245831489563, "learning_rate": 0.001, "loss": 2.5817, "step": 34600 }, { "epoch": 11.215255332902391, "grad_norm": 1.0358257293701172, "learning_rate": 0.001, "loss": 2.5875, "step": 34700 }, { "epoch": 11.247575953458306, "grad_norm": 1.11635160446167, "learning_rate": 0.001, "loss": 2.5677, "step": 34800 }, { "epoch": 11.279896574014222, "grad_norm": 1.115350604057312, "learning_rate": 0.001, "loss": 2.5832, "step": 34900 }, { "epoch": 11.312217194570136, "grad_norm": 1.3675668239593506, "learning_rate": 0.001, "loss": 2.5801, "step": 35000 }, { "epoch": 11.344537815126051, "grad_norm": 0.9039455652236938, "learning_rate": 0.001, "loss": 2.581, "step": 35100 }, { "epoch": 11.376858435681966, "grad_norm": 0.9431731700897217, "learning_rate": 0.001, "loss": 2.5871, "step": 35200 }, { "epoch": 11.40917905623788, "grad_norm": 1.5077314376831055, "learning_rate": 0.001, "loss": 2.5942, "step": 35300 }, { "epoch": 11.441499676793795, "grad_norm": 0.9280950427055359, "learning_rate": 0.001, "loss": 2.5854, "step": 35400 }, { "epoch": 11.47382029734971, "grad_norm": 0.8959194421768188, "learning_rate": 0.001, "loss": 2.6077, "step": 35500 }, { "epoch": 11.506140917905624, "grad_norm": 0.8173645734786987, "learning_rate": 0.001, "loss": 2.6389, "step": 35600 }, { "epoch": 11.538461538461538, "grad_norm": 0.9029143452644348, "learning_rate": 0.001, "loss": 2.6093, "step": 35700 }, { "epoch": 11.570782159017453, "grad_norm": 1.0597341060638428, "learning_rate": 0.001, "loss": 2.6155, "step": 35800 }, { "epoch": 11.603102779573367, "grad_norm": 0.7653180360794067, "learning_rate": 0.001, "loss": 2.5938, "step": 35900 }, { "epoch": 11.635423400129282, "grad_norm": 0.9608849287033081, "learning_rate": 0.001, "loss": 2.6059, "step": 36000 }, { "epoch": 11.667744020685197, "grad_norm": 0.7896739840507507, "learning_rate": 0.001, "loss": 2.6306, "step": 36100 }, { "epoch": 11.700064641241111, "grad_norm": 1.1343886852264404, "learning_rate": 0.001, "loss": 2.6083, "step": 36200 }, { "epoch": 11.732385261797026, "grad_norm": 1.0658496618270874, "learning_rate": 0.001, "loss": 2.6192, "step": 36300 }, { "epoch": 11.764705882352942, "grad_norm": 0.9849185943603516, "learning_rate": 0.001, "loss": 2.6148, "step": 36400 }, { "epoch": 11.797026502908857, "grad_norm": 1.1015666723251343, "learning_rate": 0.001, "loss": 2.6081, "step": 36500 }, { "epoch": 11.829347123464771, "grad_norm": 1.2068266868591309, "learning_rate": 0.001, "loss": 2.6453, "step": 36600 }, { "epoch": 11.861667744020686, "grad_norm": 1.140753984451294, "learning_rate": 0.001, "loss": 2.6327, "step": 36700 }, { "epoch": 11.8939883645766, "grad_norm": 1.2157039642333984, "learning_rate": 0.001, "loss": 2.6198, "step": 36800 }, { "epoch": 11.926308985132515, "grad_norm": 1.1962515115737915, "learning_rate": 0.001, "loss": 2.6495, "step": 36900 }, { "epoch": 11.95862960568843, "grad_norm": 1.0149519443511963, "learning_rate": 0.001, "loss": 2.6221, "step": 37000 }, { "epoch": 11.990950226244344, "grad_norm": 1.1916568279266357, "learning_rate": 0.001, "loss": 2.6462, "step": 37100 }, { "epoch": 12.023270846800258, "grad_norm": 1.198426365852356, "learning_rate": 0.001, "loss": 2.5254, "step": 37200 }, { "epoch": 12.055591467356173, "grad_norm": 1.0675870180130005, "learning_rate": 0.001, "loss": 2.4878, "step": 37300 }, { "epoch": 12.087912087912088, "grad_norm": 0.934251070022583, "learning_rate": 0.001, "loss": 2.4979, "step": 37400 }, { "epoch": 12.120232708468002, "grad_norm": 1.2349053621292114, "learning_rate": 0.001, "loss": 2.4913, "step": 37500 }, { "epoch": 12.152553329023917, "grad_norm": 1.235815405845642, "learning_rate": 0.001, "loss": 2.5046, "step": 37600 }, { "epoch": 12.184873949579831, "grad_norm": 1.257660150527954, "learning_rate": 0.001, "loss": 2.5259, "step": 37700 }, { "epoch": 12.217194570135746, "grad_norm": 1.246017336845398, "learning_rate": 0.001, "loss": 2.5421, "step": 37800 }, { "epoch": 12.24951519069166, "grad_norm": 1.1210070848464966, "learning_rate": 0.001, "loss": 2.5366, "step": 37900 }, { "epoch": 12.281835811247577, "grad_norm": 1.0047426223754883, "learning_rate": 0.001, "loss": 2.5445, "step": 38000 }, { "epoch": 12.314156431803491, "grad_norm": 0.8410885334014893, "learning_rate": 0.001, "loss": 2.5369, "step": 38100 }, { "epoch": 12.346477052359406, "grad_norm": 1.1904544830322266, "learning_rate": 0.001, "loss": 2.5295, "step": 38200 }, { "epoch": 12.37879767291532, "grad_norm": 1.5273857116699219, "learning_rate": 0.001, "loss": 2.5166, "step": 38300 }, { "epoch": 12.411118293471235, "grad_norm": 0.9626736640930176, "learning_rate": 0.001, "loss": 2.5535, "step": 38400 }, { "epoch": 12.44343891402715, "grad_norm": 0.8150332570075989, "learning_rate": 0.001, "loss": 2.5595, "step": 38500 }, { "epoch": 12.475759534583064, "grad_norm": 1.0832295417785645, "learning_rate": 0.001, "loss": 2.5457, "step": 38600 }, { "epoch": 12.508080155138979, "grad_norm": 1.0489040613174438, "learning_rate": 0.001, "loss": 2.5534, "step": 38700 }, { "epoch": 12.540400775694893, "grad_norm": 0.9710071682929993, "learning_rate": 0.001, "loss": 2.5536, "step": 38800 }, { "epoch": 12.572721396250808, "grad_norm": 1.0966761112213135, "learning_rate": 0.001, "loss": 2.5589, "step": 38900 }, { "epoch": 12.605042016806722, "grad_norm": 1.2801822423934937, "learning_rate": 0.001, "loss": 2.5783, "step": 39000 }, { "epoch": 12.637362637362637, "grad_norm": 0.9837306141853333, "learning_rate": 0.001, "loss": 2.578, "step": 39100 }, { "epoch": 12.669683257918551, "grad_norm": 0.9963669776916504, "learning_rate": 0.001, "loss": 2.5728, "step": 39200 }, { "epoch": 12.702003878474466, "grad_norm": 1.0618131160736084, "learning_rate": 0.001, "loss": 2.5602, "step": 39300 }, { "epoch": 12.73432449903038, "grad_norm": 1.1699451208114624, "learning_rate": 0.001, "loss": 2.6021, "step": 39400 }, { "epoch": 12.766645119586297, "grad_norm": 1.159548282623291, "learning_rate": 0.001, "loss": 2.5887, "step": 39500 }, { "epoch": 12.798965740142211, "grad_norm": 1.0369675159454346, "learning_rate": 0.001, "loss": 2.5738, "step": 39600 }, { "epoch": 12.831286360698126, "grad_norm": 1.0057525634765625, "learning_rate": 0.001, "loss": 2.593, "step": 39700 }, { "epoch": 12.86360698125404, "grad_norm": 1.0129728317260742, "learning_rate": 0.001, "loss": 2.5799, "step": 39800 }, { "epoch": 12.895927601809955, "grad_norm": 1.3273283243179321, "learning_rate": 0.001, "loss": 2.5779, "step": 39900 }, { "epoch": 12.92824822236587, "grad_norm": 0.9598479866981506, "learning_rate": 0.001, "loss": 2.5914, "step": 40000 }, { "epoch": 12.960568842921784, "grad_norm": 1.2933992147445679, "learning_rate": 0.001, "loss": 2.5951, "step": 40100 }, { "epoch": 12.992889463477699, "grad_norm": 1.0155681371688843, "learning_rate": 0.001, "loss": 2.5899, "step": 40200 }, { "epoch": 13.025210084033613, "grad_norm": 1.4213255643844604, "learning_rate": 0.001, "loss": 2.4736, "step": 40300 }, { "epoch": 13.057530704589528, "grad_norm": 1.0931323766708374, "learning_rate": 0.001, "loss": 2.458, "step": 40400 }, { "epoch": 13.089851325145442, "grad_norm": 1.0388720035552979, "learning_rate": 0.001, "loss": 2.4759, "step": 40500 }, { "epoch": 13.122171945701357, "grad_norm": 1.0456798076629639, "learning_rate": 0.001, "loss": 2.4536, "step": 40600 }, { "epoch": 13.154492566257272, "grad_norm": 0.8500736355781555, "learning_rate": 0.001, "loss": 2.465, "step": 40700 }, { "epoch": 13.186813186813186, "grad_norm": 1.1477937698364258, "learning_rate": 0.001, "loss": 2.4896, "step": 40800 }, { "epoch": 13.2191338073691, "grad_norm": 0.9674694538116455, "learning_rate": 0.001, "loss": 2.4841, "step": 40900 }, { "epoch": 13.251454427925015, "grad_norm": 0.861808717250824, "learning_rate": 0.001, "loss": 2.4665, "step": 41000 }, { "epoch": 13.283775048480932, "grad_norm": 0.9526507258415222, "learning_rate": 0.001, "loss": 2.4698, "step": 41100 }, { "epoch": 13.316095669036846, "grad_norm": 1.0938525199890137, "learning_rate": 0.001, "loss": 2.4909, "step": 41200 }, { "epoch": 13.34841628959276, "grad_norm": 0.9371994137763977, "learning_rate": 0.001, "loss": 2.4832, "step": 41300 }, { "epoch": 13.380736910148675, "grad_norm": 1.1339445114135742, "learning_rate": 0.001, "loss": 2.4994, "step": 41400 }, { "epoch": 13.41305753070459, "grad_norm": 0.8687131404876709, "learning_rate": 0.001, "loss": 2.5068, "step": 41500 }, { "epoch": 13.445378151260504, "grad_norm": 0.8617784380912781, "learning_rate": 0.001, "loss": 2.5108, "step": 41600 }, { "epoch": 13.477698771816419, "grad_norm": 1.063913345336914, "learning_rate": 0.001, "loss": 2.5043, "step": 41700 }, { "epoch": 13.510019392372334, "grad_norm": 1.0324054956436157, "learning_rate": 0.001, "loss": 2.5288, "step": 41800 }, { "epoch": 13.542340012928248, "grad_norm": 0.934737503528595, "learning_rate": 0.001, "loss": 2.5024, "step": 41900 }, { "epoch": 13.574660633484163, "grad_norm": 0.7680659294128418, "learning_rate": 0.001, "loss": 2.5058, "step": 42000 }, { "epoch": 13.606981254040077, "grad_norm": 1.341858148574829, "learning_rate": 0.001, "loss": 2.5329, "step": 42100 }, { "epoch": 13.639301874595992, "grad_norm": 0.8703786730766296, "learning_rate": 0.001, "loss": 2.524, "step": 42200 }, { "epoch": 13.671622495151906, "grad_norm": 1.0334969758987427, "learning_rate": 0.001, "loss": 2.5282, "step": 42300 }, { "epoch": 13.70394311570782, "grad_norm": 0.9755718111991882, "learning_rate": 0.001, "loss": 2.5121, "step": 42400 }, { "epoch": 13.736263736263737, "grad_norm": 0.9405517578125, "learning_rate": 0.001, "loss": 2.5202, "step": 42500 }, { "epoch": 13.768584356819652, "grad_norm": 1.0205954313278198, "learning_rate": 0.001, "loss": 2.5402, "step": 42600 }, { "epoch": 13.800904977375566, "grad_norm": 0.7872236371040344, "learning_rate": 0.001, "loss": 2.5279, "step": 42700 }, { "epoch": 13.83322559793148, "grad_norm": 1.1544864177703857, "learning_rate": 0.001, "loss": 2.548, "step": 42800 }, { "epoch": 13.865546218487395, "grad_norm": 0.7300485372543335, "learning_rate": 0.001, "loss": 2.5433, "step": 42900 }, { "epoch": 13.89786683904331, "grad_norm": 1.3647654056549072, "learning_rate": 0.001, "loss": 2.5652, "step": 43000 }, { "epoch": 13.930187459599225, "grad_norm": 1.4036544561386108, "learning_rate": 0.001, "loss": 2.5485, "step": 43100 }, { "epoch": 13.96250808015514, "grad_norm": 1.0153886079788208, "learning_rate": 0.001, "loss": 2.5488, "step": 43200 }, { "epoch": 13.994828700711054, "grad_norm": 1.1712045669555664, "learning_rate": 0.001, "loss": 2.5506, "step": 43300 }, { "epoch": 14.027149321266968, "grad_norm": 0.9621208310127258, "learning_rate": 0.001, "loss": 2.4262, "step": 43400 }, { "epoch": 14.059469941822883, "grad_norm": 0.9850360751152039, "learning_rate": 0.001, "loss": 2.405, "step": 43500 }, { "epoch": 14.091790562378797, "grad_norm": 1.406996726989746, "learning_rate": 0.001, "loss": 2.4143, "step": 43600 }, { "epoch": 14.124111182934712, "grad_norm": 1.0328227281570435, "learning_rate": 0.001, "loss": 2.4296, "step": 43700 }, { "epoch": 14.156431803490626, "grad_norm": 1.3105231523513794, "learning_rate": 0.001, "loss": 2.4179, "step": 43800 }, { "epoch": 14.188752424046541, "grad_norm": 1.309824824333191, "learning_rate": 0.001, "loss": 2.4447, "step": 43900 }, { "epoch": 14.221073044602456, "grad_norm": 1.170006275177002, "learning_rate": 0.001, "loss": 2.443, "step": 44000 }, { "epoch": 14.25339366515837, "grad_norm": 0.92409348487854, "learning_rate": 0.001, "loss": 2.4174, "step": 44100 }, { "epoch": 14.285714285714286, "grad_norm": 1.0731486082077026, "learning_rate": 0.001, "loss": 2.4297, "step": 44200 }, { "epoch": 14.318034906270201, "grad_norm": 0.9552510380744934, "learning_rate": 0.001, "loss": 2.4626, "step": 44300 }, { "epoch": 14.350355526826116, "grad_norm": 1.3744381666183472, "learning_rate": 0.001, "loss": 2.4487, "step": 44400 }, { "epoch": 14.38267614738203, "grad_norm": 0.9880512356758118, "learning_rate": 0.001, "loss": 2.4556, "step": 44500 }, { "epoch": 14.414996767937945, "grad_norm": 1.1186575889587402, "learning_rate": 0.001, "loss": 2.4665, "step": 44600 }, { "epoch": 14.44731738849386, "grad_norm": 1.2342720031738281, "learning_rate": 0.001, "loss": 2.4458, "step": 44700 }, { "epoch": 14.479638009049774, "grad_norm": 1.1874266862869263, "learning_rate": 0.001, "loss": 2.462, "step": 44800 }, { "epoch": 14.511958629605688, "grad_norm": 0.9831111431121826, "learning_rate": 0.001, "loss": 2.4848, "step": 44900 }, { "epoch": 14.544279250161603, "grad_norm": 1.2371684312820435, "learning_rate": 0.001, "loss": 2.4589, "step": 45000 }, { "epoch": 14.576599870717518, "grad_norm": 0.7964707016944885, "learning_rate": 0.001, "loss": 2.4648, "step": 45100 }, { "epoch": 14.608920491273432, "grad_norm": 1.0098804235458374, "learning_rate": 0.001, "loss": 2.4696, "step": 45200 }, { "epoch": 14.641241111829347, "grad_norm": 1.0485209226608276, "learning_rate": 0.001, "loss": 2.482, "step": 45300 }, { "epoch": 14.673561732385261, "grad_norm": 1.0087767839431763, "learning_rate": 0.001, "loss": 2.4725, "step": 45400 }, { "epoch": 14.705882352941176, "grad_norm": 1.2038854360580444, "learning_rate": 0.001, "loss": 2.479, "step": 45500 }, { "epoch": 14.738202973497092, "grad_norm": 0.9472567439079285, "learning_rate": 0.001, "loss": 2.4873, "step": 45600 }, { "epoch": 14.770523594053007, "grad_norm": 1.1870334148406982, "learning_rate": 0.001, "loss": 2.5014, "step": 45700 }, { "epoch": 14.802844214608921, "grad_norm": 0.9796122908592224, "learning_rate": 0.001, "loss": 2.4833, "step": 45800 }, { "epoch": 14.835164835164836, "grad_norm": 1.093863606452942, "learning_rate": 0.001, "loss": 2.5106, "step": 45900 }, { "epoch": 14.86748545572075, "grad_norm": 0.8866999745368958, "learning_rate": 0.001, "loss": 2.5075, "step": 46000 }, { "epoch": 14.899806076276665, "grad_norm": 1.0653350353240967, "learning_rate": 0.001, "loss": 2.4932, "step": 46100 }, { "epoch": 14.93212669683258, "grad_norm": 1.0493881702423096, "learning_rate": 0.001, "loss": 2.5202, "step": 46200 }, { "epoch": 14.964447317388494, "grad_norm": 0.9885337948799133, "learning_rate": 0.001, "loss": 2.4977, "step": 46300 }, { "epoch": 14.996767937944409, "grad_norm": 1.1102535724639893, "learning_rate": 0.001, "loss": 2.5095, "step": 46400 }, { "epoch": 15.029088558500323, "grad_norm": 1.1449183225631714, "learning_rate": 0.001, "loss": 2.3813, "step": 46500 }, { "epoch": 15.061409179056238, "grad_norm": 1.0444215536117554, "learning_rate": 0.001, "loss": 2.3625, "step": 46600 }, { "epoch": 15.093729799612152, "grad_norm": 1.1632866859436035, "learning_rate": 0.001, "loss": 2.3935, "step": 46700 }, { "epoch": 15.126050420168067, "grad_norm": 1.1026229858398438, "learning_rate": 0.001, "loss": 2.357, "step": 46800 }, { "epoch": 15.158371040723981, "grad_norm": 1.2819719314575195, "learning_rate": 0.001, "loss": 2.3705, "step": 46900 }, { "epoch": 15.190691661279896, "grad_norm": 1.118762493133545, "learning_rate": 0.001, "loss": 2.3738, "step": 47000 }, { "epoch": 15.22301228183581, "grad_norm": 0.9931496977806091, "learning_rate": 0.001, "loss": 2.3871, "step": 47100 }, { "epoch": 15.255332902391725, "grad_norm": 1.0876235961914062, "learning_rate": 0.001, "loss": 2.4186, "step": 47200 }, { "epoch": 15.287653522947641, "grad_norm": 1.0590667724609375, "learning_rate": 0.001, "loss": 2.3973, "step": 47300 }, { "epoch": 15.319974143503556, "grad_norm": 1.3229364156723022, "learning_rate": 0.001, "loss": 2.4038, "step": 47400 }, { "epoch": 15.35229476405947, "grad_norm": 1.4679980278015137, "learning_rate": 0.001, "loss": 2.4207, "step": 47500 }, { "epoch": 15.384615384615385, "grad_norm": 1.6495345830917358, "learning_rate": 0.001, "loss": 2.398, "step": 47600 }, { "epoch": 15.4169360051713, "grad_norm": 1.0480834245681763, "learning_rate": 0.001, "loss": 2.4145, "step": 47700 }, { "epoch": 15.449256625727214, "grad_norm": 1.340477705001831, "learning_rate": 0.001, "loss": 2.4233, "step": 47800 }, { "epoch": 15.481577246283129, "grad_norm": 1.1183497905731201, "learning_rate": 0.001, "loss": 2.4189, "step": 47900 }, { "epoch": 15.513897866839043, "grad_norm": 0.9559330344200134, "learning_rate": 0.001, "loss": 2.4346, "step": 48000 }, { "epoch": 15.546218487394958, "grad_norm": 1.090479850769043, "learning_rate": 0.001, "loss": 2.4432, "step": 48100 }, { "epoch": 15.578539107950872, "grad_norm": 1.1318349838256836, "learning_rate": 0.001, "loss": 2.4252, "step": 48200 }, { "epoch": 15.610859728506787, "grad_norm": 1.2161760330200195, "learning_rate": 0.001, "loss": 2.4305, "step": 48300 }, { "epoch": 15.643180349062701, "grad_norm": 1.0710586309432983, "learning_rate": 0.001, "loss": 2.4512, "step": 48400 }, { "epoch": 15.675500969618616, "grad_norm": 1.1996996402740479, "learning_rate": 0.001, "loss": 2.4546, "step": 48500 }, { "epoch": 15.70782159017453, "grad_norm": 1.1999651193618774, "learning_rate": 0.001, "loss": 2.4391, "step": 48600 }, { "epoch": 15.740142210730447, "grad_norm": 1.0343999862670898, "learning_rate": 0.001, "loss": 2.4512, "step": 48700 }, { "epoch": 15.772462831286362, "grad_norm": 1.038922667503357, "learning_rate": 0.001, "loss": 2.4524, "step": 48800 }, { "epoch": 15.804783451842276, "grad_norm": 1.0192667245864868, "learning_rate": 0.001, "loss": 2.4479, "step": 48900 }, { "epoch": 15.83710407239819, "grad_norm": 1.0764349699020386, "learning_rate": 0.001, "loss": 2.4616, "step": 49000 }, { "epoch": 15.869424692954105, "grad_norm": 1.4133694171905518, "learning_rate": 0.001, "loss": 2.4489, "step": 49100 }, { "epoch": 15.90174531351002, "grad_norm": 1.2370002269744873, "learning_rate": 0.001, "loss": 2.4603, "step": 49200 }, { "epoch": 15.934065934065934, "grad_norm": 0.9169979691505432, "learning_rate": 0.001, "loss": 2.4641, "step": 49300 }, { "epoch": 15.966386554621849, "grad_norm": 1.161568522453308, "learning_rate": 0.001, "loss": 2.471, "step": 49400 }, { "epoch": 15.998707175177763, "grad_norm": 1.4521310329437256, "learning_rate": 0.001, "loss": 2.456, "step": 49500 }, { "epoch": 16.031027795733678, "grad_norm": 1.418137550354004, "learning_rate": 0.001, "loss": 2.3271, "step": 49600 }, { "epoch": 16.063348416289593, "grad_norm": 1.6356244087219238, "learning_rate": 0.001, "loss": 2.3257, "step": 49700 }, { "epoch": 16.095669036845507, "grad_norm": 1.5366765260696411, "learning_rate": 0.001, "loss": 2.3292, "step": 49800 }, { "epoch": 16.12798965740142, "grad_norm": 1.386238694190979, "learning_rate": 0.001, "loss": 2.3504, "step": 49900 }, { "epoch": 16.160310277957336, "grad_norm": 1.2431461811065674, "learning_rate": 0.001, "loss": 2.3428, "step": 50000 }, { "epoch": 16.19263089851325, "grad_norm": 1.676797866821289, "learning_rate": 0.001, "loss": 2.3447, "step": 50100 }, { "epoch": 16.224951519069165, "grad_norm": 1.3970446586608887, "learning_rate": 0.001, "loss": 2.36, "step": 50200 }, { "epoch": 16.25727213962508, "grad_norm": 1.542039394378662, "learning_rate": 0.001, "loss": 2.3574, "step": 50300 }, { "epoch": 16.289592760180994, "grad_norm": 1.7507634162902832, "learning_rate": 0.001, "loss": 2.3575, "step": 50400 }, { "epoch": 16.32191338073691, "grad_norm": 1.0830522775650024, "learning_rate": 0.001, "loss": 2.3557, "step": 50500 }, { "epoch": 16.354234001292824, "grad_norm": 1.1822794675827026, "learning_rate": 0.001, "loss": 2.3762, "step": 50600 }, { "epoch": 16.386554621848738, "grad_norm": 1.2072137594223022, "learning_rate": 0.001, "loss": 2.3763, "step": 50700 }, { "epoch": 16.418875242404653, "grad_norm": 1.2439135313034058, "learning_rate": 0.001, "loss": 2.3854, "step": 50800 }, { "epoch": 16.451195862960567, "grad_norm": 1.6300753355026245, "learning_rate": 0.001, "loss": 2.3863, "step": 50900 }, { "epoch": 16.483516483516482, "grad_norm": 1.2830435037612915, "learning_rate": 0.001, "loss": 2.3932, "step": 51000 }, { "epoch": 16.5158371040724, "grad_norm": 1.2407660484313965, "learning_rate": 0.001, "loss": 2.3846, "step": 51100 }, { "epoch": 16.548157724628314, "grad_norm": 1.5125946998596191, "learning_rate": 0.001, "loss": 2.38, "step": 51200 }, { "epoch": 16.58047834518423, "grad_norm": 1.2218705415725708, "learning_rate": 0.001, "loss": 2.4112, "step": 51300 }, { "epoch": 16.612798965740144, "grad_norm": 1.1899499893188477, "learning_rate": 0.001, "loss": 2.3926, "step": 51400 }, { "epoch": 16.645119586296058, "grad_norm": 1.378842830657959, "learning_rate": 0.001, "loss": 2.3945, "step": 51500 }, { "epoch": 16.677440206851973, "grad_norm": 1.0433300733566284, "learning_rate": 0.001, "loss": 2.4003, "step": 51600 }, { "epoch": 16.709760827407887, "grad_norm": 1.6604347229003906, "learning_rate": 0.001, "loss": 2.4054, "step": 51700 }, { "epoch": 16.742081447963802, "grad_norm": 1.0785212516784668, "learning_rate": 0.001, "loss": 2.4039, "step": 51800 }, { "epoch": 16.774402068519716, "grad_norm": 1.3013510704040527, "learning_rate": 0.001, "loss": 2.4119, "step": 51900 }, { "epoch": 16.80672268907563, "grad_norm": 1.3156054019927979, "learning_rate": 0.001, "loss": 2.4216, "step": 52000 }, { "epoch": 16.839043309631545, "grad_norm": 1.7108491659164429, "learning_rate": 0.001, "loss": 2.4079, "step": 52100 }, { "epoch": 16.87136393018746, "grad_norm": 1.443699836730957, "learning_rate": 0.001, "loss": 2.4053, "step": 52200 }, { "epoch": 16.903684550743375, "grad_norm": 1.277143955230713, "learning_rate": 0.001, "loss": 2.421, "step": 52300 }, { "epoch": 16.93600517129929, "grad_norm": 1.3698267936706543, "learning_rate": 0.001, "loss": 2.4179, "step": 52400 }, { "epoch": 16.968325791855204, "grad_norm": 1.3238378763198853, "learning_rate": 0.001, "loss": 2.4443, "step": 52500 }, { "epoch": 17.00064641241112, "grad_norm": 1.0544229745864868, "learning_rate": 0.001, "loss": 2.4145, "step": 52600 }, { "epoch": 17.032967032967033, "grad_norm": 1.4758795499801636, "learning_rate": 0.001, "loss": 2.2783, "step": 52700 }, { "epoch": 17.065287653522947, "grad_norm": 1.10501229763031, "learning_rate": 0.001, "loss": 2.2856, "step": 52800 }, { "epoch": 17.097608274078862, "grad_norm": 1.2699426412582397, "learning_rate": 0.001, "loss": 2.273, "step": 52900 }, { "epoch": 17.129928894634777, "grad_norm": 1.629236102104187, "learning_rate": 0.001, "loss": 2.294, "step": 53000 }, { "epoch": 17.16224951519069, "grad_norm": 2.1197991371154785, "learning_rate": 0.001, "loss": 2.3205, "step": 53100 }, { "epoch": 17.194570135746606, "grad_norm": 1.8824236392974854, "learning_rate": 0.001, "loss": 2.3119, "step": 53200 }, { "epoch": 17.22689075630252, "grad_norm": 1.3362033367156982, "learning_rate": 0.001, "loss": 2.3331, "step": 53300 }, { "epoch": 17.259211376858435, "grad_norm": 1.6379491090774536, "learning_rate": 0.001, "loss": 2.3171, "step": 53400 }, { "epoch": 17.29153199741435, "grad_norm": 2.098562240600586, "learning_rate": 0.001, "loss": 2.3238, "step": 53500 }, { "epoch": 17.323852617970264, "grad_norm": 2.088083505630493, "learning_rate": 0.001, "loss": 2.3207, "step": 53600 }, { "epoch": 17.35617323852618, "grad_norm": 1.4209333658218384, "learning_rate": 0.001, "loss": 2.3402, "step": 53700 }, { "epoch": 17.388493859082093, "grad_norm": 1.4178082942962646, "learning_rate": 0.001, "loss": 2.3379, "step": 53800 }, { "epoch": 17.420814479638008, "grad_norm": 1.5361058712005615, "learning_rate": 0.001, "loss": 2.3434, "step": 53900 }, { "epoch": 17.453135100193922, "grad_norm": 1.3857008218765259, "learning_rate": 0.001, "loss": 2.3418, "step": 54000 }, { "epoch": 17.485455720749837, "grad_norm": 1.328792691230774, "learning_rate": 0.001, "loss": 2.351, "step": 54100 }, { "epoch": 17.517776341305755, "grad_norm": 1.2982306480407715, "learning_rate": 0.001, "loss": 2.3433, "step": 54200 }, { "epoch": 17.55009696186167, "grad_norm": 1.5809420347213745, "learning_rate": 0.001, "loss": 2.3577, "step": 54300 }, { "epoch": 17.582417582417584, "grad_norm": 1.4151922464370728, "learning_rate": 0.001, "loss": 2.3852, "step": 54400 }, { "epoch": 17.6147382029735, "grad_norm": 1.4341883659362793, "learning_rate": 0.001, "loss": 2.3577, "step": 54500 }, { "epoch": 17.647058823529413, "grad_norm": 1.6244245767593384, "learning_rate": 0.001, "loss": 2.3682, "step": 54600 }, { "epoch": 17.679379444085328, "grad_norm": 1.7657774686813354, "learning_rate": 0.001, "loss": 2.3503, "step": 54700 }, { "epoch": 17.711700064641242, "grad_norm": 1.8639613389968872, "learning_rate": 0.001, "loss": 2.3719, "step": 54800 }, { "epoch": 17.744020685197157, "grad_norm": 1.389522910118103, "learning_rate": 0.001, "loss": 2.3989, "step": 54900 }, { "epoch": 17.77634130575307, "grad_norm": 1.4958171844482422, "learning_rate": 0.001, "loss": 2.3753, "step": 55000 }, { "epoch": 17.808661926308986, "grad_norm": 1.6785283088684082, "learning_rate": 0.001, "loss": 2.3798, "step": 55100 }, { "epoch": 17.8409825468649, "grad_norm": 1.7483515739440918, "learning_rate": 0.001, "loss": 2.3783, "step": 55200 }, { "epoch": 17.873303167420815, "grad_norm": 1.5856812000274658, "learning_rate": 0.001, "loss": 2.3854, "step": 55300 }, { "epoch": 17.90562378797673, "grad_norm": 1.6331095695495605, "learning_rate": 0.001, "loss": 2.369, "step": 55400 }, { "epoch": 17.937944408532644, "grad_norm": 1.7543635368347168, "learning_rate": 0.001, "loss": 2.3685, "step": 55500 }, { "epoch": 17.97026502908856, "grad_norm": 1.5347305536270142, "learning_rate": 0.001, "loss": 2.3902, "step": 55600 }, { "epoch": 18.002585649644473, "grad_norm": 1.2079260349273682, "learning_rate": 0.001, "loss": 2.4057, "step": 55700 }, { "epoch": 18.034906270200388, "grad_norm": 1.109870195388794, "learning_rate": 0.001, "loss": 2.2369, "step": 55800 }, { "epoch": 18.067226890756302, "grad_norm": 0.941036581993103, "learning_rate": 0.001, "loss": 2.2422, "step": 55900 }, { "epoch": 18.099547511312217, "grad_norm": 1.3007092475891113, "learning_rate": 0.001, "loss": 2.2524, "step": 56000 }, { "epoch": 18.13186813186813, "grad_norm": 1.3963851928710938, "learning_rate": 0.001, "loss": 2.2683, "step": 56100 }, { "epoch": 18.164188752424046, "grad_norm": 1.2644718885421753, "learning_rate": 0.001, "loss": 2.293, "step": 56200 }, { "epoch": 18.19650937297996, "grad_norm": 1.173424243927002, "learning_rate": 0.001, "loss": 2.279, "step": 56300 }, { "epoch": 18.228829993535875, "grad_norm": 1.003485918045044, "learning_rate": 0.001, "loss": 2.2928, "step": 56400 }, { "epoch": 18.26115061409179, "grad_norm": 0.9148826599121094, "learning_rate": 0.001, "loss": 2.2939, "step": 56500 }, { "epoch": 18.293471234647704, "grad_norm": 1.1833020448684692, "learning_rate": 0.001, "loss": 2.2878, "step": 56600 }, { "epoch": 18.32579185520362, "grad_norm": 0.9342957735061646, "learning_rate": 0.001, "loss": 2.3092, "step": 56700 }, { "epoch": 18.358112475759533, "grad_norm": 1.3121254444122314, "learning_rate": 0.001, "loss": 2.269, "step": 56800 }, { "epoch": 18.390433096315448, "grad_norm": 1.312016248703003, "learning_rate": 0.001, "loss": 2.3006, "step": 56900 }, { "epoch": 18.422753716871362, "grad_norm": 1.1071215867996216, "learning_rate": 0.001, "loss": 2.3017, "step": 57000 }, { "epoch": 18.455074337427277, "grad_norm": 1.495505690574646, "learning_rate": 0.001, "loss": 2.3203, "step": 57100 }, { "epoch": 18.48739495798319, "grad_norm": 1.095226526260376, "learning_rate": 0.001, "loss": 2.2802, "step": 57200 }, { "epoch": 18.51971557853911, "grad_norm": 1.1401058435440063, "learning_rate": 0.001, "loss": 2.3278, "step": 57300 }, { "epoch": 18.552036199095024, "grad_norm": 1.0433752536773682, "learning_rate": 0.001, "loss": 2.341, "step": 57400 }, { "epoch": 18.58435681965094, "grad_norm": 1.3284567594528198, "learning_rate": 0.001, "loss": 2.3327, "step": 57500 }, { "epoch": 18.616677440206853, "grad_norm": 1.3075090646743774, "learning_rate": 0.001, "loss": 2.3224, "step": 57600 }, { "epoch": 18.648998060762768, "grad_norm": 1.1280314922332764, "learning_rate": 0.001, "loss": 2.3369, "step": 57700 }, { "epoch": 18.681318681318682, "grad_norm": 1.4304474592208862, "learning_rate": 0.001, "loss": 2.3305, "step": 57800 }, { "epoch": 18.713639301874597, "grad_norm": 1.4093077182769775, "learning_rate": 0.001, "loss": 2.3169, "step": 57900 }, { "epoch": 18.74595992243051, "grad_norm": 0.8837575912475586, "learning_rate": 0.001, "loss": 2.344, "step": 58000 }, { "epoch": 18.778280542986426, "grad_norm": 1.1318713426589966, "learning_rate": 0.001, "loss": 2.3622, "step": 58100 }, { "epoch": 18.81060116354234, "grad_norm": 1.010046124458313, "learning_rate": 0.001, "loss": 2.3521, "step": 58200 }, { "epoch": 18.842921784098255, "grad_norm": 1.1944527626037598, "learning_rate": 0.001, "loss": 2.3417, "step": 58300 }, { "epoch": 18.87524240465417, "grad_norm": 1.2252675294876099, "learning_rate": 0.001, "loss": 2.3597, "step": 58400 }, { "epoch": 18.907563025210084, "grad_norm": 1.2115516662597656, "learning_rate": 0.001, "loss": 2.3281, "step": 58500 }, { "epoch": 18.939883645766, "grad_norm": 1.3430302143096924, "learning_rate": 0.001, "loss": 2.3678, "step": 58600 }, { "epoch": 18.972204266321913, "grad_norm": 1.059554934501648, "learning_rate": 0.001, "loss": 2.3558, "step": 58700 }, { "epoch": 19.004524886877828, "grad_norm": 1.26351797580719, "learning_rate": 0.001, "loss": 2.3541, "step": 58800 }, { "epoch": 19.036845507433743, "grad_norm": 1.0600953102111816, "learning_rate": 0.001, "loss": 2.219, "step": 58900 }, { "epoch": 19.069166127989657, "grad_norm": 1.5005885362625122, "learning_rate": 0.001, "loss": 2.2073, "step": 59000 }, { "epoch": 19.10148674854557, "grad_norm": 1.4235773086547852, "learning_rate": 0.001, "loss": 2.2287, "step": 59100 }, { "epoch": 19.133807369101486, "grad_norm": 1.0466982126235962, "learning_rate": 0.001, "loss": 2.2534, "step": 59200 }, { "epoch": 19.1661279896574, "grad_norm": 1.4175443649291992, "learning_rate": 0.001, "loss": 2.2323, "step": 59300 }, { "epoch": 19.198448610213315, "grad_norm": 1.1985677480697632, "learning_rate": 0.001, "loss": 2.2415, "step": 59400 }, { "epoch": 19.23076923076923, "grad_norm": 1.0888758897781372, "learning_rate": 0.001, "loss": 2.2418, "step": 59500 }, { "epoch": 19.263089851325145, "grad_norm": 0.9708693027496338, "learning_rate": 0.001, "loss": 2.2625, "step": 59600 }, { "epoch": 19.29541047188106, "grad_norm": 1.3424566984176636, "learning_rate": 0.001, "loss": 2.2504, "step": 59700 }, { "epoch": 19.327731092436974, "grad_norm": 1.1441328525543213, "learning_rate": 0.001, "loss": 2.2535, "step": 59800 }, { "epoch": 19.360051712992888, "grad_norm": 1.1090998649597168, "learning_rate": 0.001, "loss": 2.2799, "step": 59900 }, { "epoch": 19.392372333548803, "grad_norm": 0.9804235696792603, "learning_rate": 0.001, "loss": 2.2683, "step": 60000 }, { "epoch": 19.424692954104717, "grad_norm": 1.1989473104476929, "learning_rate": 0.001, "loss": 2.2779, "step": 60100 }, { "epoch": 19.457013574660632, "grad_norm": 1.6978230476379395, "learning_rate": 0.001, "loss": 2.2626, "step": 60200 }, { "epoch": 19.489334195216546, "grad_norm": 1.3387566804885864, "learning_rate": 0.001, "loss": 2.2908, "step": 60300 }, { "epoch": 19.521654815772465, "grad_norm": 1.1720887422561646, "learning_rate": 0.001, "loss": 2.2893, "step": 60400 }, { "epoch": 19.55397543632838, "grad_norm": 1.0618383884429932, "learning_rate": 0.001, "loss": 2.2833, "step": 60500 }, { "epoch": 19.586296056884294, "grad_norm": 0.9412760734558105, "learning_rate": 0.001, "loss": 2.283, "step": 60600 }, { "epoch": 19.618616677440208, "grad_norm": 1.4370046854019165, "learning_rate": 0.001, "loss": 2.2991, "step": 60700 }, { "epoch": 19.650937297996123, "grad_norm": 1.4847232103347778, "learning_rate": 0.001, "loss": 2.2973, "step": 60800 }, { "epoch": 19.683257918552037, "grad_norm": 0.9826359152793884, "learning_rate": 0.001, "loss": 2.2933, "step": 60900 }, { "epoch": 19.715578539107952, "grad_norm": 1.269861102104187, "learning_rate": 0.001, "loss": 2.3051, "step": 61000 }, { "epoch": 19.747899159663866, "grad_norm": 1.4577744007110596, "learning_rate": 0.001, "loss": 2.2907, "step": 61100 }, { "epoch": 19.78021978021978, "grad_norm": 1.3713974952697754, "learning_rate": 0.001, "loss": 2.2964, "step": 61200 }, { "epoch": 19.812540400775696, "grad_norm": 0.8531525731086731, "learning_rate": 0.001, "loss": 2.322, "step": 61300 }, { "epoch": 19.84486102133161, "grad_norm": 0.8501541614532471, "learning_rate": 0.001, "loss": 2.2948, "step": 61400 }, { "epoch": 19.877181641887525, "grad_norm": 1.437203288078308, "learning_rate": 0.001, "loss": 2.3299, "step": 61500 }, { "epoch": 19.90950226244344, "grad_norm": 1.27197265625, "learning_rate": 0.001, "loss": 2.3322, "step": 61600 }, { "epoch": 19.941822882999354, "grad_norm": 0.863344132900238, "learning_rate": 0.001, "loss": 2.3453, "step": 61700 }, { "epoch": 19.97414350355527, "grad_norm": 1.3954321146011353, "learning_rate": 0.001, "loss": 2.3386, "step": 61800 }, { "epoch": 20.006464124111183, "grad_norm": 1.2504416704177856, "learning_rate": 0.001, "loss": 2.3092, "step": 61900 }, { "epoch": 20.038784744667097, "grad_norm": 1.1983377933502197, "learning_rate": 0.001, "loss": 2.2074, "step": 62000 }, { "epoch": 20.071105365223012, "grad_norm": 1.1122028827667236, "learning_rate": 0.001, "loss": 2.1786, "step": 62100 }, { "epoch": 20.103425985778927, "grad_norm": 1.2796459197998047, "learning_rate": 0.001, "loss": 2.2029, "step": 62200 }, { "epoch": 20.13574660633484, "grad_norm": 1.1963897943496704, "learning_rate": 0.001, "loss": 2.1912, "step": 62300 }, { "epoch": 20.168067226890756, "grad_norm": 1.084876537322998, "learning_rate": 0.001, "loss": 2.2138, "step": 62400 }, { "epoch": 20.20038784744667, "grad_norm": 1.357012391090393, "learning_rate": 0.001, "loss": 2.2203, "step": 62500 }, { "epoch": 20.232708468002585, "grad_norm": 1.0108733177185059, "learning_rate": 0.001, "loss": 2.2175, "step": 62600 }, { "epoch": 20.2650290885585, "grad_norm": 1.3037816286087036, "learning_rate": 0.001, "loss": 2.2083, "step": 62700 }, { "epoch": 20.297349709114414, "grad_norm": 1.2153654098510742, "learning_rate": 0.001, "loss": 2.2167, "step": 62800 }, { "epoch": 20.32967032967033, "grad_norm": 1.1544158458709717, "learning_rate": 0.001, "loss": 2.2183, "step": 62900 }, { "epoch": 20.361990950226243, "grad_norm": 1.18621826171875, "learning_rate": 0.001, "loss": 2.217, "step": 63000 }, { "epoch": 20.394311570782158, "grad_norm": 1.1521902084350586, "learning_rate": 0.001, "loss": 2.2508, "step": 63100 }, { "epoch": 20.426632191338072, "grad_norm": 1.1992689371109009, "learning_rate": 0.001, "loss": 2.2515, "step": 63200 }, { "epoch": 20.458952811893987, "grad_norm": 1.1839091777801514, "learning_rate": 0.001, "loss": 2.2541, "step": 63300 }, { "epoch": 20.4912734324499, "grad_norm": 1.1625888347625732, "learning_rate": 0.001, "loss": 2.2538, "step": 63400 }, { "epoch": 20.52359405300582, "grad_norm": 1.188076376914978, "learning_rate": 0.001, "loss": 2.2535, "step": 63500 }, { "epoch": 20.555914673561734, "grad_norm": 1.0718660354614258, "learning_rate": 0.001, "loss": 2.2624, "step": 63600 }, { "epoch": 20.58823529411765, "grad_norm": 1.2299270629882812, "learning_rate": 0.001, "loss": 2.2516, "step": 63700 }, { "epoch": 20.620555914673563, "grad_norm": 1.1382802724838257, "learning_rate": 0.001, "loss": 2.2649, "step": 63800 }, { "epoch": 20.652876535229478, "grad_norm": 1.2164167165756226, "learning_rate": 0.001, "loss": 2.2592, "step": 63900 }, { "epoch": 20.685197155785392, "grad_norm": 1.0841768980026245, "learning_rate": 0.001, "loss": 2.2691, "step": 64000 }, { "epoch": 20.717517776341307, "grad_norm": 1.0041121244430542, "learning_rate": 0.001, "loss": 2.2783, "step": 64100 }, { "epoch": 20.74983839689722, "grad_norm": 1.1556775569915771, "learning_rate": 0.001, "loss": 2.2603, "step": 64200 }, { "epoch": 20.782159017453136, "grad_norm": 0.9205794930458069, "learning_rate": 0.001, "loss": 2.3004, "step": 64300 }, { "epoch": 20.81447963800905, "grad_norm": 1.3644750118255615, "learning_rate": 0.001, "loss": 2.2996, "step": 64400 }, { "epoch": 20.846800258564965, "grad_norm": 0.9331674575805664, "learning_rate": 0.001, "loss": 2.2768, "step": 64500 }, { "epoch": 20.87912087912088, "grad_norm": 1.0455474853515625, "learning_rate": 0.001, "loss": 2.2781, "step": 64600 }, { "epoch": 20.911441499676794, "grad_norm": 1.4594848155975342, "learning_rate": 0.001, "loss": 2.2765, "step": 64700 }, { "epoch": 20.94376212023271, "grad_norm": 1.0385990142822266, "learning_rate": 0.001, "loss": 2.2935, "step": 64800 }, { "epoch": 20.976082740788623, "grad_norm": 1.577081561088562, "learning_rate": 0.001, "loss": 2.2826, "step": 64900 }, { "epoch": 21.008403361344538, "grad_norm": 1.0930685997009277, "learning_rate": 0.001, "loss": 2.2722, "step": 65000 }, { "epoch": 21.040723981900452, "grad_norm": 1.1612266302108765, "learning_rate": 0.001, "loss": 2.1529, "step": 65100 }, { "epoch": 21.073044602456367, "grad_norm": 1.321443796157837, "learning_rate": 0.001, "loss": 2.1513, "step": 65200 }, { "epoch": 21.10536522301228, "grad_norm": 0.8242465257644653, "learning_rate": 0.001, "loss": 2.1593, "step": 65300 }, { "epoch": 21.137685843568196, "grad_norm": 1.2424418926239014, "learning_rate": 0.001, "loss": 2.1824, "step": 65400 }, { "epoch": 21.17000646412411, "grad_norm": 1.1042993068695068, "learning_rate": 0.001, "loss": 2.1891, "step": 65500 }, { "epoch": 21.202327084680025, "grad_norm": 0.9918628334999084, "learning_rate": 0.001, "loss": 2.1892, "step": 65600 }, { "epoch": 21.23464770523594, "grad_norm": 1.0536214113235474, "learning_rate": 0.001, "loss": 2.1884, "step": 65700 }, { "epoch": 21.266968325791854, "grad_norm": 1.1322977542877197, "learning_rate": 0.001, "loss": 2.1944, "step": 65800 }, { "epoch": 21.29928894634777, "grad_norm": 1.2656852006912231, "learning_rate": 0.001, "loss": 2.1963, "step": 65900 }, { "epoch": 21.331609566903683, "grad_norm": 1.2672237157821655, "learning_rate": 0.001, "loss": 2.2046, "step": 66000 }, { "epoch": 21.363930187459598, "grad_norm": 1.2874271869659424, "learning_rate": 0.001, "loss": 2.1938, "step": 66100 }, { "epoch": 21.396250808015512, "grad_norm": 0.881458044052124, "learning_rate": 0.001, "loss": 2.1936, "step": 66200 }, { "epoch": 21.428571428571427, "grad_norm": 1.2030190229415894, "learning_rate": 0.001, "loss": 2.2008, "step": 66300 }, { "epoch": 21.46089204912734, "grad_norm": 0.9939525127410889, "learning_rate": 0.001, "loss": 2.2307, "step": 66400 }, { "epoch": 21.49321266968326, "grad_norm": 0.927545428276062, "learning_rate": 0.001, "loss": 2.2114, "step": 66500 }, { "epoch": 21.525533290239174, "grad_norm": 1.0802602767944336, "learning_rate": 0.001, "loss": 2.2072, "step": 66600 }, { "epoch": 21.55785391079509, "grad_norm": 1.0599498748779297, "learning_rate": 0.001, "loss": 2.2306, "step": 66700 }, { "epoch": 21.590174531351003, "grad_norm": 1.3177958726882935, "learning_rate": 0.001, "loss": 2.2166, "step": 66800 }, { "epoch": 21.622495151906918, "grad_norm": 1.34926438331604, "learning_rate": 0.001, "loss": 2.2222, "step": 66900 }, { "epoch": 21.654815772462833, "grad_norm": 1.0225821733474731, "learning_rate": 0.001, "loss": 2.247, "step": 67000 }, { "epoch": 21.687136393018747, "grad_norm": 1.190183401107788, "learning_rate": 0.001, "loss": 2.2398, "step": 67100 }, { "epoch": 21.71945701357466, "grad_norm": 1.215757131576538, "learning_rate": 0.001, "loss": 2.2526, "step": 67200 }, { "epoch": 21.751777634130576, "grad_norm": 1.2403733730316162, "learning_rate": 0.001, "loss": 2.2506, "step": 67300 }, { "epoch": 21.78409825468649, "grad_norm": 1.2998617887496948, "learning_rate": 0.001, "loss": 2.2337, "step": 67400 }, { "epoch": 21.816418875242405, "grad_norm": 0.9788816571235657, "learning_rate": 0.001, "loss": 2.2501, "step": 67500 }, { "epoch": 21.84873949579832, "grad_norm": 1.0035640001296997, "learning_rate": 0.001, "loss": 2.2456, "step": 67600 }, { "epoch": 21.881060116354234, "grad_norm": 0.9567862749099731, "learning_rate": 0.001, "loss": 2.2563, "step": 67700 }, { "epoch": 21.91338073691015, "grad_norm": 1.2522963285446167, "learning_rate": 0.001, "loss": 2.2691, "step": 67800 }, { "epoch": 21.945701357466064, "grad_norm": 1.3323355913162231, "learning_rate": 0.001, "loss": 2.2682, "step": 67900 }, { "epoch": 21.978021978021978, "grad_norm": 1.022301197052002, "learning_rate": 0.001, "loss": 2.2829, "step": 68000 }, { "epoch": 22.010342598577893, "grad_norm": 0.9632961750030518, "learning_rate": 0.001, "loss": 2.2159, "step": 68100 }, { "epoch": 22.042663219133807, "grad_norm": 0.9349167346954346, "learning_rate": 0.001, "loss": 2.1377, "step": 68200 }, { "epoch": 22.07498383968972, "grad_norm": 1.2619801759719849, "learning_rate": 0.001, "loss": 2.1479, "step": 68300 }, { "epoch": 22.107304460245636, "grad_norm": 1.0707018375396729, "learning_rate": 0.001, "loss": 2.1212, "step": 68400 }, { "epoch": 22.13962508080155, "grad_norm": 1.2993378639221191, "learning_rate": 0.001, "loss": 2.1276, "step": 68500 }, { "epoch": 22.171945701357465, "grad_norm": 1.2283509969711304, "learning_rate": 0.001, "loss": 2.1508, "step": 68600 }, { "epoch": 22.20426632191338, "grad_norm": 1.0965924263000488, "learning_rate": 0.001, "loss": 2.1642, "step": 68700 }, { "epoch": 22.236586942469295, "grad_norm": 1.5436381101608276, "learning_rate": 0.001, "loss": 2.1672, "step": 68800 }, { "epoch": 22.26890756302521, "grad_norm": 1.2528314590454102, "learning_rate": 0.001, "loss": 2.1684, "step": 68900 }, { "epoch": 22.301228183581124, "grad_norm": 1.1026721000671387, "learning_rate": 0.001, "loss": 2.1799, "step": 69000 }, { "epoch": 22.33354880413704, "grad_norm": 1.2602219581604004, "learning_rate": 0.001, "loss": 2.1647, "step": 69100 }, { "epoch": 22.365869424692953, "grad_norm": 1.118823528289795, "learning_rate": 0.001, "loss": 2.1881, "step": 69200 }, { "epoch": 22.398190045248867, "grad_norm": 1.1456882953643799, "learning_rate": 0.001, "loss": 2.1736, "step": 69300 }, { "epoch": 22.430510665804782, "grad_norm": 0.9726011157035828, "learning_rate": 0.001, "loss": 2.1651, "step": 69400 }, { "epoch": 22.462831286360696, "grad_norm": 0.9580657482147217, "learning_rate": 0.001, "loss": 2.1747, "step": 69500 }, { "epoch": 22.49515190691661, "grad_norm": 1.0189781188964844, "learning_rate": 0.001, "loss": 2.1691, "step": 69600 }, { "epoch": 22.52747252747253, "grad_norm": 0.9164172410964966, "learning_rate": 0.001, "loss": 2.1956, "step": 69700 }, { "epoch": 22.559793148028444, "grad_norm": 1.3433533906936646, "learning_rate": 0.001, "loss": 2.1881, "step": 69800 }, { "epoch": 22.59211376858436, "grad_norm": 1.2436511516571045, "learning_rate": 0.001, "loss": 2.1895, "step": 69900 }, { "epoch": 22.624434389140273, "grad_norm": 1.0203531980514526, "learning_rate": 0.001, "loss": 2.2088, "step": 70000 }, { "epoch": 22.656755009696187, "grad_norm": 0.9450057744979858, "learning_rate": 0.001, "loss": 2.2119, "step": 70100 }, { "epoch": 22.689075630252102, "grad_norm": 1.4776325225830078, "learning_rate": 0.001, "loss": 2.2145, "step": 70200 }, { "epoch": 22.721396250808017, "grad_norm": 1.3227741718292236, "learning_rate": 0.001, "loss": 2.2034, "step": 70300 }, { "epoch": 22.75371687136393, "grad_norm": 1.2652865648269653, "learning_rate": 0.001, "loss": 2.2091, "step": 70400 }, { "epoch": 22.786037491919846, "grad_norm": 1.1568591594696045, "learning_rate": 0.001, "loss": 2.2176, "step": 70500 }, { "epoch": 22.81835811247576, "grad_norm": 1.1600966453552246, "learning_rate": 0.001, "loss": 2.2186, "step": 70600 }, { "epoch": 22.850678733031675, "grad_norm": 1.0852841138839722, "learning_rate": 0.001, "loss": 2.2318, "step": 70700 }, { "epoch": 22.88299935358759, "grad_norm": 1.266211748123169, "learning_rate": 0.001, "loss": 2.2282, "step": 70800 }, { "epoch": 22.915319974143504, "grad_norm": 1.2288943529129028, "learning_rate": 0.001, "loss": 2.228, "step": 70900 }, { "epoch": 22.94764059469942, "grad_norm": 1.3902039527893066, "learning_rate": 0.001, "loss": 2.2269, "step": 71000 }, { "epoch": 22.979961215255333, "grad_norm": 0.937089741230011, "learning_rate": 0.001, "loss": 2.2499, "step": 71100 }, { "epoch": 23.012281835811248, "grad_norm": 1.1552362442016602, "learning_rate": 0.001, "loss": 2.1682, "step": 71200 }, { "epoch": 23.044602456367162, "grad_norm": 1.3249483108520508, "learning_rate": 0.001, "loss": 2.0807, "step": 71300 }, { "epoch": 23.076923076923077, "grad_norm": 1.5028316974639893, "learning_rate": 0.001, "loss": 2.1071, "step": 71400 }, { "epoch": 23.10924369747899, "grad_norm": 1.7541735172271729, "learning_rate": 0.001, "loss": 2.1075, "step": 71500 }, { "epoch": 23.141564318034906, "grad_norm": 1.3183276653289795, "learning_rate": 0.001, "loss": 2.1169, "step": 71600 }, { "epoch": 23.17388493859082, "grad_norm": 1.4954915046691895, "learning_rate": 0.001, "loss": 2.1122, "step": 71700 }, { "epoch": 23.206205559146735, "grad_norm": 1.128176212310791, "learning_rate": 0.001, "loss": 2.1316, "step": 71800 }, { "epoch": 23.23852617970265, "grad_norm": 1.1584060192108154, "learning_rate": 0.001, "loss": 2.1386, "step": 71900 }, { "epoch": 23.270846800258564, "grad_norm": 1.363810658454895, "learning_rate": 0.001, "loss": 2.1218, "step": 72000 }, { "epoch": 23.30316742081448, "grad_norm": 1.2617250680923462, "learning_rate": 0.001, "loss": 2.1395, "step": 72100 }, { "epoch": 23.335488041370393, "grad_norm": 1.2358677387237549, "learning_rate": 0.001, "loss": 2.1483, "step": 72200 }, { "epoch": 23.367808661926308, "grad_norm": 1.4727267026901245, "learning_rate": 0.001, "loss": 2.1461, "step": 72300 }, { "epoch": 23.400129282482222, "grad_norm": 1.2978276014328003, "learning_rate": 0.001, "loss": 2.164, "step": 72400 }, { "epoch": 23.432449903038137, "grad_norm": 1.1162272691726685, "learning_rate": 0.001, "loss": 2.1393, "step": 72500 }, { "epoch": 23.46477052359405, "grad_norm": 1.5267865657806396, "learning_rate": 0.001, "loss": 2.1532, "step": 72600 }, { "epoch": 23.49709114414997, "grad_norm": 1.349844217300415, "learning_rate": 0.001, "loss": 2.1577, "step": 72700 }, { "epoch": 23.529411764705884, "grad_norm": 1.7891136407852173, "learning_rate": 0.001, "loss": 2.1496, "step": 72800 }, { "epoch": 23.5617323852618, "grad_norm": 1.613793134689331, "learning_rate": 0.001, "loss": 2.162, "step": 72900 }, { "epoch": 23.594053005817713, "grad_norm": 1.121187686920166, "learning_rate": 0.001, "loss": 2.159, "step": 73000 }, { "epoch": 23.626373626373628, "grad_norm": 1.2575411796569824, "learning_rate": 0.001, "loss": 2.1651, "step": 73100 }, { "epoch": 23.658694246929542, "grad_norm": 1.5651819705963135, "learning_rate": 0.001, "loss": 2.1736, "step": 73200 }, { "epoch": 23.691014867485457, "grad_norm": 1.4974477291107178, "learning_rate": 0.001, "loss": 2.1858, "step": 73300 }, { "epoch": 23.72333548804137, "grad_norm": 1.1631637811660767, "learning_rate": 0.001, "loss": 2.1891, "step": 73400 }, { "epoch": 23.755656108597286, "grad_norm": 1.5709234476089478, "learning_rate": 0.001, "loss": 2.1827, "step": 73500 }, { "epoch": 23.7879767291532, "grad_norm": 1.447441577911377, "learning_rate": 0.001, "loss": 2.2088, "step": 73600 }, { "epoch": 23.820297349709115, "grad_norm": 1.2315850257873535, "learning_rate": 0.001, "loss": 2.2009, "step": 73700 }, { "epoch": 23.85261797026503, "grad_norm": 1.4171655178070068, "learning_rate": 0.001, "loss": 2.1866, "step": 73800 }, { "epoch": 23.884938590820944, "grad_norm": 1.5399742126464844, "learning_rate": 0.001, "loss": 2.1967, "step": 73900 }, { "epoch": 23.91725921137686, "grad_norm": 1.135227918624878, "learning_rate": 0.001, "loss": 2.1932, "step": 74000 }, { "epoch": 23.949579831932773, "grad_norm": 1.1006965637207031, "learning_rate": 0.001, "loss": 2.2132, "step": 74100 }, { "epoch": 23.981900452488688, "grad_norm": 1.530867576599121, "learning_rate": 0.001, "loss": 2.2207, "step": 74200 }, { "epoch": 24.014221073044602, "grad_norm": 1.686764121055603, "learning_rate": 0.001, "loss": 2.1332, "step": 74300 }, { "epoch": 24.046541693600517, "grad_norm": 1.4816681146621704, "learning_rate": 0.001, "loss": 2.0556, "step": 74400 }, { "epoch": 24.07886231415643, "grad_norm": 1.2083070278167725, "learning_rate": 0.001, "loss": 2.0783, "step": 74500 }, { "epoch": 24.111182934712346, "grad_norm": 1.873100757598877, "learning_rate": 0.001, "loss": 2.0883, "step": 74600 }, { "epoch": 24.14350355526826, "grad_norm": 1.3434606790542603, "learning_rate": 0.001, "loss": 2.0975, "step": 74700 }, { "epoch": 24.175824175824175, "grad_norm": 1.4237688779830933, "learning_rate": 0.001, "loss": 2.0818, "step": 74800 }, { "epoch": 24.20814479638009, "grad_norm": 1.3739618062973022, "learning_rate": 0.001, "loss": 2.0801, "step": 74900 }, { "epoch": 24.240465416936004, "grad_norm": 1.2694149017333984, "learning_rate": 0.001, "loss": 2.1075, "step": 75000 }, { "epoch": 24.27278603749192, "grad_norm": 1.557441234588623, "learning_rate": 0.001, "loss": 2.0868, "step": 75100 }, { "epoch": 24.305106658047833, "grad_norm": 1.2402006387710571, "learning_rate": 0.001, "loss": 2.1188, "step": 75200 }, { "epoch": 24.337427278603748, "grad_norm": 1.2395163774490356, "learning_rate": 0.001, "loss": 2.1081, "step": 75300 }, { "epoch": 24.369747899159663, "grad_norm": 1.3019064664840698, "learning_rate": 0.001, "loss": 2.1262, "step": 75400 }, { "epoch": 24.402068519715577, "grad_norm": 1.385576605796814, "learning_rate": 0.001, "loss": 2.1431, "step": 75500 }, { "epoch": 24.43438914027149, "grad_norm": 1.610942006111145, "learning_rate": 0.001, "loss": 2.1083, "step": 75600 }, { "epoch": 24.466709760827406, "grad_norm": 1.2178122997283936, "learning_rate": 0.001, "loss": 2.1371, "step": 75700 }, { "epoch": 24.49903038138332, "grad_norm": 1.3105348348617554, "learning_rate": 0.001, "loss": 2.1256, "step": 75800 }, { "epoch": 24.53135100193924, "grad_norm": 1.3397716283798218, "learning_rate": 0.001, "loss": 2.1286, "step": 75900 }, { "epoch": 24.563671622495153, "grad_norm": 1.9577621221542358, "learning_rate": 0.001, "loss": 2.1523, "step": 76000 }, { "epoch": 24.595992243051068, "grad_norm": 1.2940340042114258, "learning_rate": 0.001, "loss": 2.1451, "step": 76100 }, { "epoch": 24.628312863606983, "grad_norm": 1.330095648765564, "learning_rate": 0.001, "loss": 2.1563, "step": 76200 }, { "epoch": 24.660633484162897, "grad_norm": 1.4415050745010376, "learning_rate": 0.001, "loss": 2.1577, "step": 76300 }, { "epoch": 24.69295410471881, "grad_norm": 1.3965532779693604, "learning_rate": 0.001, "loss": 2.1528, "step": 76400 }, { "epoch": 24.725274725274726, "grad_norm": 1.4386045932769775, "learning_rate": 0.001, "loss": 2.1542, "step": 76500 }, { "epoch": 24.75759534583064, "grad_norm": 1.6891236305236816, "learning_rate": 0.001, "loss": 2.1646, "step": 76600 }, { "epoch": 24.789915966386555, "grad_norm": 1.355452537536621, "learning_rate": 0.001, "loss": 2.1637, "step": 76700 }, { "epoch": 24.82223658694247, "grad_norm": 1.5033726692199707, "learning_rate": 0.001, "loss": 2.1679, "step": 76800 }, { "epoch": 24.854557207498384, "grad_norm": 1.8620679378509521, "learning_rate": 0.001, "loss": 2.152, "step": 76900 }, { "epoch": 24.8868778280543, "grad_norm": 1.3260308504104614, "learning_rate": 0.001, "loss": 2.1783, "step": 77000 }, { "epoch": 24.919198448610214, "grad_norm": 1.2607606649398804, "learning_rate": 0.001, "loss": 2.1856, "step": 77100 }, { "epoch": 24.951519069166128, "grad_norm": 1.466381311416626, "learning_rate": 0.001, "loss": 2.1857, "step": 77200 }, { "epoch": 24.983839689722043, "grad_norm": 1.292051911354065, "learning_rate": 0.001, "loss": 2.1806, "step": 77300 }, { "epoch": 25.016160310277957, "grad_norm": 1.7954028844833374, "learning_rate": 0.001, "loss": 2.0906, "step": 77400 }, { "epoch": 25.048480930833872, "grad_norm": 1.6722086668014526, "learning_rate": 0.001, "loss": 2.0501, "step": 77500 }, { "epoch": 25.080801551389786, "grad_norm": 2.243319272994995, "learning_rate": 0.001, "loss": 2.0422, "step": 77600 }, { "epoch": 25.1131221719457, "grad_norm": 2.0324037075042725, "learning_rate": 0.001, "loss": 2.064, "step": 77700 }, { "epoch": 25.145442792501616, "grad_norm": 1.8791396617889404, "learning_rate": 0.001, "loss": 2.0754, "step": 77800 }, { "epoch": 25.17776341305753, "grad_norm": 2.5221776962280273, "learning_rate": 0.001, "loss": 2.0617, "step": 77900 }, { "epoch": 25.210084033613445, "grad_norm": 2.2298007011413574, "learning_rate": 0.001, "loss": 2.059, "step": 78000 }, { "epoch": 25.24240465416936, "grad_norm": 2.844670057296753, "learning_rate": 0.001, "loss": 2.055, "step": 78100 }, { "epoch": 25.274725274725274, "grad_norm": 1.8920618295669556, "learning_rate": 0.001, "loss": 2.1047, "step": 78200 }, { "epoch": 25.30704589528119, "grad_norm": 2.7440826892852783, "learning_rate": 0.001, "loss": 2.0731, "step": 78300 }, { "epoch": 25.339366515837103, "grad_norm": 1.8846522569656372, "learning_rate": 0.001, "loss": 2.0671, "step": 78400 }, { "epoch": 25.371687136393017, "grad_norm": 2.325923204421997, "learning_rate": 0.001, "loss": 2.0712, "step": 78500 }, { "epoch": 25.404007756948932, "grad_norm": 1.8636107444763184, "learning_rate": 0.001, "loss": 2.0903, "step": 78600 }, { "epoch": 25.436328377504847, "grad_norm": 1.8899035453796387, "learning_rate": 0.001, "loss": 2.0982, "step": 78700 }, { "epoch": 25.46864899806076, "grad_norm": 2.493457317352295, "learning_rate": 0.001, "loss": 2.1104, "step": 78800 }, { "epoch": 25.50096961861668, "grad_norm": 1.93169367313385, "learning_rate": 0.001, "loss": 2.1031, "step": 78900 }, { "epoch": 25.533290239172594, "grad_norm": 2.1825897693634033, "learning_rate": 0.001, "loss": 2.1127, "step": 79000 }, { "epoch": 25.56561085972851, "grad_norm": 1.7733447551727295, "learning_rate": 0.001, "loss": 2.1179, "step": 79100 }, { "epoch": 25.597931480284423, "grad_norm": 1.6951693296432495, "learning_rate": 0.001, "loss": 2.1353, "step": 79200 }, { "epoch": 25.630252100840337, "grad_norm": 1.6487561464309692, "learning_rate": 0.001, "loss": 2.1199, "step": 79300 }, { "epoch": 25.662572721396252, "grad_norm": 2.3216216564178467, "learning_rate": 0.001, "loss": 2.1412, "step": 79400 }, { "epoch": 25.694893341952167, "grad_norm": 2.1835920810699463, "learning_rate": 0.001, "loss": 2.1387, "step": 79500 }, { "epoch": 25.72721396250808, "grad_norm": 2.5037333965301514, "learning_rate": 0.001, "loss": 2.1345, "step": 79600 }, { "epoch": 25.759534583063996, "grad_norm": 1.924839735031128, "learning_rate": 0.001, "loss": 2.1384, "step": 79700 }, { "epoch": 25.79185520361991, "grad_norm": 2.1649904251098633, "learning_rate": 0.001, "loss": 2.1526, "step": 79800 }, { "epoch": 25.824175824175825, "grad_norm": 1.911528468132019, "learning_rate": 0.001, "loss": 2.1241, "step": 79900 }, { "epoch": 25.85649644473174, "grad_norm": 1.9926867485046387, "learning_rate": 0.001, "loss": 2.1524, "step": 80000 }, { "epoch": 25.888817065287654, "grad_norm": 1.946124792098999, "learning_rate": 0.001, "loss": 2.1363, "step": 80100 }, { "epoch": 25.92113768584357, "grad_norm": 1.9511570930480957, "learning_rate": 0.001, "loss": 2.1686, "step": 80200 }, { "epoch": 25.953458306399483, "grad_norm": 2.4123759269714355, "learning_rate": 0.001, "loss": 2.1474, "step": 80300 }, { "epoch": 25.985778926955398, "grad_norm": 1.9220401048660278, "learning_rate": 0.001, "loss": 2.1668, "step": 80400 }, { "epoch": 26.018099547511312, "grad_norm": 1.3430564403533936, "learning_rate": 0.001, "loss": 2.0692, "step": 80500 }, { "epoch": 26.050420168067227, "grad_norm": 1.3940428495407104, "learning_rate": 0.001, "loss": 1.9975, "step": 80600 }, { "epoch": 26.08274078862314, "grad_norm": 1.4681583642959595, "learning_rate": 0.001, "loss": 2.0075, "step": 80700 }, { "epoch": 26.115061409179056, "grad_norm": 0.9912952780723572, "learning_rate": 0.001, "loss": 2.0133, "step": 80800 }, { "epoch": 26.14738202973497, "grad_norm": 1.3190398216247559, "learning_rate": 0.001, "loss": 2.0483, "step": 80900 }, { "epoch": 26.179702650290885, "grad_norm": 1.2165215015411377, "learning_rate": 0.001, "loss": 2.0469, "step": 81000 }, { "epoch": 26.2120232708468, "grad_norm": 1.2185070514678955, "learning_rate": 0.001, "loss": 2.0529, "step": 81100 }, { "epoch": 26.244343891402714, "grad_norm": 1.1104484796524048, "learning_rate": 0.001, "loss": 2.0501, "step": 81200 }, { "epoch": 26.27666451195863, "grad_norm": 1.1159557104110718, "learning_rate": 0.001, "loss": 2.056, "step": 81300 }, { "epoch": 26.308985132514543, "grad_norm": 1.1061629056930542, "learning_rate": 0.001, "loss": 2.072, "step": 81400 }, { "epoch": 26.341305753070458, "grad_norm": 1.1867953538894653, "learning_rate": 0.001, "loss": 2.075, "step": 81500 }, { "epoch": 26.373626373626372, "grad_norm": 1.3672994375228882, "learning_rate": 0.001, "loss": 2.0571, "step": 81600 }, { "epoch": 26.405946994182287, "grad_norm": 1.466471791267395, "learning_rate": 0.001, "loss": 2.0804, "step": 81700 }, { "epoch": 26.4382676147382, "grad_norm": 1.4500887393951416, "learning_rate": 0.001, "loss": 2.0849, "step": 81800 }, { "epoch": 26.470588235294116, "grad_norm": 1.4271094799041748, "learning_rate": 0.001, "loss": 2.074, "step": 81900 }, { "epoch": 26.50290885585003, "grad_norm": 1.4136507511138916, "learning_rate": 0.001, "loss": 2.0754, "step": 82000 }, { "epoch": 26.53522947640595, "grad_norm": 1.851994276046753, "learning_rate": 0.001, "loss": 2.1002, "step": 82100 }, { "epoch": 26.567550096961863, "grad_norm": 1.2131636142730713, "learning_rate": 0.001, "loss": 2.0822, "step": 82200 }, { "epoch": 26.599870717517778, "grad_norm": 1.9536834955215454, "learning_rate": 0.001, "loss": 2.0967, "step": 82300 }, { "epoch": 26.632191338073692, "grad_norm": 1.3947795629501343, "learning_rate": 0.001, "loss": 2.0994, "step": 82400 }, { "epoch": 26.664511958629607, "grad_norm": 1.3950212001800537, "learning_rate": 0.001, "loss": 2.1009, "step": 82500 }, { "epoch": 26.69683257918552, "grad_norm": 1.3273499011993408, "learning_rate": 0.001, "loss": 2.1082, "step": 82600 }, { "epoch": 26.729153199741436, "grad_norm": 1.3609721660614014, "learning_rate": 0.001, "loss": 2.1004, "step": 82700 }, { "epoch": 26.76147382029735, "grad_norm": 1.6250184774398804, "learning_rate": 0.001, "loss": 2.1293, "step": 82800 }, { "epoch": 26.793794440853265, "grad_norm": 1.179201602935791, "learning_rate": 0.001, "loss": 2.1137, "step": 82900 }, { "epoch": 26.82611506140918, "grad_norm": 1.2400749921798706, "learning_rate": 0.001, "loss": 2.1199, "step": 83000 }, { "epoch": 26.858435681965094, "grad_norm": 1.4110934734344482, "learning_rate": 0.001, "loss": 2.1298, "step": 83100 }, { "epoch": 26.89075630252101, "grad_norm": 1.213017463684082, "learning_rate": 0.001, "loss": 2.1216, "step": 83200 }, { "epoch": 26.923076923076923, "grad_norm": 1.3616845607757568, "learning_rate": 0.001, "loss": 2.1267, "step": 83300 }, { "epoch": 26.955397543632838, "grad_norm": 1.1601148843765259, "learning_rate": 0.001, "loss": 2.1222, "step": 83400 }, { "epoch": 26.987718164188752, "grad_norm": 1.6087828874588013, "learning_rate": 0.001, "loss": 2.121, "step": 83500 }, { "epoch": 27.020038784744667, "grad_norm": 1.7387524843215942, "learning_rate": 0.001, "loss": 2.0578, "step": 83600 }, { "epoch": 27.05235940530058, "grad_norm": 1.2267652750015259, "learning_rate": 0.001, "loss": 1.9945, "step": 83700 }, { "epoch": 27.084680025856496, "grad_norm": 1.6405795812606812, "learning_rate": 0.001, "loss": 2.0002, "step": 83800 }, { "epoch": 27.11700064641241, "grad_norm": 1.1666873693466187, "learning_rate": 0.001, "loss": 2.015, "step": 83900 }, { "epoch": 27.149321266968325, "grad_norm": 1.1755584478378296, "learning_rate": 0.001, "loss": 2.0102, "step": 84000 }, { "epoch": 27.18164188752424, "grad_norm": 1.324985384941101, "learning_rate": 0.001, "loss": 2.0359, "step": 84100 }, { "epoch": 27.213962508080154, "grad_norm": 1.079712152481079, "learning_rate": 0.001, "loss": 1.9905, "step": 84200 }, { "epoch": 27.24628312863607, "grad_norm": 1.401971697807312, "learning_rate": 0.001, "loss": 2.0207, "step": 84300 }, { "epoch": 27.278603749191983, "grad_norm": 1.2873399257659912, "learning_rate": 0.001, "loss": 2.0211, "step": 84400 }, { "epoch": 27.310924369747898, "grad_norm": 1.115574598312378, "learning_rate": 0.001, "loss": 2.0332, "step": 84500 }, { "epoch": 27.343244990303813, "grad_norm": 1.2931740283966064, "learning_rate": 0.001, "loss": 2.0606, "step": 84600 }, { "epoch": 27.375565610859727, "grad_norm": 1.0610228776931763, "learning_rate": 0.001, "loss": 2.0494, "step": 84700 }, { "epoch": 27.40788623141564, "grad_norm": 1.0473644733428955, "learning_rate": 0.001, "loss": 2.0388, "step": 84800 }, { "epoch": 27.440206851971556, "grad_norm": 1.187235713005066, "learning_rate": 0.001, "loss": 2.0494, "step": 84900 }, { "epoch": 27.47252747252747, "grad_norm": 1.115997076034546, "learning_rate": 0.001, "loss": 2.0444, "step": 85000 }, { "epoch": 27.50484809308339, "grad_norm": 1.2527693510055542, "learning_rate": 0.001, "loss": 2.0718, "step": 85100 }, { "epoch": 27.537168713639304, "grad_norm": 1.5422104597091675, "learning_rate": 0.001, "loss": 2.0718, "step": 85200 }, { "epoch": 27.569489334195218, "grad_norm": 1.2862259149551392, "learning_rate": 0.001, "loss": 2.0728, "step": 85300 }, { "epoch": 27.601809954751133, "grad_norm": 1.295546293258667, "learning_rate": 0.001, "loss": 2.0683, "step": 85400 }, { "epoch": 27.634130575307047, "grad_norm": 1.2638643980026245, "learning_rate": 0.001, "loss": 2.0751, "step": 85500 }, { "epoch": 27.66645119586296, "grad_norm": 1.4461342096328735, "learning_rate": 0.001, "loss": 2.071, "step": 85600 }, { "epoch": 27.698771816418876, "grad_norm": 1.4849462509155273, "learning_rate": 0.001, "loss": 2.0922, "step": 85700 }, { "epoch": 27.73109243697479, "grad_norm": 1.451640009880066, "learning_rate": 0.001, "loss": 2.0789, "step": 85800 }, { "epoch": 27.763413057530705, "grad_norm": 1.152021884918213, "learning_rate": 0.001, "loss": 2.0748, "step": 85900 }, { "epoch": 27.79573367808662, "grad_norm": 1.290688395500183, "learning_rate": 0.001, "loss": 2.0829, "step": 86000 }, { "epoch": 27.828054298642535, "grad_norm": 1.3290468454360962, "learning_rate": 0.001, "loss": 2.0807, "step": 86100 }, { "epoch": 27.86037491919845, "grad_norm": 1.049330472946167, "learning_rate": 0.001, "loss": 2.0989, "step": 86200 }, { "epoch": 27.892695539754364, "grad_norm": 1.2603390216827393, "learning_rate": 0.001, "loss": 2.0839, "step": 86300 }, { "epoch": 27.92501616031028, "grad_norm": 1.6097103357315063, "learning_rate": 0.001, "loss": 2.0925, "step": 86400 }, { "epoch": 27.957336780866193, "grad_norm": 1.294921875, "learning_rate": 0.001, "loss": 2.1088, "step": 86500 }, { "epoch": 27.989657401422107, "grad_norm": 1.0576322078704834, "learning_rate": 0.001, "loss": 2.1014, "step": 86600 }, { "epoch": 28.021978021978022, "grad_norm": 1.2913296222686768, "learning_rate": 0.001, "loss": 2.0329, "step": 86700 }, { "epoch": 28.054298642533936, "grad_norm": 1.1250892877578735, "learning_rate": 0.001, "loss": 1.9698, "step": 86800 }, { "epoch": 28.08661926308985, "grad_norm": 1.4452911615371704, "learning_rate": 0.001, "loss": 1.9609, "step": 86900 }, { "epoch": 28.118939883645766, "grad_norm": 1.107289433479309, "learning_rate": 0.001, "loss": 1.9791, "step": 87000 }, { "epoch": 28.15126050420168, "grad_norm": 1.0976715087890625, "learning_rate": 0.001, "loss": 1.9739, "step": 87100 }, { "epoch": 28.183581124757595, "grad_norm": 1.0388271808624268, "learning_rate": 0.001, "loss": 1.9921, "step": 87200 }, { "epoch": 28.21590174531351, "grad_norm": 0.9731875061988831, "learning_rate": 0.001, "loss": 2.0035, "step": 87300 }, { "epoch": 28.248222365869424, "grad_norm": 1.587350845336914, "learning_rate": 0.001, "loss": 2.0067, "step": 87400 }, { "epoch": 28.28054298642534, "grad_norm": 1.0208741426467896, "learning_rate": 0.001, "loss": 2.0131, "step": 87500 }, { "epoch": 28.312863606981253, "grad_norm": 1.1413222551345825, "learning_rate": 0.001, "loss": 1.9982, "step": 87600 }, { "epoch": 28.345184227537167, "grad_norm": 1.7160639762878418, "learning_rate": 0.001, "loss": 1.9979, "step": 87700 }, { "epoch": 28.377504848093082, "grad_norm": 1.1650629043579102, "learning_rate": 0.001, "loss": 2.0399, "step": 87800 }, { "epoch": 28.409825468648997, "grad_norm": 1.4147566556930542, "learning_rate": 0.001, "loss": 2.0203, "step": 87900 }, { "epoch": 28.44214608920491, "grad_norm": 1.616251826286316, "learning_rate": 0.001, "loss": 2.0179, "step": 88000 }, { "epoch": 28.474466709760826, "grad_norm": 1.2545949220657349, "learning_rate": 0.001, "loss": 2.0283, "step": 88100 }, { "epoch": 28.50678733031674, "grad_norm": 1.191787838935852, "learning_rate": 0.001, "loss": 2.0339, "step": 88200 }, { "epoch": 28.53910795087266, "grad_norm": 1.0043002367019653, "learning_rate": 0.001, "loss": 2.0448, "step": 88300 }, { "epoch": 28.571428571428573, "grad_norm": 1.0854709148406982, "learning_rate": 0.001, "loss": 2.0488, "step": 88400 }, { "epoch": 28.603749191984488, "grad_norm": 1.4309149980545044, "learning_rate": 0.001, "loss": 2.0632, "step": 88500 }, { "epoch": 28.636069812540402, "grad_norm": 1.083609700202942, "learning_rate": 0.001, "loss": 2.0538, "step": 88600 }, { "epoch": 28.668390433096317, "grad_norm": 1.0403717756271362, "learning_rate": 0.001, "loss": 2.0595, "step": 88700 }, { "epoch": 28.70071105365223, "grad_norm": 1.0985143184661865, "learning_rate": 0.001, "loss": 2.0622, "step": 88800 }, { "epoch": 28.733031674208146, "grad_norm": 1.4709973335266113, "learning_rate": 0.001, "loss": 2.0665, "step": 88900 }, { "epoch": 28.76535229476406, "grad_norm": 1.333082675933838, "learning_rate": 0.001, "loss": 2.0751, "step": 89000 }, { "epoch": 28.797672915319975, "grad_norm": 1.4253976345062256, "learning_rate": 0.001, "loss": 2.0654, "step": 89100 }, { "epoch": 28.82999353587589, "grad_norm": 0.9215937852859497, "learning_rate": 0.001, "loss": 2.0716, "step": 89200 }, { "epoch": 28.862314156431804, "grad_norm": 1.0642420053482056, "learning_rate": 0.001, "loss": 2.0668, "step": 89300 }, { "epoch": 28.89463477698772, "grad_norm": 1.1206321716308594, "learning_rate": 0.001, "loss": 2.0576, "step": 89400 }, { "epoch": 28.926955397543633, "grad_norm": 1.4976261854171753, "learning_rate": 0.001, "loss": 2.0805, "step": 89500 }, { "epoch": 28.959276018099548, "grad_norm": 1.3443635702133179, "learning_rate": 0.001, "loss": 2.0714, "step": 89600 }, { "epoch": 28.991596638655462, "grad_norm": 0.9091135859489441, "learning_rate": 0.001, "loss": 2.0872, "step": 89700 }, { "epoch": 29.023917259211377, "grad_norm": 1.4851621389389038, "learning_rate": 0.001, "loss": 1.9905, "step": 89800 }, { "epoch": 29.05623787976729, "grad_norm": 2.0039730072021484, "learning_rate": 0.001, "loss": 1.9387, "step": 89900 }, { "epoch": 29.088558500323206, "grad_norm": 1.3306325674057007, "learning_rate": 0.001, "loss": 1.9439, "step": 90000 }, { "epoch": 29.12087912087912, "grad_norm": 1.3993875980377197, "learning_rate": 0.001, "loss": 1.9633, "step": 90100 }, { "epoch": 29.153199741435035, "grad_norm": 1.2553470134735107, "learning_rate": 0.001, "loss": 1.9568, "step": 90200 }, { "epoch": 29.18552036199095, "grad_norm": 1.2187873125076294, "learning_rate": 0.001, "loss": 1.9731, "step": 90300 }, { "epoch": 29.217840982546864, "grad_norm": 1.3381234407424927, "learning_rate": 0.001, "loss": 1.9704, "step": 90400 }, { "epoch": 29.25016160310278, "grad_norm": 1.2848703861236572, "learning_rate": 0.001, "loss": 1.9669, "step": 90500 }, { "epoch": 29.282482223658693, "grad_norm": 1.2281534671783447, "learning_rate": 0.001, "loss": 1.9694, "step": 90600 }, { "epoch": 29.314802844214608, "grad_norm": 1.4465069770812988, "learning_rate": 0.001, "loss": 1.9953, "step": 90700 }, { "epoch": 29.347123464770522, "grad_norm": 1.279419183731079, "learning_rate": 0.001, "loss": 1.9941, "step": 90800 }, { "epoch": 29.379444085326437, "grad_norm": 0.9530718326568604, "learning_rate": 0.001, "loss": 1.9907, "step": 90900 }, { "epoch": 29.41176470588235, "grad_norm": 1.0085426568984985, "learning_rate": 0.001, "loss": 2.0044, "step": 91000 }, { "epoch": 29.444085326438266, "grad_norm": 0.9455175995826721, "learning_rate": 0.001, "loss": 2.0158, "step": 91100 }, { "epoch": 29.47640594699418, "grad_norm": 1.0811785459518433, "learning_rate": 0.001, "loss": 2.0083, "step": 91200 }, { "epoch": 29.5087265675501, "grad_norm": 1.1927356719970703, "learning_rate": 0.001, "loss": 2.0165, "step": 91300 }, { "epoch": 29.541047188106013, "grad_norm": 1.1006433963775635, "learning_rate": 0.001, "loss": 2.0339, "step": 91400 }, { "epoch": 29.573367808661928, "grad_norm": 1.3249398469924927, "learning_rate": 0.001, "loss": 2.0158, "step": 91500 }, { "epoch": 29.605688429217842, "grad_norm": 0.9829372763633728, "learning_rate": 0.001, "loss": 2.0007, "step": 91600 }, { "epoch": 29.638009049773757, "grad_norm": 1.1615664958953857, "learning_rate": 0.001, "loss": 2.0224, "step": 91700 }, { "epoch": 29.67032967032967, "grad_norm": 0.927891194820404, "learning_rate": 0.001, "loss": 2.0409, "step": 91800 }, { "epoch": 29.702650290885586, "grad_norm": 1.3460584878921509, "learning_rate": 0.001, "loss": 2.0524, "step": 91900 }, { "epoch": 29.7349709114415, "grad_norm": 1.1224879026412964, "learning_rate": 0.001, "loss": 2.0366, "step": 92000 }, { "epoch": 29.767291531997415, "grad_norm": 1.0611379146575928, "learning_rate": 0.001, "loss": 2.0434, "step": 92100 }, { "epoch": 29.79961215255333, "grad_norm": 0.874144434928894, "learning_rate": 0.001, "loss": 2.0535, "step": 92200 }, { "epoch": 29.831932773109244, "grad_norm": 1.5935416221618652, "learning_rate": 0.001, "loss": 2.0518, "step": 92300 }, { "epoch": 29.86425339366516, "grad_norm": 1.1362969875335693, "learning_rate": 0.001, "loss": 2.0513, "step": 92400 }, { "epoch": 29.896574014221073, "grad_norm": 1.1391018629074097, "learning_rate": 0.001, "loss": 2.0559, "step": 92500 }, { "epoch": 29.928894634776988, "grad_norm": 1.5134310722351074, "learning_rate": 0.001, "loss": 2.0598, "step": 92600 }, { "epoch": 29.961215255332903, "grad_norm": 1.102142572402954, "learning_rate": 0.001, "loss": 2.0483, "step": 92700 }, { "epoch": 29.993535875888817, "grad_norm": 1.028840184211731, "learning_rate": 0.001, "loss": 2.068, "step": 92800 }, { "epoch": 30.02585649644473, "grad_norm": 1.2381364107131958, "learning_rate": 0.001, "loss": 1.9441, "step": 92900 }, { "epoch": 30.058177117000646, "grad_norm": 1.2475801706314087, "learning_rate": 0.001, "loss": 1.9169, "step": 93000 }, { "epoch": 30.09049773755656, "grad_norm": 1.440842866897583, "learning_rate": 0.001, "loss": 1.9157, "step": 93100 }, { "epoch": 30.122818358112475, "grad_norm": 1.1453300714492798, "learning_rate": 0.001, "loss": 1.9223, "step": 93200 }, { "epoch": 30.15513897866839, "grad_norm": 1.0690298080444336, "learning_rate": 0.001, "loss": 1.9323, "step": 93300 }, { "epoch": 30.187459599224304, "grad_norm": 1.7190358638763428, "learning_rate": 0.001, "loss": 1.9623, "step": 93400 }, { "epoch": 30.21978021978022, "grad_norm": 1.0190703868865967, "learning_rate": 0.001, "loss": 1.9552, "step": 93500 }, { "epoch": 30.252100840336134, "grad_norm": 1.0576410293579102, "learning_rate": 0.001, "loss": 1.9724, "step": 93600 }, { "epoch": 30.284421460892048, "grad_norm": 1.1603765487670898, "learning_rate": 0.001, "loss": 1.9563, "step": 93700 }, { "epoch": 30.316742081447963, "grad_norm": 1.3184884786605835, "learning_rate": 0.001, "loss": 1.9375, "step": 93800 }, { "epoch": 30.349062702003877, "grad_norm": 0.9913526177406311, "learning_rate": 0.001, "loss": 1.967, "step": 93900 }, { "epoch": 30.381383322559792, "grad_norm": 1.3647373914718628, "learning_rate": 0.001, "loss": 1.9812, "step": 94000 }, { "epoch": 30.413703943115706, "grad_norm": 1.1337854862213135, "learning_rate": 0.001, "loss": 1.9927, "step": 94100 }, { "epoch": 30.44602456367162, "grad_norm": 1.3728899955749512, "learning_rate": 0.001, "loss": 1.9755, "step": 94200 }, { "epoch": 30.478345184227535, "grad_norm": 1.633349061012268, "learning_rate": 0.001, "loss": 2.0141, "step": 94300 }, { "epoch": 30.51066580478345, "grad_norm": 1.2653772830963135, "learning_rate": 0.001, "loss": 2.0159, "step": 94400 }, { "epoch": 30.542986425339368, "grad_norm": 1.3831946849822998, "learning_rate": 0.001, "loss": 1.9999, "step": 94500 }, { "epoch": 30.575307045895283, "grad_norm": 1.022870659828186, "learning_rate": 0.001, "loss": 2.0143, "step": 94600 }, { "epoch": 30.607627666451197, "grad_norm": 1.227567195892334, "learning_rate": 0.001, "loss": 2.0142, "step": 94700 }, { "epoch": 30.639948287007112, "grad_norm": 1.2944329977035522, "learning_rate": 0.001, "loss": 1.9835, "step": 94800 }, { "epoch": 30.672268907563026, "grad_norm": 1.64191472530365, "learning_rate": 0.001, "loss": 1.9901, "step": 94900 }, { "epoch": 30.70458952811894, "grad_norm": 1.162553071975708, "learning_rate": 0.001, "loss": 2.0045, "step": 95000 }, { "epoch": 30.736910148674855, "grad_norm": 1.1812092065811157, "learning_rate": 0.001, "loss": 2.0317, "step": 95100 }, { "epoch": 30.76923076923077, "grad_norm": 1.1358566284179688, "learning_rate": 0.001, "loss": 2.0088, "step": 95200 }, { "epoch": 30.801551389786685, "grad_norm": 1.4215623140335083, "learning_rate": 0.001, "loss": 2.0188, "step": 95300 }, { "epoch": 30.8338720103426, "grad_norm": 1.0207571983337402, "learning_rate": 0.001, "loss": 2.0421, "step": 95400 }, { "epoch": 30.866192630898514, "grad_norm": 1.1766799688339233, "learning_rate": 0.001, "loss": 2.0261, "step": 95500 }, { "epoch": 30.89851325145443, "grad_norm": 1.167883276939392, "learning_rate": 0.001, "loss": 2.0145, "step": 95600 }, { "epoch": 30.930833872010343, "grad_norm": 1.4235504865646362, "learning_rate": 0.001, "loss": 2.0521, "step": 95700 }, { "epoch": 30.963154492566257, "grad_norm": 0.9883701205253601, "learning_rate": 0.001, "loss": 2.028, "step": 95800 }, { "epoch": 30.995475113122172, "grad_norm": 1.138373613357544, "learning_rate": 0.001, "loss": 2.0154, "step": 95900 }, { "epoch": 31.027795733678087, "grad_norm": 1.3630508184432983, "learning_rate": 0.001, "loss": 1.8929, "step": 96000 }, { "epoch": 31.060116354234, "grad_norm": 1.895157814025879, "learning_rate": 0.001, "loss": 1.9025, "step": 96100 }, { "epoch": 31.092436974789916, "grad_norm": 1.3133440017700195, "learning_rate": 0.001, "loss": 1.9233, "step": 96200 }, { "epoch": 31.12475759534583, "grad_norm": 1.2981374263763428, "learning_rate": 0.001, "loss": 1.9224, "step": 96300 }, { "epoch": 31.157078215901745, "grad_norm": 1.547069787979126, "learning_rate": 0.001, "loss": 1.9235, "step": 96400 }, { "epoch": 31.18939883645766, "grad_norm": 1.1872363090515137, "learning_rate": 0.001, "loss": 1.9043, "step": 96500 }, { "epoch": 31.221719457013574, "grad_norm": 1.2956496477127075, "learning_rate": 0.001, "loss": 1.941, "step": 96600 }, { "epoch": 31.25404007756949, "grad_norm": 1.8166218996047974, "learning_rate": 0.001, "loss": 1.9277, "step": 96700 }, { "epoch": 31.286360698125403, "grad_norm": 1.355141520500183, "learning_rate": 0.001, "loss": 1.9502, "step": 96800 }, { "epoch": 31.318681318681318, "grad_norm": 1.3605191707611084, "learning_rate": 0.001, "loss": 1.9556, "step": 96900 }, { "epoch": 31.351001939237232, "grad_norm": 1.4461206197738647, "learning_rate": 0.001, "loss": 1.949, "step": 97000 }, { "epoch": 31.383322559793147, "grad_norm": 1.3594260215759277, "learning_rate": 0.001, "loss": 1.9486, "step": 97100 }, { "epoch": 31.41564318034906, "grad_norm": 1.4502143859863281, "learning_rate": 0.001, "loss": 1.969, "step": 97200 }, { "epoch": 31.447963800904976, "grad_norm": 1.5335756540298462, "learning_rate": 0.001, "loss": 1.9664, "step": 97300 }, { "epoch": 31.48028442146089, "grad_norm": 1.6835445165634155, "learning_rate": 0.001, "loss": 1.9612, "step": 97400 }, { "epoch": 31.51260504201681, "grad_norm": 1.0857112407684326, "learning_rate": 0.001, "loss": 1.9835, "step": 97500 }, { "epoch": 31.544925662572723, "grad_norm": 1.4577279090881348, "learning_rate": 0.001, "loss": 1.9784, "step": 97600 }, { "epoch": 31.577246283128638, "grad_norm": 1.52741539478302, "learning_rate": 0.001, "loss": 1.9788, "step": 97700 }, { "epoch": 31.609566903684552, "grad_norm": 1.308344841003418, "learning_rate": 0.001, "loss": 1.9791, "step": 97800 }, { "epoch": 31.641887524240467, "grad_norm": 1.5521067380905151, "learning_rate": 0.001, "loss": 1.9852, "step": 97900 }, { "epoch": 31.67420814479638, "grad_norm": 1.5745213031768799, "learning_rate": 0.001, "loss": 1.9962, "step": 98000 }, { "epoch": 31.706528765352296, "grad_norm": 1.4261987209320068, "learning_rate": 0.001, "loss": 2.0037, "step": 98100 }, { "epoch": 31.73884938590821, "grad_norm": 1.3460932970046997, "learning_rate": 0.001, "loss": 1.985, "step": 98200 }, { "epoch": 31.771170006464125, "grad_norm": 1.37859308719635, "learning_rate": 0.001, "loss": 2.0048, "step": 98300 }, { "epoch": 31.80349062702004, "grad_norm": 1.5813380479812622, "learning_rate": 0.001, "loss": 1.9964, "step": 98400 }, { "epoch": 31.835811247575954, "grad_norm": 1.2057660818099976, "learning_rate": 0.001, "loss": 2.0085, "step": 98500 }, { "epoch": 31.86813186813187, "grad_norm": 1.7127104997634888, "learning_rate": 0.001, "loss": 2.0064, "step": 98600 }, { "epoch": 31.900452488687783, "grad_norm": 1.3543590307235718, "learning_rate": 0.001, "loss": 2.0216, "step": 98700 }, { "epoch": 31.932773109243698, "grad_norm": 1.3376524448394775, "learning_rate": 0.001, "loss": 2.0191, "step": 98800 }, { "epoch": 31.965093729799612, "grad_norm": 1.2518678903579712, "learning_rate": 0.001, "loss": 2.0098, "step": 98900 }, { "epoch": 31.997414350355527, "grad_norm": 1.4121952056884766, "learning_rate": 0.001, "loss": 2.006, "step": 99000 }, { "epoch": 32.02973497091144, "grad_norm": 1.2436646223068237, "learning_rate": 0.001, "loss": 1.9104, "step": 99100 }, { "epoch": 32.062055591467356, "grad_norm": 1.7367206811904907, "learning_rate": 0.001, "loss": 1.8809, "step": 99200 }, { "epoch": 32.09437621202327, "grad_norm": 1.4106762409210205, "learning_rate": 0.001, "loss": 1.871, "step": 99300 }, { "epoch": 32.126696832579185, "grad_norm": 1.8812962770462036, "learning_rate": 0.001, "loss": 1.8948, "step": 99400 }, { "epoch": 32.1590174531351, "grad_norm": 1.5063000917434692, "learning_rate": 0.001, "loss": 1.9008, "step": 99500 }, { "epoch": 32.191338073691014, "grad_norm": 1.3886702060699463, "learning_rate": 0.001, "loss": 1.9098, "step": 99600 }, { "epoch": 32.22365869424693, "grad_norm": 1.7176951169967651, "learning_rate": 0.001, "loss": 1.911, "step": 99700 }, { "epoch": 32.25597931480284, "grad_norm": 1.4107688665390015, "learning_rate": 0.001, "loss": 1.9095, "step": 99800 }, { "epoch": 32.28829993535876, "grad_norm": 1.4663777351379395, "learning_rate": 0.001, "loss": 1.9209, "step": 99900 }, { "epoch": 32.32062055591467, "grad_norm": 1.4027447700500488, "learning_rate": 0.001, "loss": 1.9371, "step": 100000 }, { "epoch": 32.35294117647059, "grad_norm": 1.5440044403076172, "learning_rate": 0.001, "loss": 1.9344, "step": 100100 }, { "epoch": 32.3852617970265, "grad_norm": 1.1808867454528809, "learning_rate": 0.001, "loss": 1.9392, "step": 100200 }, { "epoch": 32.417582417582416, "grad_norm": 1.8131669759750366, "learning_rate": 0.001, "loss": 1.9282, "step": 100300 }, { "epoch": 32.44990303813833, "grad_norm": 1.7838051319122314, "learning_rate": 0.001, "loss": 1.939, "step": 100400 }, { "epoch": 32.482223658694245, "grad_norm": 1.6978235244750977, "learning_rate": 0.001, "loss": 1.9372, "step": 100500 }, { "epoch": 32.51454427925016, "grad_norm": 1.7364856004714966, "learning_rate": 0.001, "loss": 1.9371, "step": 100600 }, { "epoch": 32.546864899806074, "grad_norm": 1.515915870666504, "learning_rate": 0.001, "loss": 1.9544, "step": 100700 }, { "epoch": 32.57918552036199, "grad_norm": 1.5563019514083862, "learning_rate": 0.001, "loss": 1.9404, "step": 100800 }, { "epoch": 32.6115061409179, "grad_norm": 1.6745736598968506, "learning_rate": 0.001, "loss": 1.959, "step": 100900 }, { "epoch": 32.64382676147382, "grad_norm": 1.301689863204956, "learning_rate": 0.001, "loss": 1.9624, "step": 101000 }, { "epoch": 32.67614738202973, "grad_norm": 1.9219502210617065, "learning_rate": 0.001, "loss": 1.9617, "step": 101100 }, { "epoch": 32.70846800258565, "grad_norm": 1.2895933389663696, "learning_rate": 0.001, "loss": 1.9755, "step": 101200 }, { "epoch": 32.74078862314156, "grad_norm": 1.6644096374511719, "learning_rate": 0.001, "loss": 1.9767, "step": 101300 }, { "epoch": 32.773109243697476, "grad_norm": 1.2009501457214355, "learning_rate": 0.001, "loss": 1.9859, "step": 101400 }, { "epoch": 32.80542986425339, "grad_norm": 1.6504448652267456, "learning_rate": 0.001, "loss": 1.9893, "step": 101500 }, { "epoch": 32.837750484809305, "grad_norm": 1.294108510017395, "learning_rate": 0.001, "loss": 1.9821, "step": 101600 }, { "epoch": 32.87007110536522, "grad_norm": 1.3442059755325317, "learning_rate": 0.001, "loss": 1.9768, "step": 101700 }, { "epoch": 32.902391725921134, "grad_norm": 1.323499321937561, "learning_rate": 0.001, "loss": 1.9862, "step": 101800 }, { "epoch": 32.93471234647705, "grad_norm": 1.263298749923706, "learning_rate": 0.001, "loss": 1.9949, "step": 101900 }, { "epoch": 32.967032967032964, "grad_norm": 1.1355193853378296, "learning_rate": 0.001, "loss": 1.9974, "step": 102000 }, { "epoch": 32.999353587588885, "grad_norm": 1.4146533012390137, "learning_rate": 0.001, "loss": 1.989, "step": 102100 }, { "epoch": 33.0316742081448, "grad_norm": 1.960270643234253, "learning_rate": 0.001, "loss": 1.8564, "step": 102200 }, { "epoch": 33.063994828700714, "grad_norm": 1.4034452438354492, "learning_rate": 0.001, "loss": 1.8468, "step": 102300 }, { "epoch": 33.09631544925663, "grad_norm": 1.3908497095108032, "learning_rate": 0.001, "loss": 1.8627, "step": 102400 }, { "epoch": 33.12863606981254, "grad_norm": 1.7255970239639282, "learning_rate": 0.001, "loss": 1.8636, "step": 102500 }, { "epoch": 33.16095669036846, "grad_norm": 1.5471359491348267, "learning_rate": 0.001, "loss": 1.8713, "step": 102600 }, { "epoch": 33.19327731092437, "grad_norm": 1.8626443147659302, "learning_rate": 0.001, "loss": 1.8739, "step": 102700 }, { "epoch": 33.22559793148029, "grad_norm": 2.0910379886627197, "learning_rate": 0.001, "loss": 1.8891, "step": 102800 }, { "epoch": 33.2579185520362, "grad_norm": 1.719954013824463, "learning_rate": 0.001, "loss": 1.8907, "step": 102900 }, { "epoch": 33.290239172592116, "grad_norm": 1.8111015558242798, "learning_rate": 0.001, "loss": 1.8979, "step": 103000 }, { "epoch": 33.32255979314803, "grad_norm": 1.9263817071914673, "learning_rate": 0.001, "loss": 1.8959, "step": 103100 }, { "epoch": 33.354880413703945, "grad_norm": 1.8665021657943726, "learning_rate": 0.001, "loss": 1.9126, "step": 103200 }, { "epoch": 33.38720103425986, "grad_norm": 1.5710049867630005, "learning_rate": 0.001, "loss": 1.9129, "step": 103300 }, { "epoch": 33.419521654815775, "grad_norm": 1.529770016670227, "learning_rate": 0.001, "loss": 1.9401, "step": 103400 }, { "epoch": 33.45184227537169, "grad_norm": 1.5240123271942139, "learning_rate": 0.001, "loss": 1.9269, "step": 103500 }, { "epoch": 33.484162895927604, "grad_norm": 1.6057839393615723, "learning_rate": 0.001, "loss": 1.9255, "step": 103600 }, { "epoch": 33.51648351648352, "grad_norm": 1.86514151096344, "learning_rate": 0.001, "loss": 1.9281, "step": 103700 }, { "epoch": 33.54880413703943, "grad_norm": 1.2546050548553467, "learning_rate": 0.001, "loss": 1.9388, "step": 103800 }, { "epoch": 33.58112475759535, "grad_norm": 1.3954681158065796, "learning_rate": 0.001, "loss": 1.9281, "step": 103900 }, { "epoch": 33.61344537815126, "grad_norm": 1.4027971029281616, "learning_rate": 0.001, "loss": 1.9497, "step": 104000 }, { "epoch": 33.645765998707176, "grad_norm": 1.707762598991394, "learning_rate": 0.001, "loss": 1.9408, "step": 104100 }, { "epoch": 33.67808661926309, "grad_norm": 1.932440161705017, "learning_rate": 0.001, "loss": 1.952, "step": 104200 }, { "epoch": 33.710407239819006, "grad_norm": 1.7187687158584595, "learning_rate": 0.001, "loss": 1.9449, "step": 104300 }, { "epoch": 33.74272786037492, "grad_norm": 1.6193797588348389, "learning_rate": 0.001, "loss": 1.9647, "step": 104400 }, { "epoch": 33.775048480930835, "grad_norm": 1.7175681591033936, "learning_rate": 0.001, "loss": 1.9666, "step": 104500 }, { "epoch": 33.80736910148675, "grad_norm": 1.404217004776001, "learning_rate": 0.001, "loss": 1.9607, "step": 104600 }, { "epoch": 33.839689722042664, "grad_norm": 1.708268165588379, "learning_rate": 0.001, "loss": 1.9749, "step": 104700 }, { "epoch": 33.87201034259858, "grad_norm": 1.5690302848815918, "learning_rate": 0.001, "loss": 1.9723, "step": 104800 }, { "epoch": 33.90433096315449, "grad_norm": 1.7933155298233032, "learning_rate": 0.001, "loss": 1.9668, "step": 104900 }, { "epoch": 33.93665158371041, "grad_norm": 1.6681078672409058, "learning_rate": 0.001, "loss": 1.9825, "step": 105000 }, { "epoch": 33.96897220426632, "grad_norm": 1.6313607692718506, "learning_rate": 0.001, "loss": 2.0023, "step": 105100 }, { "epoch": 34.00129282482224, "grad_norm": 1.3496909141540527, "learning_rate": 0.001, "loss": 1.97, "step": 105200 }, { "epoch": 34.03361344537815, "grad_norm": 1.725895881652832, "learning_rate": 0.001, "loss": 1.8485, "step": 105300 }, { "epoch": 34.065934065934066, "grad_norm": 1.2111670970916748, "learning_rate": 0.001, "loss": 1.8429, "step": 105400 }, { "epoch": 34.09825468648998, "grad_norm": 1.6606335639953613, "learning_rate": 0.001, "loss": 1.8409, "step": 105500 }, { "epoch": 34.130575307045895, "grad_norm": 1.2555725574493408, "learning_rate": 0.001, "loss": 1.8649, "step": 105600 }, { "epoch": 34.16289592760181, "grad_norm": 1.5818908214569092, "learning_rate": 0.001, "loss": 1.8675, "step": 105700 }, { "epoch": 34.195216548157724, "grad_norm": 1.6778271198272705, "learning_rate": 0.001, "loss": 1.8732, "step": 105800 }, { "epoch": 34.22753716871364, "grad_norm": 1.4768215417861938, "learning_rate": 0.001, "loss": 1.868, "step": 105900 }, { "epoch": 34.25985778926955, "grad_norm": 1.2955535650253296, "learning_rate": 0.001, "loss": 1.8781, "step": 106000 }, { "epoch": 34.29217840982547, "grad_norm": 1.585286259651184, "learning_rate": 0.001, "loss": 1.8739, "step": 106100 }, { "epoch": 34.32449903038138, "grad_norm": 1.5454331636428833, "learning_rate": 0.001, "loss": 1.8881, "step": 106200 }, { "epoch": 34.3568196509373, "grad_norm": 1.4111459255218506, "learning_rate": 0.001, "loss": 1.9013, "step": 106300 }, { "epoch": 34.38914027149321, "grad_norm": 1.7052496671676636, "learning_rate": 0.001, "loss": 1.883, "step": 106400 }, { "epoch": 34.421460892049126, "grad_norm": 1.6497286558151245, "learning_rate": 0.001, "loss": 1.9053, "step": 106500 }, { "epoch": 34.45378151260504, "grad_norm": 1.4096091985702515, "learning_rate": 0.001, "loss": 1.9002, "step": 106600 }, { "epoch": 34.486102133160955, "grad_norm": 1.349039077758789, "learning_rate": 0.001, "loss": 1.8935, "step": 106700 }, { "epoch": 34.51842275371687, "grad_norm": 1.6230989694595337, "learning_rate": 0.001, "loss": 1.9151, "step": 106800 }, { "epoch": 34.550743374272784, "grad_norm": 1.2353782653808594, "learning_rate": 0.001, "loss": 1.904, "step": 106900 }, { "epoch": 34.5830639948287, "grad_norm": 1.5027496814727783, "learning_rate": 0.001, "loss": 1.9106, "step": 107000 }, { "epoch": 34.61538461538461, "grad_norm": 1.5522241592407227, "learning_rate": 0.001, "loss": 1.9283, "step": 107100 }, { "epoch": 34.64770523594053, "grad_norm": 1.550193428993225, "learning_rate": 0.001, "loss": 1.9336, "step": 107200 }, { "epoch": 34.68002585649644, "grad_norm": 1.786367654800415, "learning_rate": 0.001, "loss": 1.9277, "step": 107300 }, { "epoch": 34.71234647705236, "grad_norm": 1.2258667945861816, "learning_rate": 0.001, "loss": 1.9287, "step": 107400 }, { "epoch": 34.74466709760827, "grad_norm": 1.552863597869873, "learning_rate": 0.001, "loss": 1.9324, "step": 107500 }, { "epoch": 34.776987718164186, "grad_norm": 1.5470998287200928, "learning_rate": 0.001, "loss": 1.9377, "step": 107600 }, { "epoch": 34.8093083387201, "grad_norm": 1.2013700008392334, "learning_rate": 0.001, "loss": 1.9421, "step": 107700 }, { "epoch": 34.841628959276015, "grad_norm": 1.5143885612487793, "learning_rate": 0.001, "loss": 1.9404, "step": 107800 }, { "epoch": 34.87394957983193, "grad_norm": 1.606493592262268, "learning_rate": 0.001, "loss": 1.9494, "step": 107900 }, { "epoch": 34.906270200387844, "grad_norm": 1.654229998588562, "learning_rate": 0.001, "loss": 1.946, "step": 108000 }, { "epoch": 34.93859082094376, "grad_norm": 1.496848225593567, "learning_rate": 0.001, "loss": 1.965, "step": 108100 }, { "epoch": 34.97091144149967, "grad_norm": 1.171495795249939, "learning_rate": 0.001, "loss": 1.9583, "step": 108200 }, { "epoch": 35.003232062055595, "grad_norm": 1.484670877456665, "learning_rate": 0.001, "loss": 1.9489, "step": 108300 }, { "epoch": 35.03555268261151, "grad_norm": 1.6827988624572754, "learning_rate": 0.001, "loss": 1.8172, "step": 108400 }, { "epoch": 35.067873303167424, "grad_norm": 1.5099338293075562, "learning_rate": 0.001, "loss": 1.8169, "step": 108500 }, { "epoch": 35.10019392372334, "grad_norm": 1.0843244791030884, "learning_rate": 0.001, "loss": 1.8134, "step": 108600 }, { "epoch": 35.13251454427925, "grad_norm": 1.0883989334106445, "learning_rate": 0.001, "loss": 1.8469, "step": 108700 }, { "epoch": 35.16483516483517, "grad_norm": 1.2726517915725708, "learning_rate": 0.001, "loss": 1.8401, "step": 108800 }, { "epoch": 35.19715578539108, "grad_norm": 1.3099430799484253, "learning_rate": 0.001, "loss": 1.8624, "step": 108900 }, { "epoch": 35.229476405947, "grad_norm": 1.0547666549682617, "learning_rate": 0.001, "loss": 1.8532, "step": 109000 }, { "epoch": 35.26179702650291, "grad_norm": 1.3312631845474243, "learning_rate": 0.001, "loss": 1.8733, "step": 109100 }, { "epoch": 35.294117647058826, "grad_norm": 1.3960740566253662, "learning_rate": 0.001, "loss": 1.8592, "step": 109200 }, { "epoch": 35.32643826761474, "grad_norm": 0.9560843110084534, "learning_rate": 0.001, "loss": 1.8542, "step": 109300 }, { "epoch": 35.358758888170655, "grad_norm": 1.2486746311187744, "learning_rate": 0.001, "loss": 1.8638, "step": 109400 }, { "epoch": 35.39107950872657, "grad_norm": 1.9921132326126099, "learning_rate": 0.001, "loss": 1.8619, "step": 109500 }, { "epoch": 35.423400129282484, "grad_norm": 1.6659047603607178, "learning_rate": 0.001, "loss": 1.861, "step": 109600 }, { "epoch": 35.4557207498384, "grad_norm": 1.2826552391052246, "learning_rate": 0.001, "loss": 1.8915, "step": 109700 }, { "epoch": 35.48804137039431, "grad_norm": 1.251529335975647, "learning_rate": 0.001, "loss": 1.8829, "step": 109800 }, { "epoch": 35.52036199095023, "grad_norm": 1.6887863874435425, "learning_rate": 0.001, "loss": 1.9092, "step": 109900 }, { "epoch": 35.55268261150614, "grad_norm": 1.3254314661026, "learning_rate": 0.001, "loss": 1.9104, "step": 110000 }, { "epoch": 35.58500323206206, "grad_norm": 1.2733975648880005, "learning_rate": 0.001, "loss": 1.8877, "step": 110100 }, { "epoch": 35.61732385261797, "grad_norm": 1.3033257722854614, "learning_rate": 0.001, "loss": 1.8943, "step": 110200 }, { "epoch": 35.649644473173886, "grad_norm": 1.5639281272888184, "learning_rate": 0.001, "loss": 1.9241, "step": 110300 }, { "epoch": 35.6819650937298, "grad_norm": 1.1277391910552979, "learning_rate": 0.001, "loss": 1.9131, "step": 110400 }, { "epoch": 35.714285714285715, "grad_norm": 1.113537311553955, "learning_rate": 0.001, "loss": 1.9023, "step": 110500 }, { "epoch": 35.74660633484163, "grad_norm": 1.3119865655899048, "learning_rate": 0.001, "loss": 1.9088, "step": 110600 }, { "epoch": 35.778926955397544, "grad_norm": 1.0738476514816284, "learning_rate": 0.001, "loss": 1.9163, "step": 110700 }, { "epoch": 35.81124757595346, "grad_norm": 1.345321774482727, "learning_rate": 0.001, "loss": 1.897, "step": 110800 }, { "epoch": 35.84356819650937, "grad_norm": 0.9614481329917908, "learning_rate": 0.001, "loss": 1.951, "step": 110900 }, { "epoch": 35.87588881706529, "grad_norm": 1.6291916370391846, "learning_rate": 0.001, "loss": 1.9186, "step": 111000 }, { "epoch": 35.9082094376212, "grad_norm": 1.0890250205993652, "learning_rate": 0.001, "loss": 1.9356, "step": 111100 }, { "epoch": 35.94053005817712, "grad_norm": 1.4325668811798096, "learning_rate": 0.001, "loss": 1.9342, "step": 111200 }, { "epoch": 35.97285067873303, "grad_norm": 1.6891958713531494, "learning_rate": 0.001, "loss": 1.9331, "step": 111300 }, { "epoch": 36.005171299288946, "grad_norm": 1.3368526697158813, "learning_rate": 0.001, "loss": 1.9427, "step": 111400 }, { "epoch": 36.03749191984486, "grad_norm": 1.5092604160308838, "learning_rate": 0.001, "loss": 1.8021, "step": 111500 }, { "epoch": 36.069812540400775, "grad_norm": 1.243878722190857, "learning_rate": 0.001, "loss": 1.8109, "step": 111600 }, { "epoch": 36.10213316095669, "grad_norm": 0.9875638484954834, "learning_rate": 0.001, "loss": 1.8099, "step": 111700 }, { "epoch": 36.134453781512605, "grad_norm": 1.0712621212005615, "learning_rate": 0.001, "loss": 1.8081, "step": 111800 }, { "epoch": 36.16677440206852, "grad_norm": 1.3209693431854248, "learning_rate": 0.001, "loss": 1.828, "step": 111900 }, { "epoch": 36.199095022624434, "grad_norm": 1.3973124027252197, "learning_rate": 0.001, "loss": 1.8319, "step": 112000 }, { "epoch": 36.23141564318035, "grad_norm": 1.2643061876296997, "learning_rate": 0.001, "loss": 1.8468, "step": 112100 }, { "epoch": 36.26373626373626, "grad_norm": 1.2189161777496338, "learning_rate": 0.001, "loss": 1.8274, "step": 112200 }, { "epoch": 36.29605688429218, "grad_norm": 1.1872079372406006, "learning_rate": 0.001, "loss": 1.8441, "step": 112300 }, { "epoch": 36.32837750484809, "grad_norm": 1.401799201965332, "learning_rate": 0.001, "loss": 1.86, "step": 112400 }, { "epoch": 36.36069812540401, "grad_norm": 1.4873459339141846, "learning_rate": 0.001, "loss": 1.8611, "step": 112500 }, { "epoch": 36.39301874595992, "grad_norm": 1.5051405429840088, "learning_rate": 0.001, "loss": 1.8512, "step": 112600 }, { "epoch": 36.425339366515836, "grad_norm": 1.6135274171829224, "learning_rate": 0.001, "loss": 1.8497, "step": 112700 }, { "epoch": 36.45765998707175, "grad_norm": 1.05423104763031, "learning_rate": 0.001, "loss": 1.8806, "step": 112800 }, { "epoch": 36.489980607627665, "grad_norm": 1.1396740674972534, "learning_rate": 0.001, "loss": 1.8848, "step": 112900 }, { "epoch": 36.52230122818358, "grad_norm": 1.095083236694336, "learning_rate": 0.001, "loss": 1.8578, "step": 113000 }, { "epoch": 36.554621848739494, "grad_norm": 1.0797603130340576, "learning_rate": 0.001, "loss": 1.8792, "step": 113100 }, { "epoch": 36.58694246929541, "grad_norm": 1.4871982336044312, "learning_rate": 0.001, "loss": 1.8817, "step": 113200 }, { "epoch": 36.61926308985132, "grad_norm": 1.134158730506897, "learning_rate": 0.001, "loss": 1.9027, "step": 113300 }, { "epoch": 36.65158371040724, "grad_norm": 1.3541128635406494, "learning_rate": 0.001, "loss": 1.8867, "step": 113400 }, { "epoch": 36.68390433096315, "grad_norm": 0.9839674234390259, "learning_rate": 0.001, "loss": 1.8863, "step": 113500 }, { "epoch": 36.71622495151907, "grad_norm": 1.0756222009658813, "learning_rate": 0.001, "loss": 1.8909, "step": 113600 }, { "epoch": 36.74854557207498, "grad_norm": 1.3785662651062012, "learning_rate": 0.001, "loss": 1.9015, "step": 113700 }, { "epoch": 36.780866192630896, "grad_norm": 1.2404823303222656, "learning_rate": 0.001, "loss": 1.8972, "step": 113800 }, { "epoch": 36.81318681318681, "grad_norm": 1.2794967889785767, "learning_rate": 0.001, "loss": 1.9081, "step": 113900 }, { "epoch": 36.845507433742725, "grad_norm": 1.1256643533706665, "learning_rate": 0.001, "loss": 1.904, "step": 114000 }, { "epoch": 36.87782805429864, "grad_norm": 1.3804694414138794, "learning_rate": 0.001, "loss": 1.9057, "step": 114100 }, { "epoch": 36.910148674854554, "grad_norm": 1.4928176403045654, "learning_rate": 0.001, "loss": 1.9147, "step": 114200 }, { "epoch": 36.94246929541047, "grad_norm": 1.1267145872116089, "learning_rate": 0.001, "loss": 1.9063, "step": 114300 }, { "epoch": 36.97478991596638, "grad_norm": 1.179831862449646, "learning_rate": 0.001, "loss": 1.9184, "step": 114400 }, { "epoch": 37.007110536522305, "grad_norm": 1.2279536724090576, "learning_rate": 0.001, "loss": 1.8946, "step": 114500 }, { "epoch": 37.03943115707822, "grad_norm": 1.1896717548370361, "learning_rate": 0.001, "loss": 1.7819, "step": 114600 }, { "epoch": 37.071751777634134, "grad_norm": 1.3132928609848022, "learning_rate": 0.001, "loss": 1.7896, "step": 114700 }, { "epoch": 37.10407239819005, "grad_norm": 1.1400185823440552, "learning_rate": 0.001, "loss": 1.7985, "step": 114800 }, { "epoch": 37.13639301874596, "grad_norm": 1.4554778337478638, "learning_rate": 0.001, "loss": 1.7952, "step": 114900 }, { "epoch": 37.16871363930188, "grad_norm": 1.4060354232788086, "learning_rate": 0.001, "loss": 1.7948, "step": 115000 }, { "epoch": 37.20103425985779, "grad_norm": 0.9965196847915649, "learning_rate": 0.001, "loss": 1.8039, "step": 115100 }, { "epoch": 37.23335488041371, "grad_norm": 1.0781031847000122, "learning_rate": 0.001, "loss": 1.8393, "step": 115200 }, { "epoch": 37.26567550096962, "grad_norm": 1.2023656368255615, "learning_rate": 0.001, "loss": 1.818, "step": 115300 }, { "epoch": 37.297996121525536, "grad_norm": 1.4046202898025513, "learning_rate": 0.001, "loss": 1.8312, "step": 115400 }, { "epoch": 37.33031674208145, "grad_norm": 1.2398176193237305, "learning_rate": 0.001, "loss": 1.8338, "step": 115500 }, { "epoch": 37.362637362637365, "grad_norm": 1.5853986740112305, "learning_rate": 0.001, "loss": 1.8268, "step": 115600 }, { "epoch": 37.39495798319328, "grad_norm": 1.4195340871810913, "learning_rate": 0.001, "loss": 1.8459, "step": 115700 }, { "epoch": 37.427278603749194, "grad_norm": 1.4397501945495605, "learning_rate": 0.001, "loss": 1.851, "step": 115800 }, { "epoch": 37.45959922430511, "grad_norm": 1.0061242580413818, "learning_rate": 0.001, "loss": 1.8391, "step": 115900 }, { "epoch": 37.49191984486102, "grad_norm": 1.672114610671997, "learning_rate": 0.001, "loss": 1.8614, "step": 116000 }, { "epoch": 37.52424046541694, "grad_norm": 1.0445514917373657, "learning_rate": 0.001, "loss": 1.8528, "step": 116100 }, { "epoch": 37.55656108597285, "grad_norm": 1.3120521306991577, "learning_rate": 0.001, "loss": 1.8488, "step": 116200 }, { "epoch": 37.58888170652877, "grad_norm": 1.2624495029449463, "learning_rate": 0.001, "loss": 1.8653, "step": 116300 }, { "epoch": 37.62120232708468, "grad_norm": 0.9890035390853882, "learning_rate": 0.001, "loss": 1.8696, "step": 116400 }, { "epoch": 37.653522947640596, "grad_norm": 1.3105789422988892, "learning_rate": 0.001, "loss": 1.8667, "step": 116500 }, { "epoch": 37.68584356819651, "grad_norm": 1.5599737167358398, "learning_rate": 0.001, "loss": 1.867, "step": 116600 }, { "epoch": 37.718164188752425, "grad_norm": 1.437135934829712, "learning_rate": 0.001, "loss": 1.8869, "step": 116700 }, { "epoch": 37.75048480930834, "grad_norm": 1.0562933683395386, "learning_rate": 0.001, "loss": 1.8696, "step": 116800 }, { "epoch": 37.782805429864254, "grad_norm": 1.3576165437698364, "learning_rate": 0.001, "loss": 1.9007, "step": 116900 }, { "epoch": 37.81512605042017, "grad_norm": 1.1719582080841064, "learning_rate": 0.001, "loss": 1.8788, "step": 117000 }, { "epoch": 37.84744667097608, "grad_norm": 1.3145471811294556, "learning_rate": 0.001, "loss": 1.8952, "step": 117100 }, { "epoch": 37.879767291532, "grad_norm": 1.1600955724716187, "learning_rate": 0.001, "loss": 1.8729, "step": 117200 }, { "epoch": 37.91208791208791, "grad_norm": 1.1656097173690796, "learning_rate": 0.001, "loss": 1.9102, "step": 117300 }, { "epoch": 37.94440853264383, "grad_norm": 1.3269482851028442, "learning_rate": 0.001, "loss": 1.9104, "step": 117400 }, { "epoch": 37.97672915319974, "grad_norm": 1.4651037454605103, "learning_rate": 0.001, "loss": 1.9153, "step": 117500 }, { "epoch": 38.009049773755656, "grad_norm": 1.4630742073059082, "learning_rate": 0.001, "loss": 1.8721, "step": 117600 }, { "epoch": 38.04137039431157, "grad_norm": 1.5355961322784424, "learning_rate": 0.001, "loss": 1.779, "step": 117700 }, { "epoch": 38.073691014867485, "grad_norm": 1.068429708480835, "learning_rate": 0.001, "loss": 1.7664, "step": 117800 }, { "epoch": 38.1060116354234, "grad_norm": 1.7566649913787842, "learning_rate": 0.001, "loss": 1.786, "step": 117900 }, { "epoch": 38.138332255979314, "grad_norm": 1.3657581806182861, "learning_rate": 0.001, "loss": 1.7981, "step": 118000 }, { "epoch": 38.17065287653523, "grad_norm": 1.031551480293274, "learning_rate": 0.001, "loss": 1.7941, "step": 118100 }, { "epoch": 38.20297349709114, "grad_norm": 1.4861186742782593, "learning_rate": 0.001, "loss": 1.8068, "step": 118200 }, { "epoch": 38.23529411764706, "grad_norm": 1.6767009496688843, "learning_rate": 0.001, "loss": 1.7795, "step": 118300 }, { "epoch": 38.26761473820297, "grad_norm": 1.299914836883545, "learning_rate": 0.001, "loss": 1.8062, "step": 118400 }, { "epoch": 38.29993535875889, "grad_norm": 1.477315902709961, "learning_rate": 0.001, "loss": 1.8114, "step": 118500 }, { "epoch": 38.3322559793148, "grad_norm": 1.3180029392242432, "learning_rate": 0.001, "loss": 1.8187, "step": 118600 }, { "epoch": 38.364576599870716, "grad_norm": 1.2591907978057861, "learning_rate": 0.001, "loss": 1.8163, "step": 118700 }, { "epoch": 38.39689722042663, "grad_norm": 0.9934723377227783, "learning_rate": 0.001, "loss": 1.8256, "step": 118800 }, { "epoch": 38.429217840982545, "grad_norm": 1.316521406173706, "learning_rate": 0.001, "loss": 1.8199, "step": 118900 }, { "epoch": 38.46153846153846, "grad_norm": 1.5410053730010986, "learning_rate": 0.001, "loss": 1.8195, "step": 119000 }, { "epoch": 38.493859082094374, "grad_norm": 1.301754355430603, "learning_rate": 0.001, "loss": 1.8209, "step": 119100 }, { "epoch": 38.52617970265029, "grad_norm": 1.0874885320663452, "learning_rate": 0.001, "loss": 1.8474, "step": 119200 }, { "epoch": 38.558500323206204, "grad_norm": 1.8527356386184692, "learning_rate": 0.001, "loss": 1.8503, "step": 119300 }, { "epoch": 38.59082094376212, "grad_norm": 1.24700927734375, "learning_rate": 0.001, "loss": 1.8372, "step": 119400 }, { "epoch": 38.62314156431803, "grad_norm": 1.1928914785385132, "learning_rate": 0.001, "loss": 1.8499, "step": 119500 }, { "epoch": 38.65546218487395, "grad_norm": 1.2841598987579346, "learning_rate": 0.001, "loss": 1.8461, "step": 119600 }, { "epoch": 38.68778280542986, "grad_norm": 1.2930001020431519, "learning_rate": 0.001, "loss": 1.8546, "step": 119700 }, { "epoch": 38.720103425985776, "grad_norm": 1.3524971008300781, "learning_rate": 0.001, "loss": 1.8654, "step": 119800 }, { "epoch": 38.75242404654169, "grad_norm": 1.240613341331482, "learning_rate": 0.001, "loss": 1.8599, "step": 119900 }, { "epoch": 38.784744667097605, "grad_norm": 1.337829351425171, "learning_rate": 0.001, "loss": 1.8665, "step": 120000 }, { "epoch": 38.81706528765352, "grad_norm": 1.1188397407531738, "learning_rate": 0.001, "loss": 1.8555, "step": 120100 }, { "epoch": 38.849385908209435, "grad_norm": 1.049929141998291, "learning_rate": 0.001, "loss": 1.8901, "step": 120200 }, { "epoch": 38.88170652876535, "grad_norm": 1.0788209438323975, "learning_rate": 0.001, "loss": 1.8839, "step": 120300 }, { "epoch": 38.914027149321264, "grad_norm": 1.2696701288223267, "learning_rate": 0.001, "loss": 1.8861, "step": 120400 }, { "epoch": 38.94634776987718, "grad_norm": 1.7940024137496948, "learning_rate": 0.001, "loss": 1.8673, "step": 120500 }, { "epoch": 38.97866839043309, "grad_norm": 1.25827956199646, "learning_rate": 0.001, "loss": 1.8943, "step": 120600 }, { "epoch": 39.010989010989015, "grad_norm": 1.3485623598098755, "learning_rate": 0.001, "loss": 1.8329, "step": 120700 }, { "epoch": 39.04330963154493, "grad_norm": 1.3273441791534424, "learning_rate": 0.001, "loss": 1.7511, "step": 120800 }, { "epoch": 39.075630252100844, "grad_norm": 1.3480987548828125, "learning_rate": 0.001, "loss": 1.7612, "step": 120900 }, { "epoch": 39.10795087265676, "grad_norm": 1.6776913404464722, "learning_rate": 0.001, "loss": 1.7685, "step": 121000 }, { "epoch": 39.14027149321267, "grad_norm": 1.483688473701477, "learning_rate": 0.001, "loss": 1.7646, "step": 121100 }, { "epoch": 39.17259211376859, "grad_norm": 1.494076132774353, "learning_rate": 0.001, "loss": 1.7574, "step": 121200 }, { "epoch": 39.2049127343245, "grad_norm": 1.4595417976379395, "learning_rate": 0.001, "loss": 1.7649, "step": 121300 }, { "epoch": 39.237233354880416, "grad_norm": 1.205970048904419, "learning_rate": 0.001, "loss": 1.8028, "step": 121400 }, { "epoch": 39.26955397543633, "grad_norm": 1.363794207572937, "learning_rate": 0.001, "loss": 1.7898, "step": 121500 }, { "epoch": 39.301874595992246, "grad_norm": 1.1740237474441528, "learning_rate": 0.001, "loss": 1.78, "step": 121600 }, { "epoch": 39.33419521654816, "grad_norm": 1.4625617265701294, "learning_rate": 0.001, "loss": 1.8087, "step": 121700 }, { "epoch": 39.366515837104075, "grad_norm": 1.282263994216919, "learning_rate": 0.001, "loss": 1.8007, "step": 121800 }, { "epoch": 39.39883645765999, "grad_norm": 1.1302987337112427, "learning_rate": 0.001, "loss": 1.7909, "step": 121900 }, { "epoch": 39.431157078215904, "grad_norm": 1.6036306619644165, "learning_rate": 0.001, "loss": 1.809, "step": 122000 }, { "epoch": 39.46347769877182, "grad_norm": 1.3770872354507446, "learning_rate": 0.001, "loss": 1.8171, "step": 122100 }, { "epoch": 39.49579831932773, "grad_norm": 1.3180655241012573, "learning_rate": 0.001, "loss": 1.8166, "step": 122200 }, { "epoch": 39.52811893988365, "grad_norm": 1.7276040315628052, "learning_rate": 0.001, "loss": 1.8245, "step": 122300 }, { "epoch": 39.56043956043956, "grad_norm": 1.8373006582260132, "learning_rate": 0.001, "loss": 1.8152, "step": 122400 }, { "epoch": 39.59276018099548, "grad_norm": 1.5962961912155151, "learning_rate": 0.001, "loss": 1.8338, "step": 122500 }, { "epoch": 39.62508080155139, "grad_norm": 1.576045036315918, "learning_rate": 0.001, "loss": 1.8291, "step": 122600 }, { "epoch": 39.657401422107306, "grad_norm": 1.198539137840271, "learning_rate": 0.001, "loss": 1.8487, "step": 122700 }, { "epoch": 39.68972204266322, "grad_norm": 1.3989983797073364, "learning_rate": 0.001, "loss": 1.8487, "step": 122800 }, { "epoch": 39.722042663219135, "grad_norm": 1.213088870048523, "learning_rate": 0.001, "loss": 1.838, "step": 122900 }, { "epoch": 39.75436328377505, "grad_norm": 1.3717889785766602, "learning_rate": 0.001, "loss": 1.8456, "step": 123000 }, { "epoch": 39.786683904330964, "grad_norm": 1.225083827972412, "learning_rate": 0.001, "loss": 1.8628, "step": 123100 }, { "epoch": 39.81900452488688, "grad_norm": 1.5054234266281128, "learning_rate": 0.001, "loss": 1.8628, "step": 123200 }, { "epoch": 39.85132514544279, "grad_norm": 1.5282232761383057, "learning_rate": 0.001, "loss": 1.8504, "step": 123300 }, { "epoch": 39.88364576599871, "grad_norm": 1.2803972959518433, "learning_rate": 0.001, "loss": 1.8651, "step": 123400 }, { "epoch": 39.91596638655462, "grad_norm": 1.4066797494888306, "learning_rate": 0.001, "loss": 1.8671, "step": 123500 }, { "epoch": 39.94828700711054, "grad_norm": 1.4010391235351562, "learning_rate": 0.001, "loss": 1.843, "step": 123600 }, { "epoch": 39.98060762766645, "grad_norm": 1.6911927461624146, "learning_rate": 0.001, "loss": 1.8662, "step": 123700 }, { "epoch": 40.012928248222366, "grad_norm": 1.6860558986663818, "learning_rate": 0.001, "loss": 1.8256, "step": 123800 }, { "epoch": 40.04524886877828, "grad_norm": 1.307278037071228, "learning_rate": 0.001, "loss": 1.7505, "step": 123900 }, { "epoch": 40.077569489334195, "grad_norm": 1.5941931009292603, "learning_rate": 0.001, "loss": 1.731, "step": 124000 }, { "epoch": 40.10989010989011, "grad_norm": 1.1873581409454346, "learning_rate": 0.001, "loss": 1.7325, "step": 124100 }, { "epoch": 40.142210730446024, "grad_norm": 1.153531551361084, "learning_rate": 0.001, "loss": 1.7385, "step": 124200 }, { "epoch": 40.17453135100194, "grad_norm": 1.6365715265274048, "learning_rate": 0.001, "loss": 1.7542, "step": 124300 }, { "epoch": 40.20685197155785, "grad_norm": 1.7263197898864746, "learning_rate": 0.001, "loss": 1.7778, "step": 124400 }, { "epoch": 40.23917259211377, "grad_norm": 1.5471076965332031, "learning_rate": 0.001, "loss": 1.7632, "step": 124500 }, { "epoch": 40.27149321266968, "grad_norm": 2.015866279602051, "learning_rate": 0.001, "loss": 1.7633, "step": 124600 }, { "epoch": 40.3038138332256, "grad_norm": 1.406688928604126, "learning_rate": 0.001, "loss": 1.7786, "step": 124700 }, { "epoch": 40.33613445378151, "grad_norm": 1.5543419122695923, "learning_rate": 0.001, "loss": 1.7788, "step": 124800 }, { "epoch": 40.368455074337426, "grad_norm": 1.744920253753662, "learning_rate": 0.001, "loss": 1.7981, "step": 124900 }, { "epoch": 40.40077569489334, "grad_norm": 1.4804073572158813, "learning_rate": 0.001, "loss": 1.7952, "step": 125000 }, { "epoch": 40.433096315449255, "grad_norm": 1.7059993743896484, "learning_rate": 0.001, "loss": 1.7812, "step": 125100 }, { "epoch": 40.46541693600517, "grad_norm": 1.6757292747497559, "learning_rate": 0.001, "loss": 1.812, "step": 125200 }, { "epoch": 40.497737556561084, "grad_norm": 1.4929311275482178, "learning_rate": 0.001, "loss": 1.8132, "step": 125300 }, { "epoch": 40.530058177117, "grad_norm": 1.617465615272522, "learning_rate": 0.001, "loss": 1.8102, "step": 125400 }, { "epoch": 40.56237879767291, "grad_norm": 1.3344014883041382, "learning_rate": 0.001, "loss": 1.8099, "step": 125500 }, { "epoch": 40.59469941822883, "grad_norm": 1.1278409957885742, "learning_rate": 0.001, "loss": 1.8197, "step": 125600 }, { "epoch": 40.62702003878474, "grad_norm": 1.408752202987671, "learning_rate": 0.001, "loss": 1.8128, "step": 125700 }, { "epoch": 40.65934065934066, "grad_norm": 1.6868847608566284, "learning_rate": 0.001, "loss": 1.8067, "step": 125800 }, { "epoch": 40.69166127989657, "grad_norm": 1.3664335012435913, "learning_rate": 0.001, "loss": 1.8145, "step": 125900 }, { "epoch": 40.723981900452486, "grad_norm": 1.0876169204711914, "learning_rate": 0.001, "loss": 1.839, "step": 126000 }, { "epoch": 40.7563025210084, "grad_norm": 1.4990159273147583, "learning_rate": 0.001, "loss": 1.8261, "step": 126100 }, { "epoch": 40.788623141564315, "grad_norm": 1.2511619329452515, "learning_rate": 0.001, "loss": 1.8379, "step": 126200 }, { "epoch": 40.82094376212023, "grad_norm": 1.3371134996414185, "learning_rate": 0.001, "loss": 1.835, "step": 126300 }, { "epoch": 40.853264382676144, "grad_norm": 1.6547143459320068, "learning_rate": 0.001, "loss": 1.8423, "step": 126400 }, { "epoch": 40.88558500323206, "grad_norm": 1.372111201286316, "learning_rate": 0.001, "loss": 1.8324, "step": 126500 }, { "epoch": 40.91790562378797, "grad_norm": 1.5143040418624878, "learning_rate": 0.001, "loss": 1.8382, "step": 126600 }, { "epoch": 40.95022624434389, "grad_norm": 1.6955279111862183, "learning_rate": 0.001, "loss": 1.8533, "step": 126700 }, { "epoch": 40.9825468648998, "grad_norm": 1.4255764484405518, "learning_rate": 0.001, "loss": 1.8535, "step": 126800 }, { "epoch": 41.014867485455724, "grad_norm": 1.5626089572906494, "learning_rate": 0.001, "loss": 1.765, "step": 126900 }, { "epoch": 41.04718810601164, "grad_norm": 1.950334906578064, "learning_rate": 0.001, "loss": 1.7285, "step": 127000 }, { "epoch": 41.07950872656755, "grad_norm": 1.411240816116333, "learning_rate": 0.001, "loss": 1.735, "step": 127100 }, { "epoch": 41.11182934712347, "grad_norm": 1.5985175371170044, "learning_rate": 0.001, "loss": 1.7289, "step": 127200 }, { "epoch": 41.14414996767938, "grad_norm": 1.399863600730896, "learning_rate": 0.001, "loss": 1.7341, "step": 127300 }, { "epoch": 41.1764705882353, "grad_norm": 2.0055899620056152, "learning_rate": 0.001, "loss": 1.72, "step": 127400 }, { "epoch": 41.20879120879121, "grad_norm": 1.681624174118042, "learning_rate": 0.001, "loss": 1.7439, "step": 127500 }, { "epoch": 41.241111829347126, "grad_norm": 1.5897300243377686, "learning_rate": 0.001, "loss": 1.7625, "step": 127600 }, { "epoch": 41.27343244990304, "grad_norm": 1.7503191232681274, "learning_rate": 0.001, "loss": 1.7712, "step": 127700 }, { "epoch": 41.305753070458955, "grad_norm": 1.715479850769043, "learning_rate": 0.001, "loss": 1.7624, "step": 127800 }, { "epoch": 41.33807369101487, "grad_norm": 1.6322977542877197, "learning_rate": 0.001, "loss": 1.7547, "step": 127900 }, { "epoch": 41.370394311570784, "grad_norm": 1.592178463935852, "learning_rate": 0.001, "loss": 1.7963, "step": 128000 }, { "epoch": 41.4027149321267, "grad_norm": 1.2945525646209717, "learning_rate": 0.001, "loss": 1.7699, "step": 128100 }, { "epoch": 41.43503555268261, "grad_norm": 1.4757287502288818, "learning_rate": 0.001, "loss": 1.7668, "step": 128200 }, { "epoch": 41.46735617323853, "grad_norm": 1.5124675035476685, "learning_rate": 0.001, "loss": 1.7906, "step": 128300 }, { "epoch": 41.49967679379444, "grad_norm": 2.080328941345215, "learning_rate": 0.001, "loss": 1.7906, "step": 128400 }, { "epoch": 41.53199741435036, "grad_norm": 2.1160199642181396, "learning_rate": 0.001, "loss": 1.7867, "step": 128500 }, { "epoch": 41.56431803490627, "grad_norm": 1.788028359413147, "learning_rate": 0.001, "loss": 1.7829, "step": 128600 }, { "epoch": 41.596638655462186, "grad_norm": 1.5656241178512573, "learning_rate": 0.001, "loss": 1.7748, "step": 128700 }, { "epoch": 41.6289592760181, "grad_norm": 1.5667623281478882, "learning_rate": 0.001, "loss": 1.7921, "step": 128800 }, { "epoch": 41.661279896574015, "grad_norm": 1.9110908508300781, "learning_rate": 0.001, "loss": 1.7988, "step": 128900 }, { "epoch": 41.69360051712993, "grad_norm": 1.7964316606521606, "learning_rate": 0.001, "loss": 1.811, "step": 129000 }, { "epoch": 41.725921137685845, "grad_norm": 1.4739376306533813, "learning_rate": 0.001, "loss": 1.8143, "step": 129100 }, { "epoch": 41.75824175824176, "grad_norm": 1.4890472888946533, "learning_rate": 0.001, "loss": 1.8305, "step": 129200 }, { "epoch": 41.790562378797674, "grad_norm": 1.7300723791122437, "learning_rate": 0.001, "loss": 1.7989, "step": 129300 }, { "epoch": 41.82288299935359, "grad_norm": 1.9375386238098145, "learning_rate": 0.001, "loss": 1.8203, "step": 129400 }, { "epoch": 41.8552036199095, "grad_norm": 1.827304720878601, "learning_rate": 0.001, "loss": 1.8306, "step": 129500 }, { "epoch": 41.88752424046542, "grad_norm": 1.7213237285614014, "learning_rate": 0.001, "loss": 1.8174, "step": 129600 }, { "epoch": 41.91984486102133, "grad_norm": 1.3543641567230225, "learning_rate": 0.001, "loss": 1.8407, "step": 129700 }, { "epoch": 41.95216548157725, "grad_norm": 1.6443567276000977, "learning_rate": 0.001, "loss": 1.8223, "step": 129800 }, { "epoch": 41.98448610213316, "grad_norm": 1.3840230703353882, "learning_rate": 0.001, "loss": 1.8507, "step": 129900 }, { "epoch": 42.016806722689076, "grad_norm": 2.517742872238159, "learning_rate": 0.001, "loss": 1.7459, "step": 130000 }, { "epoch": 42.04912734324499, "grad_norm": 2.185734510421753, "learning_rate": 0.001, "loss": 1.7024, "step": 130100 }, { "epoch": 42.081447963800905, "grad_norm": 2.0672836303710938, "learning_rate": 0.001, "loss": 1.7158, "step": 130200 }, { "epoch": 42.11376858435682, "grad_norm": 1.973358154296875, "learning_rate": 0.001, "loss": 1.7142, "step": 130300 }, { "epoch": 42.146089204912734, "grad_norm": 1.7326325178146362, "learning_rate": 0.001, "loss": 1.7178, "step": 130400 }, { "epoch": 42.17840982546865, "grad_norm": 2.1925511360168457, "learning_rate": 0.001, "loss": 1.7355, "step": 130500 }, { "epoch": 42.21073044602456, "grad_norm": 2.043834924697876, "learning_rate": 0.001, "loss": 1.7256, "step": 130600 }, { "epoch": 42.24305106658048, "grad_norm": 1.3037394285202026, "learning_rate": 0.001, "loss": 1.731, "step": 130700 }, { "epoch": 42.27537168713639, "grad_norm": 1.4939100742340088, "learning_rate": 0.001, "loss": 1.7371, "step": 130800 }, { "epoch": 42.30769230769231, "grad_norm": 1.9814097881317139, "learning_rate": 0.001, "loss": 1.7486, "step": 130900 }, { "epoch": 42.34001292824822, "grad_norm": 1.8973731994628906, "learning_rate": 0.001, "loss": 1.7409, "step": 131000 }, { "epoch": 42.372333548804136, "grad_norm": 1.7295153141021729, "learning_rate": 0.001, "loss": 1.7641, "step": 131100 }, { "epoch": 42.40465416936005, "grad_norm": 2.118375539779663, "learning_rate": 0.001, "loss": 1.7775, "step": 131200 }, { "epoch": 42.436974789915965, "grad_norm": 2.247931480407715, "learning_rate": 0.001, "loss": 1.753, "step": 131300 }, { "epoch": 42.46929541047188, "grad_norm": 1.6591767072677612, "learning_rate": 0.001, "loss": 1.7646, "step": 131400 }, { "epoch": 42.501616031027794, "grad_norm": 1.8841537237167358, "learning_rate": 0.001, "loss": 1.7669, "step": 131500 }, { "epoch": 42.53393665158371, "grad_norm": 1.5252450704574585, "learning_rate": 0.001, "loss": 1.7684, "step": 131600 }, { "epoch": 42.56625727213962, "grad_norm": 2.3641180992126465, "learning_rate": 0.001, "loss": 1.7768, "step": 131700 }, { "epoch": 42.59857789269554, "grad_norm": 2.023132085800171, "learning_rate": 0.001, "loss": 1.8046, "step": 131800 }, { "epoch": 42.63089851325145, "grad_norm": 1.7865345478057861, "learning_rate": 0.001, "loss": 1.7726, "step": 131900 }, { "epoch": 42.66321913380737, "grad_norm": 1.8594768047332764, "learning_rate": 0.001, "loss": 1.788, "step": 132000 }, { "epoch": 42.69553975436328, "grad_norm": 1.9835621118545532, "learning_rate": 0.001, "loss": 1.7863, "step": 132100 }, { "epoch": 42.727860374919196, "grad_norm": 1.874348759651184, "learning_rate": 0.001, "loss": 1.7948, "step": 132200 }, { "epoch": 42.76018099547511, "grad_norm": 1.708947777748108, "learning_rate": 0.001, "loss": 1.7948, "step": 132300 }, { "epoch": 42.792501616031025, "grad_norm": 1.8363604545593262, "learning_rate": 0.001, "loss": 1.8011, "step": 132400 }, { "epoch": 42.82482223658694, "grad_norm": 1.9058271646499634, "learning_rate": 0.001, "loss": 1.8117, "step": 132500 }, { "epoch": 42.857142857142854, "grad_norm": 2.1362392902374268, "learning_rate": 0.001, "loss": 1.8062, "step": 132600 }, { "epoch": 42.88946347769877, "grad_norm": 1.9057139158248901, "learning_rate": 0.001, "loss": 1.7971, "step": 132700 }, { "epoch": 42.92178409825468, "grad_norm": 2.2736151218414307, "learning_rate": 0.001, "loss": 1.8343, "step": 132800 }, { "epoch": 42.9541047188106, "grad_norm": 1.8226218223571777, "learning_rate": 0.001, "loss": 1.8219, "step": 132900 }, { "epoch": 42.98642533936652, "grad_norm": 1.5797466039657593, "learning_rate": 0.001, "loss": 1.8071, "step": 133000 }, { "epoch": 43.018745959922434, "grad_norm": 1.521481990814209, "learning_rate": 0.001, "loss": 1.753, "step": 133100 }, { "epoch": 43.05106658047835, "grad_norm": 1.5325391292572021, "learning_rate": 0.001, "loss": 1.6839, "step": 133200 }, { "epoch": 43.08338720103426, "grad_norm": 1.285569429397583, "learning_rate": 0.001, "loss": 1.6817, "step": 133300 }, { "epoch": 43.11570782159018, "grad_norm": 1.5999805927276611, "learning_rate": 0.001, "loss": 1.7031, "step": 133400 }, { "epoch": 43.14802844214609, "grad_norm": 1.6621031761169434, "learning_rate": 0.001, "loss": 1.7072, "step": 133500 }, { "epoch": 43.18034906270201, "grad_norm": 1.1231224536895752, "learning_rate": 0.001, "loss": 1.7284, "step": 133600 }, { "epoch": 43.21266968325792, "grad_norm": 2.0875513553619385, "learning_rate": 0.001, "loss": 1.7371, "step": 133700 }, { "epoch": 43.244990303813836, "grad_norm": 1.1687034368515015, "learning_rate": 0.001, "loss": 1.7061, "step": 133800 }, { "epoch": 43.27731092436975, "grad_norm": 1.332298994064331, "learning_rate": 0.001, "loss": 1.7254, "step": 133900 }, { "epoch": 43.309631544925665, "grad_norm": 1.3487528562545776, "learning_rate": 0.001, "loss": 1.7478, "step": 134000 }, { "epoch": 43.34195216548158, "grad_norm": 1.3946985006332397, "learning_rate": 0.001, "loss": 1.7475, "step": 134100 }, { "epoch": 43.374272786037494, "grad_norm": 1.4364039897918701, "learning_rate": 0.001, "loss": 1.7416, "step": 134200 }, { "epoch": 43.40659340659341, "grad_norm": 1.4674285650253296, "learning_rate": 0.001, "loss": 1.7538, "step": 134300 }, { "epoch": 43.43891402714932, "grad_norm": 1.2243691682815552, "learning_rate": 0.001, "loss": 1.7365, "step": 134400 }, { "epoch": 43.47123464770524, "grad_norm": 1.2732268571853638, "learning_rate": 0.001, "loss": 1.7684, "step": 134500 }, { "epoch": 43.50355526826115, "grad_norm": 1.8665534257888794, "learning_rate": 0.001, "loss": 1.7515, "step": 134600 }, { "epoch": 43.53587588881707, "grad_norm": 1.401395320892334, "learning_rate": 0.001, "loss": 1.7465, "step": 134700 }, { "epoch": 43.56819650937298, "grad_norm": 1.7193236351013184, "learning_rate": 0.001, "loss": 1.7725, "step": 134800 }, { "epoch": 43.600517129928896, "grad_norm": 1.4448840618133545, "learning_rate": 0.001, "loss": 1.7486, "step": 134900 }, { "epoch": 43.63283775048481, "grad_norm": 1.8662267923355103, "learning_rate": 0.001, "loss": 1.7845, "step": 135000 }, { "epoch": 43.665158371040725, "grad_norm": 1.5399407148361206, "learning_rate": 0.001, "loss": 1.7644, "step": 135100 }, { "epoch": 43.69747899159664, "grad_norm": 1.2298833131790161, "learning_rate": 0.001, "loss": 1.7841, "step": 135200 }, { "epoch": 43.729799612152554, "grad_norm": 1.181747317314148, "learning_rate": 0.001, "loss": 1.7781, "step": 135300 }, { "epoch": 43.76212023270847, "grad_norm": 1.1132404804229736, "learning_rate": 0.001, "loss": 1.7873, "step": 135400 }, { "epoch": 43.79444085326438, "grad_norm": 1.2958418130874634, "learning_rate": 0.001, "loss": 1.7973, "step": 135500 }, { "epoch": 43.8267614738203, "grad_norm": 1.3314193487167358, "learning_rate": 0.001, "loss": 1.7729, "step": 135600 }, { "epoch": 43.85908209437621, "grad_norm": 1.2268832921981812, "learning_rate": 0.001, "loss": 1.7882, "step": 135700 }, { "epoch": 43.89140271493213, "grad_norm": 1.6641273498535156, "learning_rate": 0.001, "loss": 1.796, "step": 135800 }, { "epoch": 43.92372333548804, "grad_norm": 1.4358201026916504, "learning_rate": 0.001, "loss": 1.8006, "step": 135900 }, { "epoch": 43.956043956043956, "grad_norm": 1.683240294456482, "learning_rate": 0.001, "loss": 1.789, "step": 136000 }, { "epoch": 43.98836457659987, "grad_norm": 1.3339091539382935, "learning_rate": 0.001, "loss": 1.8012, "step": 136100 }, { "epoch": 44.020685197155785, "grad_norm": 1.1389038562774658, "learning_rate": 0.001, "loss": 1.73, "step": 136200 }, { "epoch": 44.0530058177117, "grad_norm": 1.3806034326553345, "learning_rate": 0.001, "loss": 1.6848, "step": 136300 }, { "epoch": 44.085326438267614, "grad_norm": 1.372220754623413, "learning_rate": 0.001, "loss": 1.6656, "step": 136400 }, { "epoch": 44.11764705882353, "grad_norm": 1.4831829071044922, "learning_rate": 0.001, "loss": 1.693, "step": 136500 }, { "epoch": 44.14996767937944, "grad_norm": 1.418005347251892, "learning_rate": 0.001, "loss": 1.6875, "step": 136600 }, { "epoch": 44.18228829993536, "grad_norm": 1.3556125164031982, "learning_rate": 0.001, "loss": 1.706, "step": 136700 }, { "epoch": 44.21460892049127, "grad_norm": 1.2906237840652466, "learning_rate": 0.001, "loss": 1.6978, "step": 136800 }, { "epoch": 44.24692954104719, "grad_norm": 1.6397830247879028, "learning_rate": 0.001, "loss": 1.6998, "step": 136900 }, { "epoch": 44.2792501616031, "grad_norm": 1.4387032985687256, "learning_rate": 0.001, "loss": 1.7246, "step": 137000 }, { "epoch": 44.311570782159016, "grad_norm": 1.3406988382339478, "learning_rate": 0.001, "loss": 1.7006, "step": 137100 }, { "epoch": 44.34389140271493, "grad_norm": 1.217828392982483, "learning_rate": 0.001, "loss": 1.7204, "step": 137200 }, { "epoch": 44.376212023270845, "grad_norm": 1.4825609922409058, "learning_rate": 0.001, "loss": 1.7147, "step": 137300 }, { "epoch": 44.40853264382676, "grad_norm": 1.3367348909378052, "learning_rate": 0.001, "loss": 1.7242, "step": 137400 }, { "epoch": 44.440853264382675, "grad_norm": 1.2189295291900635, "learning_rate": 0.001, "loss": 1.7252, "step": 137500 }, { "epoch": 44.47317388493859, "grad_norm": 1.3985297679901123, "learning_rate": 0.001, "loss": 1.7492, "step": 137600 }, { "epoch": 44.505494505494504, "grad_norm": 1.5942193269729614, "learning_rate": 0.001, "loss": 1.7454, "step": 137700 }, { "epoch": 44.53781512605042, "grad_norm": 1.4428784847259521, "learning_rate": 0.001, "loss": 1.7478, "step": 137800 }, { "epoch": 44.57013574660633, "grad_norm": 1.6369024515151978, "learning_rate": 0.001, "loss": 1.762, "step": 137900 }, { "epoch": 44.60245636716225, "grad_norm": 1.4189107418060303, "learning_rate": 0.001, "loss": 1.7404, "step": 138000 }, { "epoch": 44.63477698771816, "grad_norm": 1.0974513292312622, "learning_rate": 0.001, "loss": 1.7539, "step": 138100 }, { "epoch": 44.66709760827408, "grad_norm": 1.4818083047866821, "learning_rate": 0.001, "loss": 1.7615, "step": 138200 }, { "epoch": 44.69941822882999, "grad_norm": 1.5206955671310425, "learning_rate": 0.001, "loss": 1.7802, "step": 138300 }, { "epoch": 44.731738849385906, "grad_norm": 1.472645878791809, "learning_rate": 0.001, "loss": 1.7725, "step": 138400 }, { "epoch": 44.76405946994182, "grad_norm": 1.1264265775680542, "learning_rate": 0.001, "loss": 1.7725, "step": 138500 }, { "epoch": 44.796380090497735, "grad_norm": 1.462930679321289, "learning_rate": 0.001, "loss": 1.7652, "step": 138600 }, { "epoch": 44.82870071105365, "grad_norm": 1.6498340368270874, "learning_rate": 0.001, "loss": 1.7759, "step": 138700 }, { "epoch": 44.861021331609564, "grad_norm": 1.0541050434112549, "learning_rate": 0.001, "loss": 1.7729, "step": 138800 }, { "epoch": 44.89334195216548, "grad_norm": 1.2006580829620361, "learning_rate": 0.001, "loss": 1.7805, "step": 138900 }, { "epoch": 44.92566257272139, "grad_norm": 1.2335492372512817, "learning_rate": 0.001, "loss": 1.7877, "step": 139000 }, { "epoch": 44.95798319327731, "grad_norm": 1.3368200063705444, "learning_rate": 0.001, "loss": 1.8003, "step": 139100 }, { "epoch": 44.99030381383322, "grad_norm": 1.6851129531860352, "learning_rate": 0.001, "loss": 1.7872, "step": 139200 }, { "epoch": 45.022624434389144, "grad_norm": 1.1418347358703613, "learning_rate": 0.001, "loss": 1.7087, "step": 139300 }, { "epoch": 45.05494505494506, "grad_norm": 1.1400333642959595, "learning_rate": 0.001, "loss": 1.6607, "step": 139400 }, { "epoch": 45.08726567550097, "grad_norm": 1.8854553699493408, "learning_rate": 0.001, "loss": 1.6703, "step": 139500 }, { "epoch": 45.11958629605689, "grad_norm": 1.4569580554962158, "learning_rate": 0.001, "loss": 1.6802, "step": 139600 }, { "epoch": 45.1519069166128, "grad_norm": 1.248537302017212, "learning_rate": 0.001, "loss": 1.65, "step": 139700 }, { "epoch": 45.18422753716872, "grad_norm": 1.1826552152633667, "learning_rate": 0.001, "loss": 1.679, "step": 139800 }, { "epoch": 45.21654815772463, "grad_norm": 1.2726597785949707, "learning_rate": 0.001, "loss": 1.6903, "step": 139900 }, { "epoch": 45.248868778280546, "grad_norm": 1.6515134572982788, "learning_rate": 0.001, "loss": 1.6979, "step": 140000 }, { "epoch": 45.28118939883646, "grad_norm": 1.4121181964874268, "learning_rate": 0.001, "loss": 1.6842, "step": 140100 }, { "epoch": 45.313510019392375, "grad_norm": 1.5250076055526733, "learning_rate": 0.001, "loss": 1.6971, "step": 140200 }, { "epoch": 45.34583063994829, "grad_norm": 1.5536643266677856, "learning_rate": 0.001, "loss": 1.7001, "step": 140300 }, { "epoch": 45.378151260504204, "grad_norm": 1.589000940322876, "learning_rate": 0.001, "loss": 1.7172, "step": 140400 }, { "epoch": 45.41047188106012, "grad_norm": 1.004062533378601, "learning_rate": 0.001, "loss": 1.7173, "step": 140500 }, { "epoch": 45.44279250161603, "grad_norm": 1.5746830701828003, "learning_rate": 0.001, "loss": 1.7233, "step": 140600 }, { "epoch": 45.47511312217195, "grad_norm": 1.1398859024047852, "learning_rate": 0.001, "loss": 1.7258, "step": 140700 }, { "epoch": 45.50743374272786, "grad_norm": 1.3511697053909302, "learning_rate": 0.001, "loss": 1.7436, "step": 140800 }, { "epoch": 45.53975436328378, "grad_norm": 1.4493550062179565, "learning_rate": 0.001, "loss": 1.7314, "step": 140900 }, { "epoch": 45.57207498383969, "grad_norm": 1.2301589250564575, "learning_rate": 0.001, "loss": 1.7319, "step": 141000 }, { "epoch": 45.604395604395606, "grad_norm": 1.3222787380218506, "learning_rate": 0.001, "loss": 1.742, "step": 141100 }, { "epoch": 45.63671622495152, "grad_norm": 1.094870686531067, "learning_rate": 0.001, "loss": 1.7492, "step": 141200 }, { "epoch": 45.669036845507435, "grad_norm": 1.4017021656036377, "learning_rate": 0.001, "loss": 1.7559, "step": 141300 }, { "epoch": 45.70135746606335, "grad_norm": 1.5024820566177368, "learning_rate": 0.001, "loss": 1.7335, "step": 141400 }, { "epoch": 45.733678086619264, "grad_norm": 1.5428948402404785, "learning_rate": 0.001, "loss": 1.7629, "step": 141500 }, { "epoch": 45.76599870717518, "grad_norm": 1.2330937385559082, "learning_rate": 0.001, "loss": 1.771, "step": 141600 }, { "epoch": 45.79831932773109, "grad_norm": 1.2731399536132812, "learning_rate": 0.001, "loss": 1.7611, "step": 141700 }, { "epoch": 45.83063994828701, "grad_norm": 1.2634693384170532, "learning_rate": 0.001, "loss": 1.7557, "step": 141800 }, { "epoch": 45.86296056884292, "grad_norm": 1.2590782642364502, "learning_rate": 0.001, "loss": 1.7611, "step": 141900 }, { "epoch": 45.89528118939884, "grad_norm": 1.3432759046554565, "learning_rate": 0.001, "loss": 1.7746, "step": 142000 }, { "epoch": 45.92760180995475, "grad_norm": 1.6455352306365967, "learning_rate": 0.001, "loss": 1.7871, "step": 142100 }, { "epoch": 45.959922430510666, "grad_norm": 1.1180686950683594, "learning_rate": 0.001, "loss": 1.762, "step": 142200 }, { "epoch": 45.99224305106658, "grad_norm": 1.372404932975769, "learning_rate": 0.001, "loss": 1.7849, "step": 142300 }, { "epoch": 46.024563671622495, "grad_norm": 1.215472936630249, "learning_rate": 0.001, "loss": 1.6824, "step": 142400 }, { "epoch": 46.05688429217841, "grad_norm": 1.3119868040084839, "learning_rate": 0.001, "loss": 1.6581, "step": 142500 }, { "epoch": 46.089204912734324, "grad_norm": 1.2721792459487915, "learning_rate": 0.001, "loss": 1.6457, "step": 142600 }, { "epoch": 46.12152553329024, "grad_norm": 1.3706724643707275, "learning_rate": 0.001, "loss": 1.651, "step": 142700 }, { "epoch": 46.15384615384615, "grad_norm": 1.2131431102752686, "learning_rate": 0.001, "loss": 1.6697, "step": 142800 }, { "epoch": 46.18616677440207, "grad_norm": 1.4040346145629883, "learning_rate": 0.001, "loss": 1.659, "step": 142900 }, { "epoch": 46.21848739495798, "grad_norm": 1.5169869661331177, "learning_rate": 0.001, "loss": 1.6881, "step": 143000 }, { "epoch": 46.2508080155139, "grad_norm": 1.3575830459594727, "learning_rate": 0.001, "loss": 1.684, "step": 143100 }, { "epoch": 46.28312863606981, "grad_norm": 1.4115160703659058, "learning_rate": 0.001, "loss": 1.6892, "step": 143200 }, { "epoch": 46.315449256625726, "grad_norm": 1.3648401498794556, "learning_rate": 0.001, "loss": 1.6976, "step": 143300 }, { "epoch": 46.34776987718164, "grad_norm": 1.5159053802490234, "learning_rate": 0.001, "loss": 1.7084, "step": 143400 }, { "epoch": 46.380090497737555, "grad_norm": 1.3242292404174805, "learning_rate": 0.001, "loss": 1.6906, "step": 143500 }, { "epoch": 46.41241111829347, "grad_norm": 1.2734252214431763, "learning_rate": 0.001, "loss": 1.6991, "step": 143600 }, { "epoch": 46.444731738849384, "grad_norm": 1.3581430912017822, "learning_rate": 0.001, "loss": 1.7096, "step": 143700 }, { "epoch": 46.4770523594053, "grad_norm": 1.5372655391693115, "learning_rate": 0.001, "loss": 1.7244, "step": 143800 }, { "epoch": 46.50937297996121, "grad_norm": 1.160895586013794, "learning_rate": 0.001, "loss": 1.7147, "step": 143900 }, { "epoch": 46.54169360051713, "grad_norm": 1.1761196851730347, "learning_rate": 0.001, "loss": 1.7101, "step": 144000 }, { "epoch": 46.57401422107304, "grad_norm": 1.3373456001281738, "learning_rate": 0.001, "loss": 1.7102, "step": 144100 }, { "epoch": 46.60633484162896, "grad_norm": 1.1968778371810913, "learning_rate": 0.001, "loss": 1.7149, "step": 144200 }, { "epoch": 46.63865546218487, "grad_norm": 1.6086446046829224, "learning_rate": 0.001, "loss": 1.7136, "step": 144300 }, { "epoch": 46.670976082740786, "grad_norm": 1.4648292064666748, "learning_rate": 0.001, "loss": 1.7364, "step": 144400 }, { "epoch": 46.7032967032967, "grad_norm": 1.2216010093688965, "learning_rate": 0.001, "loss": 1.7496, "step": 144500 }, { "epoch": 46.735617323852615, "grad_norm": 1.333053708076477, "learning_rate": 0.001, "loss": 1.7463, "step": 144600 }, { "epoch": 46.76793794440853, "grad_norm": 1.3574227094650269, "learning_rate": 0.001, "loss": 1.7516, "step": 144700 }, { "epoch": 46.800258564964444, "grad_norm": 1.2774755954742432, "learning_rate": 0.001, "loss": 1.748, "step": 144800 }, { "epoch": 46.83257918552036, "grad_norm": 1.1617026329040527, "learning_rate": 0.001, "loss": 1.7472, "step": 144900 }, { "epoch": 46.864899806076274, "grad_norm": 1.248290777206421, "learning_rate": 0.001, "loss": 1.7516, "step": 145000 }, { "epoch": 46.89722042663219, "grad_norm": 0.983884871006012, "learning_rate": 0.001, "loss": 1.7464, "step": 145100 }, { "epoch": 46.9295410471881, "grad_norm": 1.3453645706176758, "learning_rate": 0.001, "loss": 1.7557, "step": 145200 }, { "epoch": 46.96186166774402, "grad_norm": 1.257980465888977, "learning_rate": 0.001, "loss": 1.7635, "step": 145300 }, { "epoch": 46.99418228829994, "grad_norm": 1.096062421798706, "learning_rate": 0.001, "loss": 1.7539, "step": 145400 }, { "epoch": 47.02650290885585, "grad_norm": 1.441990613937378, "learning_rate": 0.001, "loss": 1.656, "step": 145500 }, { "epoch": 47.05882352941177, "grad_norm": 1.2403453588485718, "learning_rate": 0.001, "loss": 1.6452, "step": 145600 }, { "epoch": 47.09114414996768, "grad_norm": 1.223830223083496, "learning_rate": 0.001, "loss": 1.6356, "step": 145700 }, { "epoch": 47.1234647705236, "grad_norm": 1.210126519203186, "learning_rate": 0.001, "loss": 1.645, "step": 145800 }, { "epoch": 47.15578539107951, "grad_norm": 1.6140942573547363, "learning_rate": 0.001, "loss": 1.651, "step": 145900 }, { "epoch": 47.188106011635426, "grad_norm": 1.7467138767242432, "learning_rate": 0.001, "loss": 1.6784, "step": 146000 }, { "epoch": 47.22042663219134, "grad_norm": 1.1775304079055786, "learning_rate": 0.001, "loss": 1.6649, "step": 146100 }, { "epoch": 47.252747252747255, "grad_norm": 1.9474658966064453, "learning_rate": 0.001, "loss": 1.6672, "step": 146200 }, { "epoch": 47.28506787330317, "grad_norm": 1.4976410865783691, "learning_rate": 0.001, "loss": 1.6637, "step": 146300 }, { "epoch": 47.317388493859085, "grad_norm": 1.3201075792312622, "learning_rate": 0.001, "loss": 1.676, "step": 146400 }, { "epoch": 47.349709114415, "grad_norm": 1.6384514570236206, "learning_rate": 0.001, "loss": 1.6676, "step": 146500 }, { "epoch": 47.382029734970914, "grad_norm": 1.6562048196792603, "learning_rate": 0.001, "loss": 1.6743, "step": 146600 }, { "epoch": 47.41435035552683, "grad_norm": 1.5702561140060425, "learning_rate": 0.001, "loss": 1.6871, "step": 146700 }, { "epoch": 47.44667097608274, "grad_norm": 1.4084358215332031, "learning_rate": 0.001, "loss": 1.6916, "step": 146800 }, { "epoch": 47.47899159663866, "grad_norm": 1.3545585870742798, "learning_rate": 0.001, "loss": 1.7055, "step": 146900 }, { "epoch": 47.51131221719457, "grad_norm": 1.1715108156204224, "learning_rate": 0.001, "loss": 1.6969, "step": 147000 }, { "epoch": 47.543632837750486, "grad_norm": 1.081047534942627, "learning_rate": 0.001, "loss": 1.6795, "step": 147100 }, { "epoch": 47.5759534583064, "grad_norm": 1.5891271829605103, "learning_rate": 0.001, "loss": 1.7091, "step": 147200 }, { "epoch": 47.608274078862316, "grad_norm": 1.2145577669143677, "learning_rate": 0.001, "loss": 1.7007, "step": 147300 }, { "epoch": 47.64059469941823, "grad_norm": 1.4922897815704346, "learning_rate": 0.001, "loss": 1.7244, "step": 147400 }, { "epoch": 47.672915319974145, "grad_norm": 1.6337180137634277, "learning_rate": 0.001, "loss": 1.7369, "step": 147500 }, { "epoch": 47.70523594053006, "grad_norm": 1.620103359222412, "learning_rate": 0.001, "loss": 1.7342, "step": 147600 }, { "epoch": 47.737556561085974, "grad_norm": 1.6328498125076294, "learning_rate": 0.001, "loss": 1.7209, "step": 147700 }, { "epoch": 47.76987718164189, "grad_norm": 1.3009995222091675, "learning_rate": 0.001, "loss": 1.7132, "step": 147800 }, { "epoch": 47.8021978021978, "grad_norm": 1.401782512664795, "learning_rate": 0.001, "loss": 1.7481, "step": 147900 }, { "epoch": 47.83451842275372, "grad_norm": 1.2889764308929443, "learning_rate": 0.001, "loss": 1.7267, "step": 148000 }, { "epoch": 47.86683904330963, "grad_norm": 1.810412049293518, "learning_rate": 0.001, "loss": 1.7317, "step": 148100 }, { "epoch": 47.89915966386555, "grad_norm": 1.6493173837661743, "learning_rate": 0.001, "loss": 1.7424, "step": 148200 }, { "epoch": 47.93148028442146, "grad_norm": 1.5466930866241455, "learning_rate": 0.001, "loss": 1.7395, "step": 148300 }, { "epoch": 47.963800904977376, "grad_norm": 1.4534482955932617, "learning_rate": 0.001, "loss": 1.7381, "step": 148400 }, { "epoch": 47.99612152553329, "grad_norm": 1.7018836736679077, "learning_rate": 0.001, "loss": 1.744, "step": 148500 }, { "epoch": 48.028442146089205, "grad_norm": 1.7868794202804565, "learning_rate": 0.001, "loss": 1.639, "step": 148600 }, { "epoch": 48.06076276664512, "grad_norm": 1.4775114059448242, "learning_rate": 0.001, "loss": 1.6331, "step": 148700 }, { "epoch": 48.093083387201034, "grad_norm": 1.349012017250061, "learning_rate": 0.001, "loss": 1.6202, "step": 148800 }, { "epoch": 48.12540400775695, "grad_norm": 1.6332101821899414, "learning_rate": 0.001, "loss": 1.6312, "step": 148900 }, { "epoch": 48.15772462831286, "grad_norm": 1.624573826789856, "learning_rate": 0.001, "loss": 1.653, "step": 149000 }, { "epoch": 48.19004524886878, "grad_norm": 1.2802793979644775, "learning_rate": 0.001, "loss": 1.6357, "step": 149100 }, { "epoch": 48.22236586942469, "grad_norm": 1.2156689167022705, "learning_rate": 0.001, "loss": 1.656, "step": 149200 }, { "epoch": 48.25468648998061, "grad_norm": 1.5607578754425049, "learning_rate": 0.001, "loss": 1.6587, "step": 149300 }, { "epoch": 48.28700711053652, "grad_norm": 1.479413390159607, "learning_rate": 0.001, "loss": 1.6439, "step": 149400 }, { "epoch": 48.319327731092436, "grad_norm": 1.3263925313949585, "learning_rate": 0.001, "loss": 1.66, "step": 149500 }, { "epoch": 48.35164835164835, "grad_norm": 1.4417411088943481, "learning_rate": 0.001, "loss": 1.6677, "step": 149600 }, { "epoch": 48.383968972204265, "grad_norm": 1.585517168045044, "learning_rate": 0.001, "loss": 1.6621, "step": 149700 }, { "epoch": 48.41628959276018, "grad_norm": 1.6125644445419312, "learning_rate": 0.001, "loss": 1.675, "step": 149800 }, { "epoch": 48.448610213316094, "grad_norm": 1.4252173900604248, "learning_rate": 0.001, "loss": 1.6818, "step": 149900 }, { "epoch": 48.48093083387201, "grad_norm": 1.4481408596038818, "learning_rate": 0.001, "loss": 1.6734, "step": 150000 }, { "epoch": 48.51325145442792, "grad_norm": 1.7809571027755737, "learning_rate": 0.001, "loss": 1.6738, "step": 150100 }, { "epoch": 48.54557207498384, "grad_norm": 1.8779230117797852, "learning_rate": 0.001, "loss": 1.6895, "step": 150200 }, { "epoch": 48.57789269553975, "grad_norm": 1.2452175617218018, "learning_rate": 0.001, "loss": 1.6788, "step": 150300 }, { "epoch": 48.61021331609567, "grad_norm": 1.8720242977142334, "learning_rate": 0.001, "loss": 1.6983, "step": 150400 }, { "epoch": 48.64253393665158, "grad_norm": 1.5265681743621826, "learning_rate": 0.001, "loss": 1.7017, "step": 150500 }, { "epoch": 48.674854557207496, "grad_norm": 1.6183137893676758, "learning_rate": 0.001, "loss": 1.7047, "step": 150600 }, { "epoch": 48.70717517776341, "grad_norm": 1.3076919317245483, "learning_rate": 0.001, "loss": 1.7059, "step": 150700 }, { "epoch": 48.739495798319325, "grad_norm": 1.2748409509658813, "learning_rate": 0.001, "loss": 1.7215, "step": 150800 }, { "epoch": 48.77181641887524, "grad_norm": 1.4130632877349854, "learning_rate": 0.001, "loss": 1.7318, "step": 150900 }, { "epoch": 48.804137039431154, "grad_norm": 1.500614047050476, "learning_rate": 0.001, "loss": 1.7096, "step": 151000 }, { "epoch": 48.83645765998707, "grad_norm": 1.2465304136276245, "learning_rate": 0.001, "loss": 1.7191, "step": 151100 }, { "epoch": 48.86877828054298, "grad_norm": 1.5916765928268433, "learning_rate": 0.001, "loss": 1.7195, "step": 151200 }, { "epoch": 48.9010989010989, "grad_norm": 1.354617714881897, "learning_rate": 0.001, "loss": 1.7126, "step": 151300 }, { "epoch": 48.93341952165481, "grad_norm": 1.574227213859558, "learning_rate": 0.001, "loss": 1.7338, "step": 151400 }, { "epoch": 48.96574014221073, "grad_norm": 1.5032780170440674, "learning_rate": 0.001, "loss": 1.7438, "step": 151500 }, { "epoch": 48.99806076276664, "grad_norm": 1.5653413534164429, "learning_rate": 0.001, "loss": 1.7113, "step": 151600 }, { "epoch": 49.03038138332256, "grad_norm": 1.396582007408142, "learning_rate": 0.001, "loss": 1.6055, "step": 151700 }, { "epoch": 49.06270200387848, "grad_norm": 1.868228793144226, "learning_rate": 0.001, "loss": 1.6014, "step": 151800 }, { "epoch": 49.09502262443439, "grad_norm": 1.523720145225525, "learning_rate": 0.001, "loss": 1.6105, "step": 151900 }, { "epoch": 49.12734324499031, "grad_norm": 1.501035213470459, "learning_rate": 0.001, "loss": 1.6247, "step": 152000 }, { "epoch": 49.15966386554622, "grad_norm": 1.7720264196395874, "learning_rate": 0.001, "loss": 1.6375, "step": 152100 }, { "epoch": 49.191984486102136, "grad_norm": 1.2451496124267578, "learning_rate": 0.001, "loss": 1.6381, "step": 152200 }, { "epoch": 49.22430510665805, "grad_norm": 1.6157736778259277, "learning_rate": 0.001, "loss": 1.643, "step": 152300 }, { "epoch": 49.256625727213965, "grad_norm": 1.3556593656539917, "learning_rate": 0.001, "loss": 1.6422, "step": 152400 }, { "epoch": 49.28894634776988, "grad_norm": 1.572458028793335, "learning_rate": 0.001, "loss": 1.6565, "step": 152500 }, { "epoch": 49.321266968325794, "grad_norm": 1.5787707567214966, "learning_rate": 0.001, "loss": 1.651, "step": 152600 }, { "epoch": 49.35358758888171, "grad_norm": 1.7005757093429565, "learning_rate": 0.001, "loss": 1.6402, "step": 152700 }, { "epoch": 49.38590820943762, "grad_norm": 1.835395097732544, "learning_rate": 0.001, "loss": 1.6582, "step": 152800 }, { "epoch": 49.41822882999354, "grad_norm": 1.7922719717025757, "learning_rate": 0.001, "loss": 1.6683, "step": 152900 }, { "epoch": 49.45054945054945, "grad_norm": 1.5180412530899048, "learning_rate": 0.001, "loss": 1.6652, "step": 153000 }, { "epoch": 49.48287007110537, "grad_norm": 1.7080254554748535, "learning_rate": 0.001, "loss": 1.6608, "step": 153100 }, { "epoch": 49.51519069166128, "grad_norm": 1.5227432250976562, "learning_rate": 0.001, "loss": 1.6777, "step": 153200 }, { "epoch": 49.547511312217196, "grad_norm": 2.0523929595947266, "learning_rate": 0.001, "loss": 1.6808, "step": 153300 }, { "epoch": 49.57983193277311, "grad_norm": 1.4774620532989502, "learning_rate": 0.001, "loss": 1.6682, "step": 153400 }, { "epoch": 49.612152553329025, "grad_norm": 1.576160192489624, "learning_rate": 0.001, "loss": 1.6844, "step": 153500 }, { "epoch": 49.64447317388494, "grad_norm": 1.4338222742080688, "learning_rate": 0.001, "loss": 1.6934, "step": 153600 }, { "epoch": 49.676793794440854, "grad_norm": 1.6853586435317993, "learning_rate": 0.001, "loss": 1.6855, "step": 153700 }, { "epoch": 49.70911441499677, "grad_norm": 1.502305030822754, "learning_rate": 0.001, "loss": 1.6999, "step": 153800 }, { "epoch": 49.74143503555268, "grad_norm": 1.3176287412643433, "learning_rate": 0.001, "loss": 1.6802, "step": 153900 }, { "epoch": 49.7737556561086, "grad_norm": 1.6562001705169678, "learning_rate": 0.001, "loss": 1.6989, "step": 154000 }, { "epoch": 49.80607627666451, "grad_norm": 1.3866074085235596, "learning_rate": 0.001, "loss": 1.7105, "step": 154100 }, { "epoch": 49.83839689722043, "grad_norm": 1.3695135116577148, "learning_rate": 0.001, "loss": 1.685, "step": 154200 }, { "epoch": 49.87071751777634, "grad_norm": 1.4399651288986206, "learning_rate": 0.001, "loss": 1.706, "step": 154300 }, { "epoch": 49.903038138332256, "grad_norm": 1.3470892906188965, "learning_rate": 0.001, "loss": 1.7119, "step": 154400 }, { "epoch": 49.93535875888817, "grad_norm": 2.029789924621582, "learning_rate": 0.001, "loss": 1.7271, "step": 154500 }, { "epoch": 49.967679379444085, "grad_norm": 1.4181112051010132, "learning_rate": 0.001, "loss": 1.7194, "step": 154600 }, { "epoch": 50.0, "grad_norm": 3.0098884105682373, "learning_rate": 0.001, "loss": 1.6978, "step": 154700 }, { "epoch": 50.032320620555915, "grad_norm": 1.7859961986541748, "learning_rate": 0.001, "loss": 1.5883, "step": 154800 }, { "epoch": 50.06464124111183, "grad_norm": 1.9198648929595947, "learning_rate": 0.001, "loss": 1.6018, "step": 154900 }, { "epoch": 50.096961861667744, "grad_norm": 1.7625179290771484, "learning_rate": 0.001, "loss": 1.6018, "step": 155000 }, { "epoch": 50.12928248222366, "grad_norm": 1.904485821723938, "learning_rate": 0.001, "loss": 1.61, "step": 155100 }, { "epoch": 50.16160310277957, "grad_norm": 2.516070604324341, "learning_rate": 0.001, "loss": 1.5987, "step": 155200 }, { "epoch": 50.19392372333549, "grad_norm": 2.0027501583099365, "learning_rate": 0.001, "loss": 1.6137, "step": 155300 }, { "epoch": 50.2262443438914, "grad_norm": 1.994936227798462, "learning_rate": 0.001, "loss": 1.6372, "step": 155400 }, { "epoch": 50.25856496444732, "grad_norm": 2.642782688140869, "learning_rate": 0.001, "loss": 1.6412, "step": 155500 }, { "epoch": 50.29088558500323, "grad_norm": 1.691218614578247, "learning_rate": 0.001, "loss": 1.6243, "step": 155600 }, { "epoch": 50.323206205559146, "grad_norm": 1.6818300485610962, "learning_rate": 0.001, "loss": 1.6463, "step": 155700 }, { "epoch": 50.35552682611506, "grad_norm": 2.295081377029419, "learning_rate": 0.001, "loss": 1.6292, "step": 155800 }, { "epoch": 50.387847446670975, "grad_norm": 2.1138439178466797, "learning_rate": 0.001, "loss": 1.6476, "step": 155900 }, { "epoch": 50.42016806722689, "grad_norm": 2.1354689598083496, "learning_rate": 0.001, "loss": 1.6484, "step": 156000 }, { "epoch": 50.452488687782804, "grad_norm": 1.5899136066436768, "learning_rate": 0.001, "loss": 1.6417, "step": 156100 }, { "epoch": 50.48480930833872, "grad_norm": 1.873875379562378, "learning_rate": 0.001, "loss": 1.6577, "step": 156200 }, { "epoch": 50.51712992889463, "grad_norm": 1.5048267841339111, "learning_rate": 0.001, "loss": 1.657, "step": 156300 }, { "epoch": 50.54945054945055, "grad_norm": 2.189378499984741, "learning_rate": 0.001, "loss": 1.663, "step": 156400 }, { "epoch": 50.58177117000646, "grad_norm": 2.1354687213897705, "learning_rate": 0.001, "loss": 1.6667, "step": 156500 }, { "epoch": 50.61409179056238, "grad_norm": 1.819714069366455, "learning_rate": 0.001, "loss": 1.6824, "step": 156600 }, { "epoch": 50.64641241111829, "grad_norm": 2.5435047149658203, "learning_rate": 0.001, "loss": 1.6583, "step": 156700 }, { "epoch": 50.678733031674206, "grad_norm": 1.9671624898910522, "learning_rate": 0.001, "loss": 1.6705, "step": 156800 }, { "epoch": 50.71105365223012, "grad_norm": 1.9127241373062134, "learning_rate": 0.001, "loss": 1.6981, "step": 156900 }, { "epoch": 50.743374272786035, "grad_norm": 1.9809879064559937, "learning_rate": 0.001, "loss": 1.6902, "step": 157000 }, { "epoch": 50.77569489334195, "grad_norm": 2.044308662414551, "learning_rate": 0.001, "loss": 1.7047, "step": 157100 }, { "epoch": 50.808015513897864, "grad_norm": 1.9656683206558228, "learning_rate": 0.001, "loss": 1.6941, "step": 157200 }, { "epoch": 50.84033613445378, "grad_norm": 1.8355093002319336, "learning_rate": 0.001, "loss": 1.7009, "step": 157300 }, { "epoch": 50.87265675500969, "grad_norm": 2.390496015548706, "learning_rate": 0.001, "loss": 1.7003, "step": 157400 }, { "epoch": 50.90497737556561, "grad_norm": 2.4113669395446777, "learning_rate": 0.001, "loss": 1.6929, "step": 157500 }, { "epoch": 50.93729799612152, "grad_norm": 2.240635633468628, "learning_rate": 0.001, "loss": 1.6955, "step": 157600 }, { "epoch": 50.96961861667744, "grad_norm": 2.356112480163574, "learning_rate": 0.001, "loss": 1.7039, "step": 157700 }, { "epoch": 51.00193923723336, "grad_norm": 1.3113300800323486, "learning_rate": 0.001, "loss": 1.7148, "step": 157800 }, { "epoch": 51.03425985778927, "grad_norm": 1.3943908214569092, "learning_rate": 0.001, "loss": 1.5684, "step": 157900 }, { "epoch": 51.06658047834519, "grad_norm": 1.5345698595046997, "learning_rate": 0.001, "loss": 1.5841, "step": 158000 }, { "epoch": 51.0989010989011, "grad_norm": 1.376284122467041, "learning_rate": 0.001, "loss": 1.5904, "step": 158100 }, { "epoch": 51.13122171945702, "grad_norm": 1.8864432573318481, "learning_rate": 0.001, "loss": 1.5832, "step": 158200 }, { "epoch": 51.16354234001293, "grad_norm": 1.077499508857727, "learning_rate": 0.001, "loss": 1.6045, "step": 158300 }, { "epoch": 51.195862960568846, "grad_norm": 1.413830041885376, "learning_rate": 0.001, "loss": 1.622, "step": 158400 }, { "epoch": 51.22818358112476, "grad_norm": 1.317996859550476, "learning_rate": 0.001, "loss": 1.6151, "step": 158500 }, { "epoch": 51.260504201680675, "grad_norm": 1.800900936126709, "learning_rate": 0.001, "loss": 1.6143, "step": 158600 }, { "epoch": 51.29282482223659, "grad_norm": 1.4891034364700317, "learning_rate": 0.001, "loss": 1.6113, "step": 158700 }, { "epoch": 51.325145442792504, "grad_norm": 1.776503324508667, "learning_rate": 0.001, "loss": 1.6184, "step": 158800 }, { "epoch": 51.35746606334842, "grad_norm": 1.4487277269363403, "learning_rate": 0.001, "loss": 1.6159, "step": 158900 }, { "epoch": 51.38978668390433, "grad_norm": 1.4836653470993042, "learning_rate": 0.001, "loss": 1.6281, "step": 159000 }, { "epoch": 51.42210730446025, "grad_norm": 1.5971252918243408, "learning_rate": 0.001, "loss": 1.6469, "step": 159100 }, { "epoch": 51.45442792501616, "grad_norm": 1.2963101863861084, "learning_rate": 0.001, "loss": 1.6385, "step": 159200 }, { "epoch": 51.48674854557208, "grad_norm": 1.1028395891189575, "learning_rate": 0.001, "loss": 1.6423, "step": 159300 }, { "epoch": 51.51906916612799, "grad_norm": 2.0811383724212646, "learning_rate": 0.001, "loss": 1.6412, "step": 159400 }, { "epoch": 51.551389786683906, "grad_norm": 1.5059459209442139, "learning_rate": 0.001, "loss": 1.6582, "step": 159500 }, { "epoch": 51.58371040723982, "grad_norm": 1.9516657590866089, "learning_rate": 0.001, "loss": 1.6565, "step": 159600 }, { "epoch": 51.616031027795735, "grad_norm": 1.6311758756637573, "learning_rate": 0.001, "loss": 1.6644, "step": 159700 }, { "epoch": 51.64835164835165, "grad_norm": 1.7732529640197754, "learning_rate": 0.001, "loss": 1.6546, "step": 159800 }, { "epoch": 51.680672268907564, "grad_norm": 1.8546462059020996, "learning_rate": 0.001, "loss": 1.6655, "step": 159900 }, { "epoch": 51.71299288946348, "grad_norm": 1.6930227279663086, "learning_rate": 0.001, "loss": 1.6688, "step": 160000 }, { "epoch": 51.74531351001939, "grad_norm": 1.3648531436920166, "learning_rate": 0.001, "loss": 1.6633, "step": 160100 }, { "epoch": 51.77763413057531, "grad_norm": 1.6870195865631104, "learning_rate": 0.001, "loss": 1.6871, "step": 160200 }, { "epoch": 51.80995475113122, "grad_norm": 1.2742482423782349, "learning_rate": 0.001, "loss": 1.6787, "step": 160300 }, { "epoch": 51.84227537168714, "grad_norm": 1.6389504671096802, "learning_rate": 0.001, "loss": 1.6658, "step": 160400 }, { "epoch": 51.87459599224305, "grad_norm": 1.4357678890228271, "learning_rate": 0.001, "loss": 1.6879, "step": 160500 }, { "epoch": 51.906916612798966, "grad_norm": 1.522757887840271, "learning_rate": 0.001, "loss": 1.6811, "step": 160600 }, { "epoch": 51.93923723335488, "grad_norm": 1.5563929080963135, "learning_rate": 0.001, "loss": 1.6796, "step": 160700 }, { "epoch": 51.971557853910795, "grad_norm": 1.4780327081680298, "learning_rate": 0.001, "loss": 1.7072, "step": 160800 }, { "epoch": 52.00387847446671, "grad_norm": 1.1867820024490356, "learning_rate": 0.001, "loss": 1.7241, "step": 160900 }, { "epoch": 52.036199095022624, "grad_norm": 1.198205828666687, "learning_rate": 0.001, "loss": 1.5527, "step": 161000 }, { "epoch": 52.06851971557854, "grad_norm": 1.3584864139556885, "learning_rate": 0.001, "loss": 1.5493, "step": 161100 }, { "epoch": 52.10084033613445, "grad_norm": 2.0809905529022217, "learning_rate": 0.001, "loss": 1.5896, "step": 161200 }, { "epoch": 52.13316095669037, "grad_norm": 1.5358004570007324, "learning_rate": 0.001, "loss": 1.5918, "step": 161300 }, { "epoch": 52.16548157724628, "grad_norm": 1.1696178913116455, "learning_rate": 0.001, "loss": 1.5772, "step": 161400 }, { "epoch": 52.1978021978022, "grad_norm": 1.2811452150344849, "learning_rate": 0.001, "loss": 1.5905, "step": 161500 }, { "epoch": 52.23012281835811, "grad_norm": 1.4328819513320923, "learning_rate": 0.001, "loss": 1.6077, "step": 161600 }, { "epoch": 52.262443438914026, "grad_norm": 1.3378756046295166, "learning_rate": 0.001, "loss": 1.616, "step": 161700 }, { "epoch": 52.29476405946994, "grad_norm": 1.2815625667572021, "learning_rate": 0.001, "loss": 1.6112, "step": 161800 }, { "epoch": 52.327084680025855, "grad_norm": 1.7682138681411743, "learning_rate": 0.001, "loss": 1.6231, "step": 161900 }, { "epoch": 52.35940530058177, "grad_norm": 1.4320544004440308, "learning_rate": 0.001, "loss": 1.6133, "step": 162000 }, { "epoch": 52.391725921137684, "grad_norm": 1.156111717224121, "learning_rate": 0.001, "loss": 1.6199, "step": 162100 }, { "epoch": 52.4240465416936, "grad_norm": 1.2384657859802246, "learning_rate": 0.001, "loss": 1.6203, "step": 162200 }, { "epoch": 52.456367162249514, "grad_norm": 1.2537254095077515, "learning_rate": 0.001, "loss": 1.6271, "step": 162300 }, { "epoch": 52.48868778280543, "grad_norm": 1.0931706428527832, "learning_rate": 0.001, "loss": 1.6172, "step": 162400 }, { "epoch": 52.52100840336134, "grad_norm": 1.236507773399353, "learning_rate": 0.001, "loss": 1.6281, "step": 162500 }, { "epoch": 52.55332902391726, "grad_norm": 1.4299321174621582, "learning_rate": 0.001, "loss": 1.6391, "step": 162600 }, { "epoch": 52.58564964447317, "grad_norm": 1.1509603261947632, "learning_rate": 0.001, "loss": 1.6387, "step": 162700 }, { "epoch": 52.617970265029086, "grad_norm": 1.17874276638031, "learning_rate": 0.001, "loss": 1.6525, "step": 162800 }, { "epoch": 52.650290885585, "grad_norm": 1.652890682220459, "learning_rate": 0.001, "loss": 1.6555, "step": 162900 }, { "epoch": 52.682611506140915, "grad_norm": 1.334714651107788, "learning_rate": 0.001, "loss": 1.6563, "step": 163000 }, { "epoch": 52.71493212669683, "grad_norm": 1.5494811534881592, "learning_rate": 0.001, "loss": 1.657, "step": 163100 }, { "epoch": 52.747252747252745, "grad_norm": 2.0530385971069336, "learning_rate": 0.001, "loss": 1.6683, "step": 163200 }, { "epoch": 52.77957336780866, "grad_norm": 1.7362953424453735, "learning_rate": 0.001, "loss": 1.6581, "step": 163300 }, { "epoch": 52.811893988364574, "grad_norm": 1.2621122598648071, "learning_rate": 0.001, "loss": 1.6536, "step": 163400 }, { "epoch": 52.84421460892049, "grad_norm": 1.248063325881958, "learning_rate": 0.001, "loss": 1.6698, "step": 163500 }, { "epoch": 52.8765352294764, "grad_norm": 1.3443856239318848, "learning_rate": 0.001, "loss": 1.6908, "step": 163600 }, { "epoch": 52.90885585003232, "grad_norm": 1.2341312170028687, "learning_rate": 0.001, "loss": 1.6762, "step": 163700 }, { "epoch": 52.94117647058823, "grad_norm": 1.3637555837631226, "learning_rate": 0.001, "loss": 1.6815, "step": 163800 }, { "epoch": 52.97349709114415, "grad_norm": 1.547029733657837, "learning_rate": 0.001, "loss": 1.6881, "step": 163900 }, { "epoch": 53.00581771170007, "grad_norm": 1.252366065979004, "learning_rate": 0.001, "loss": 1.6504, "step": 164000 }, { "epoch": 53.03813833225598, "grad_norm": 1.2293126583099365, "learning_rate": 0.001, "loss": 1.5598, "step": 164100 }, { "epoch": 53.0704589528119, "grad_norm": 1.3912168741226196, "learning_rate": 0.001, "loss": 1.5306, "step": 164200 }, { "epoch": 53.10277957336781, "grad_norm": 1.7945741415023804, "learning_rate": 0.001, "loss": 1.5636, "step": 164300 }, { "epoch": 53.135100193923726, "grad_norm": 1.0372518301010132, "learning_rate": 0.001, "loss": 1.5817, "step": 164400 }, { "epoch": 53.16742081447964, "grad_norm": 1.2016254663467407, "learning_rate": 0.001, "loss": 1.5833, "step": 164500 }, { "epoch": 53.199741435035556, "grad_norm": 1.2256823778152466, "learning_rate": 0.001, "loss": 1.5829, "step": 164600 }, { "epoch": 53.23206205559147, "grad_norm": 1.3637040853500366, "learning_rate": 0.001, "loss": 1.5824, "step": 164700 }, { "epoch": 53.264382676147385, "grad_norm": 1.452815055847168, "learning_rate": 0.001, "loss": 1.6008, "step": 164800 }, { "epoch": 53.2967032967033, "grad_norm": 1.546164870262146, "learning_rate": 0.001, "loss": 1.6041, "step": 164900 }, { "epoch": 53.329023917259214, "grad_norm": 1.9156973361968994, "learning_rate": 0.001, "loss": 1.6182, "step": 165000 }, { "epoch": 53.36134453781513, "grad_norm": 1.5630964040756226, "learning_rate": 0.001, "loss": 1.5986, "step": 165100 }, { "epoch": 53.39366515837104, "grad_norm": 1.3350998163223267, "learning_rate": 0.001, "loss": 1.6171, "step": 165200 }, { "epoch": 53.42598577892696, "grad_norm": 1.3054630756378174, "learning_rate": 0.001, "loss": 1.6153, "step": 165300 }, { "epoch": 53.45830639948287, "grad_norm": 1.5927234888076782, "learning_rate": 0.001, "loss": 1.6113, "step": 165400 }, { "epoch": 53.49062702003879, "grad_norm": 1.5710676908493042, "learning_rate": 0.001, "loss": 1.6116, "step": 165500 }, { "epoch": 53.5229476405947, "grad_norm": 1.4166501760482788, "learning_rate": 0.001, "loss": 1.6275, "step": 165600 }, { "epoch": 53.555268261150616, "grad_norm": 1.5470216274261475, "learning_rate": 0.001, "loss": 1.6301, "step": 165700 }, { "epoch": 53.58758888170653, "grad_norm": 1.11976158618927, "learning_rate": 0.001, "loss": 1.6067, "step": 165800 }, { "epoch": 53.619909502262445, "grad_norm": 1.5103334188461304, "learning_rate": 0.001, "loss": 1.6336, "step": 165900 }, { "epoch": 53.65223012281836, "grad_norm": 1.3977969884872437, "learning_rate": 0.001, "loss": 1.6305, "step": 166000 }, { "epoch": 53.684550743374274, "grad_norm": 1.0533747673034668, "learning_rate": 0.001, "loss": 1.6577, "step": 166100 }, { "epoch": 53.71687136393019, "grad_norm": 1.1209571361541748, "learning_rate": 0.001, "loss": 1.6495, "step": 166200 }, { "epoch": 53.7491919844861, "grad_norm": 1.325560212135315, "learning_rate": 0.001, "loss": 1.6387, "step": 166300 }, { "epoch": 53.78151260504202, "grad_norm": 1.669684648513794, "learning_rate": 0.001, "loss": 1.6652, "step": 166400 }, { "epoch": 53.81383322559793, "grad_norm": 1.2255173921585083, "learning_rate": 0.001, "loss": 1.6421, "step": 166500 }, { "epoch": 53.84615384615385, "grad_norm": 1.6551024913787842, "learning_rate": 0.001, "loss": 1.6467, "step": 166600 }, { "epoch": 53.87847446670976, "grad_norm": 1.3617397546768188, "learning_rate": 0.001, "loss": 1.6615, "step": 166700 }, { "epoch": 53.910795087265676, "grad_norm": 1.486262321472168, "learning_rate": 0.001, "loss": 1.6718, "step": 166800 }, { "epoch": 53.94311570782159, "grad_norm": 1.360690951347351, "learning_rate": 0.001, "loss": 1.667, "step": 166900 }, { "epoch": 53.975436328377505, "grad_norm": 1.2477593421936035, "learning_rate": 0.001, "loss": 1.6788, "step": 167000 }, { "epoch": 54.00775694893342, "grad_norm": 1.0727769136428833, "learning_rate": 0.001, "loss": 1.6627, "step": 167100 }, { "epoch": 54.040077569489334, "grad_norm": 1.4639390707015991, "learning_rate": 0.001, "loss": 1.552, "step": 167200 }, { "epoch": 54.07239819004525, "grad_norm": 1.1440781354904175, "learning_rate": 0.001, "loss": 1.542, "step": 167300 }, { "epoch": 54.10471881060116, "grad_norm": 1.5725555419921875, "learning_rate": 0.001, "loss": 1.5539, "step": 167400 }, { "epoch": 54.13703943115708, "grad_norm": 1.2718406915664673, "learning_rate": 0.001, "loss": 1.5499, "step": 167500 }, { "epoch": 54.16936005171299, "grad_norm": 1.3445727825164795, "learning_rate": 0.001, "loss": 1.5602, "step": 167600 }, { "epoch": 54.20168067226891, "grad_norm": 1.2247686386108398, "learning_rate": 0.001, "loss": 1.5604, "step": 167700 }, { "epoch": 54.23400129282482, "grad_norm": 1.6381856203079224, "learning_rate": 0.001, "loss": 1.5846, "step": 167800 }, { "epoch": 54.266321913380736, "grad_norm": 1.3378030061721802, "learning_rate": 0.001, "loss": 1.5908, "step": 167900 }, { "epoch": 54.29864253393665, "grad_norm": 1.147397756576538, "learning_rate": 0.001, "loss": 1.5852, "step": 168000 }, { "epoch": 54.330963154492565, "grad_norm": 1.2928037643432617, "learning_rate": 0.001, "loss": 1.586, "step": 168100 }, { "epoch": 54.36328377504848, "grad_norm": 1.2967069149017334, "learning_rate": 0.001, "loss": 1.6003, "step": 168200 }, { "epoch": 54.395604395604394, "grad_norm": 1.5993214845657349, "learning_rate": 0.001, "loss": 1.5974, "step": 168300 }, { "epoch": 54.42792501616031, "grad_norm": 1.49488365650177, "learning_rate": 0.001, "loss": 1.6045, "step": 168400 }, { "epoch": 54.46024563671622, "grad_norm": 1.2498141527175903, "learning_rate": 0.001, "loss": 1.61, "step": 168500 }, { "epoch": 54.49256625727214, "grad_norm": 1.6309314966201782, "learning_rate": 0.001, "loss": 1.6046, "step": 168600 }, { "epoch": 54.52488687782805, "grad_norm": 1.0654757022857666, "learning_rate": 0.001, "loss": 1.6091, "step": 168700 }, { "epoch": 54.55720749838397, "grad_norm": 1.4608960151672363, "learning_rate": 0.001, "loss": 1.6241, "step": 168800 }, { "epoch": 54.58952811893988, "grad_norm": 1.3533505201339722, "learning_rate": 0.001, "loss": 1.6278, "step": 168900 }, { "epoch": 54.621848739495796, "grad_norm": 1.649953007698059, "learning_rate": 0.001, "loss": 1.6226, "step": 169000 }, { "epoch": 54.65416936005171, "grad_norm": 1.1779111623764038, "learning_rate": 0.001, "loss": 1.6388, "step": 169100 }, { "epoch": 54.686489980607625, "grad_norm": 1.131826639175415, "learning_rate": 0.001, "loss": 1.6333, "step": 169200 }, { "epoch": 54.71881060116354, "grad_norm": 1.634190559387207, "learning_rate": 0.001, "loss": 1.6137, "step": 169300 }, { "epoch": 54.751131221719454, "grad_norm": 1.2887123823165894, "learning_rate": 0.001, "loss": 1.6448, "step": 169400 }, { "epoch": 54.78345184227537, "grad_norm": 1.4903887510299683, "learning_rate": 0.001, "loss": 1.6377, "step": 169500 }, { "epoch": 54.81577246283128, "grad_norm": 1.4088680744171143, "learning_rate": 0.001, "loss": 1.6475, "step": 169600 }, { "epoch": 54.8480930833872, "grad_norm": 1.3768340349197388, "learning_rate": 0.001, "loss": 1.6364, "step": 169700 }, { "epoch": 54.88041370394311, "grad_norm": 1.3599269390106201, "learning_rate": 0.001, "loss": 1.6346, "step": 169800 }, { "epoch": 54.91273432449903, "grad_norm": 1.323041319847107, "learning_rate": 0.001, "loss": 1.6463, "step": 169900 }, { "epoch": 54.94505494505494, "grad_norm": 1.4620647430419922, "learning_rate": 0.001, "loss": 1.6652, "step": 170000 }, { "epoch": 54.977375565610856, "grad_norm": 1.5375910997390747, "learning_rate": 0.001, "loss": 1.6611, "step": 170100 }, { "epoch": 55.00969618616678, "grad_norm": 1.4705251455307007, "learning_rate": 0.001, "loss": 1.6323, "step": 170200 }, { "epoch": 55.04201680672269, "grad_norm": 1.2176787853240967, "learning_rate": 0.001, "loss": 1.5366, "step": 170300 }, { "epoch": 55.07433742727861, "grad_norm": 1.4176909923553467, "learning_rate": 0.001, "loss": 1.5388, "step": 170400 }, { "epoch": 55.10665804783452, "grad_norm": 1.5232542753219604, "learning_rate": 0.001, "loss": 1.5412, "step": 170500 }, { "epoch": 55.138978668390436, "grad_norm": 1.2083262205123901, "learning_rate": 0.001, "loss": 1.5462, "step": 170600 }, { "epoch": 55.17129928894635, "grad_norm": 1.532065510749817, "learning_rate": 0.001, "loss": 1.5379, "step": 170700 }, { "epoch": 55.203619909502265, "grad_norm": 1.2907582521438599, "learning_rate": 0.001, "loss": 1.5585, "step": 170800 }, { "epoch": 55.23594053005818, "grad_norm": 1.2645903825759888, "learning_rate": 0.001, "loss": 1.5902, "step": 170900 }, { "epoch": 55.268261150614094, "grad_norm": 1.5214968919754028, "learning_rate": 0.001, "loss": 1.565, "step": 171000 }, { "epoch": 55.30058177117001, "grad_norm": 1.1648329496383667, "learning_rate": 0.001, "loss": 1.5745, "step": 171100 }, { "epoch": 55.33290239172592, "grad_norm": 1.636714220046997, "learning_rate": 0.001, "loss": 1.5743, "step": 171200 }, { "epoch": 55.36522301228184, "grad_norm": 1.2977919578552246, "learning_rate": 0.001, "loss": 1.5701, "step": 171300 }, { "epoch": 55.39754363283775, "grad_norm": 1.385388970375061, "learning_rate": 0.001, "loss": 1.6057, "step": 171400 }, { "epoch": 55.42986425339367, "grad_norm": 1.5424326658248901, "learning_rate": 0.001, "loss": 1.5935, "step": 171500 }, { "epoch": 55.46218487394958, "grad_norm": 1.3697704076766968, "learning_rate": 0.001, "loss": 1.5902, "step": 171600 }, { "epoch": 55.494505494505496, "grad_norm": 1.3550012111663818, "learning_rate": 0.001, "loss": 1.5964, "step": 171700 }, { "epoch": 55.52682611506141, "grad_norm": 1.5691142082214355, "learning_rate": 0.001, "loss": 1.5953, "step": 171800 }, { "epoch": 55.559146735617325, "grad_norm": 1.4401603937149048, "learning_rate": 0.001, "loss": 1.6091, "step": 171900 }, { "epoch": 55.59146735617324, "grad_norm": 2.070875406265259, "learning_rate": 0.001, "loss": 1.6154, "step": 172000 }, { "epoch": 55.623787976729155, "grad_norm": 1.2211717367172241, "learning_rate": 0.001, "loss": 1.6156, "step": 172100 }, { "epoch": 55.65610859728507, "grad_norm": 1.0735397338867188, "learning_rate": 0.001, "loss": 1.6031, "step": 172200 }, { "epoch": 55.688429217840984, "grad_norm": 1.4228463172912598, "learning_rate": 0.001, "loss": 1.6114, "step": 172300 }, { "epoch": 55.7207498383969, "grad_norm": 1.237438440322876, "learning_rate": 0.001, "loss": 1.6255, "step": 172400 }, { "epoch": 55.75307045895281, "grad_norm": 1.0559433698654175, "learning_rate": 0.001, "loss": 1.6164, "step": 172500 }, { "epoch": 55.78539107950873, "grad_norm": 1.2364671230316162, "learning_rate": 0.001, "loss": 1.6247, "step": 172600 }, { "epoch": 55.81771170006464, "grad_norm": 1.3936049938201904, "learning_rate": 0.001, "loss": 1.6272, "step": 172700 }, { "epoch": 55.85003232062056, "grad_norm": 1.3188313245773315, "learning_rate": 0.001, "loss": 1.6322, "step": 172800 }, { "epoch": 55.88235294117647, "grad_norm": 1.413771152496338, "learning_rate": 0.001, "loss": 1.6411, "step": 172900 }, { "epoch": 55.914673561732386, "grad_norm": 1.473656415939331, "learning_rate": 0.001, "loss": 1.6318, "step": 173000 }, { "epoch": 55.9469941822883, "grad_norm": 1.2637544870376587, "learning_rate": 0.001, "loss": 1.6574, "step": 173100 }, { "epoch": 55.979314802844215, "grad_norm": 1.431186556816101, "learning_rate": 0.001, "loss": 1.655, "step": 173200 }, { "epoch": 56.01163542340013, "grad_norm": 1.8388493061065674, "learning_rate": 0.001, "loss": 1.5949, "step": 173300 }, { "epoch": 56.043956043956044, "grad_norm": 1.6828988790512085, "learning_rate": 0.001, "loss": 1.5117, "step": 173400 }, { "epoch": 56.07627666451196, "grad_norm": 1.5223714113235474, "learning_rate": 0.001, "loss": 1.5248, "step": 173500 }, { "epoch": 56.10859728506787, "grad_norm": 1.3648293018341064, "learning_rate": 0.001, "loss": 1.5379, "step": 173600 }, { "epoch": 56.14091790562379, "grad_norm": 1.7287791967391968, "learning_rate": 0.001, "loss": 1.5393, "step": 173700 }, { "epoch": 56.1732385261797, "grad_norm": 1.3508548736572266, "learning_rate": 0.001, "loss": 1.5409, "step": 173800 }, { "epoch": 56.20555914673562, "grad_norm": 1.7643887996673584, "learning_rate": 0.001, "loss": 1.5358, "step": 173900 }, { "epoch": 56.23787976729153, "grad_norm": 1.4329142570495605, "learning_rate": 0.001, "loss": 1.5356, "step": 174000 }, { "epoch": 56.270200387847446, "grad_norm": 1.8588013648986816, "learning_rate": 0.001, "loss": 1.5515, "step": 174100 }, { "epoch": 56.30252100840336, "grad_norm": 1.16867995262146, "learning_rate": 0.001, "loss": 1.5672, "step": 174200 }, { "epoch": 56.334841628959275, "grad_norm": 1.4061293601989746, "learning_rate": 0.001, "loss": 1.5661, "step": 174300 }, { "epoch": 56.36716224951519, "grad_norm": 1.90892493724823, "learning_rate": 0.001, "loss": 1.5693, "step": 174400 }, { "epoch": 56.399482870071104, "grad_norm": 1.3699308633804321, "learning_rate": 0.001, "loss": 1.5647, "step": 174500 }, { "epoch": 56.43180349062702, "grad_norm": 1.3952831029891968, "learning_rate": 0.001, "loss": 1.5815, "step": 174600 }, { "epoch": 56.46412411118293, "grad_norm": 1.5254894495010376, "learning_rate": 0.001, "loss": 1.5699, "step": 174700 }, { "epoch": 56.49644473173885, "grad_norm": 1.2883787155151367, "learning_rate": 0.001, "loss": 1.5758, "step": 174800 }, { "epoch": 56.52876535229476, "grad_norm": 1.4908933639526367, "learning_rate": 0.001, "loss": 1.5834, "step": 174900 }, { "epoch": 56.56108597285068, "grad_norm": 1.6060596704483032, "learning_rate": 0.001, "loss": 1.5951, "step": 175000 }, { "epoch": 56.59340659340659, "grad_norm": 1.373229742050171, "learning_rate": 0.001, "loss": 1.6092, "step": 175100 }, { "epoch": 56.625727213962506, "grad_norm": 1.2752670049667358, "learning_rate": 0.001, "loss": 1.5959, "step": 175200 }, { "epoch": 56.65804783451842, "grad_norm": 1.3325690031051636, "learning_rate": 0.001, "loss": 1.6177, "step": 175300 }, { "epoch": 56.690368455074335, "grad_norm": 1.4257392883300781, "learning_rate": 0.001, "loss": 1.5941, "step": 175400 }, { "epoch": 56.72268907563025, "grad_norm": 1.7372586727142334, "learning_rate": 0.001, "loss": 1.6156, "step": 175500 }, { "epoch": 56.755009696186164, "grad_norm": 1.8692024946212769, "learning_rate": 0.001, "loss": 1.6122, "step": 175600 }, { "epoch": 56.78733031674208, "grad_norm": 1.1845149993896484, "learning_rate": 0.001, "loss": 1.6191, "step": 175700 }, { "epoch": 56.81965093729799, "grad_norm": 1.329891324043274, "learning_rate": 0.001, "loss": 1.6264, "step": 175800 }, { "epoch": 56.85197155785391, "grad_norm": 1.892138123512268, "learning_rate": 0.001, "loss": 1.6295, "step": 175900 }, { "epoch": 56.88429217840982, "grad_norm": 1.3890790939331055, "learning_rate": 0.001, "loss": 1.6303, "step": 176000 }, { "epoch": 56.91661279896574, "grad_norm": 1.2890719175338745, "learning_rate": 0.001, "loss": 1.6325, "step": 176100 }, { "epoch": 56.94893341952165, "grad_norm": 1.4877135753631592, "learning_rate": 0.001, "loss": 1.636, "step": 176200 }, { "epoch": 56.981254040077566, "grad_norm": 1.6091532707214355, "learning_rate": 0.001, "loss": 1.6331, "step": 176300 }, { "epoch": 57.01357466063349, "grad_norm": 1.7103350162506104, "learning_rate": 0.001, "loss": 1.5788, "step": 176400 }, { "epoch": 57.0458952811894, "grad_norm": 1.5880162715911865, "learning_rate": 0.001, "loss": 1.504, "step": 176500 }, { "epoch": 57.07821590174532, "grad_norm": 1.7704602479934692, "learning_rate": 0.001, "loss": 1.5129, "step": 176600 }, { "epoch": 57.11053652230123, "grad_norm": 1.5123844146728516, "learning_rate": 0.001, "loss": 1.5252, "step": 176700 }, { "epoch": 57.142857142857146, "grad_norm": 1.4449516534805298, "learning_rate": 0.001, "loss": 1.5276, "step": 176800 }, { "epoch": 57.17517776341306, "grad_norm": 1.8149229288101196, "learning_rate": 0.001, "loss": 1.5328, "step": 176900 }, { "epoch": 57.207498383968975, "grad_norm": 1.581376552581787, "learning_rate": 0.001, "loss": 1.5392, "step": 177000 }, { "epoch": 57.23981900452489, "grad_norm": 1.5949654579162598, "learning_rate": 0.001, "loss": 1.5372, "step": 177100 }, { "epoch": 57.272139625080804, "grad_norm": 1.690617561340332, "learning_rate": 0.001, "loss": 1.5487, "step": 177200 }, { "epoch": 57.30446024563672, "grad_norm": 1.688520073890686, "learning_rate": 0.001, "loss": 1.5432, "step": 177300 }, { "epoch": 57.33678086619263, "grad_norm": 1.3199012279510498, "learning_rate": 0.001, "loss": 1.558, "step": 177400 }, { "epoch": 57.36910148674855, "grad_norm": 2.138596773147583, "learning_rate": 0.001, "loss": 1.5618, "step": 177500 }, { "epoch": 57.40142210730446, "grad_norm": 1.5825165510177612, "learning_rate": 0.001, "loss": 1.5702, "step": 177600 }, { "epoch": 57.43374272786038, "grad_norm": 1.6623446941375732, "learning_rate": 0.001, "loss": 1.5728, "step": 177700 }, { "epoch": 57.46606334841629, "grad_norm": 1.442103385925293, "learning_rate": 0.001, "loss": 1.5674, "step": 177800 }, { "epoch": 57.498383968972206, "grad_norm": 1.7547607421875, "learning_rate": 0.001, "loss": 1.5792, "step": 177900 }, { "epoch": 57.53070458952812, "grad_norm": 1.3971937894821167, "learning_rate": 0.001, "loss": 1.5822, "step": 178000 }, { "epoch": 57.563025210084035, "grad_norm": 1.6914944648742676, "learning_rate": 0.001, "loss": 1.5839, "step": 178100 }, { "epoch": 57.59534583063995, "grad_norm": 1.3228996992111206, "learning_rate": 0.001, "loss": 1.5744, "step": 178200 }, { "epoch": 57.627666451195864, "grad_norm": 1.7055774927139282, "learning_rate": 0.001, "loss": 1.5888, "step": 178300 }, { "epoch": 57.65998707175178, "grad_norm": 1.8422220945358276, "learning_rate": 0.001, "loss": 1.5785, "step": 178400 }, { "epoch": 57.69230769230769, "grad_norm": 1.2623335123062134, "learning_rate": 0.001, "loss": 1.6017, "step": 178500 }, { "epoch": 57.72462831286361, "grad_norm": 1.4726150035858154, "learning_rate": 0.001, "loss": 1.5917, "step": 178600 }, { "epoch": 57.75694893341952, "grad_norm": 1.7706819772720337, "learning_rate": 0.001, "loss": 1.6056, "step": 178700 }, { "epoch": 57.78926955397544, "grad_norm": 1.4827563762664795, "learning_rate": 0.001, "loss": 1.6008, "step": 178800 }, { "epoch": 57.82159017453135, "grad_norm": 1.75459885597229, "learning_rate": 0.001, "loss": 1.6035, "step": 178900 }, { "epoch": 57.853910795087266, "grad_norm": 1.9852813482284546, "learning_rate": 0.001, "loss": 1.6028, "step": 179000 }, { "epoch": 57.88623141564318, "grad_norm": 1.5619901418685913, "learning_rate": 0.001, "loss": 1.6125, "step": 179100 }, { "epoch": 57.918552036199095, "grad_norm": 1.407097339630127, "learning_rate": 0.001, "loss": 1.6248, "step": 179200 }, { "epoch": 57.95087265675501, "grad_norm": 1.6684885025024414, "learning_rate": 0.001, "loss": 1.6413, "step": 179300 }, { "epoch": 57.983193277310924, "grad_norm": 1.371626377105713, "learning_rate": 0.001, "loss": 1.6387, "step": 179400 }, { "epoch": 58.01551389786684, "grad_norm": 1.6691104173660278, "learning_rate": 0.001, "loss": 1.5481, "step": 179500 }, { "epoch": 58.04783451842275, "grad_norm": 2.0170748233795166, "learning_rate": 0.001, "loss": 1.5036, "step": 179600 }, { "epoch": 58.08015513897867, "grad_norm": 2.1037421226501465, "learning_rate": 0.001, "loss": 1.5065, "step": 179700 }, { "epoch": 58.11247575953458, "grad_norm": 1.8298563957214355, "learning_rate": 0.001, "loss": 1.4945, "step": 179800 }, { "epoch": 58.1447963800905, "grad_norm": 2.10650634765625, "learning_rate": 0.001, "loss": 1.5188, "step": 179900 }, { "epoch": 58.17711700064641, "grad_norm": 2.177595853805542, "learning_rate": 0.001, "loss": 1.5035, "step": 180000 }, { "epoch": 58.209437621202326, "grad_norm": 1.913504958152771, "learning_rate": 0.001, "loss": 1.5228, "step": 180100 }, { "epoch": 58.24175824175824, "grad_norm": 1.8941245079040527, "learning_rate": 0.001, "loss": 1.5434, "step": 180200 }, { "epoch": 58.274078862314155, "grad_norm": 1.6749768257141113, "learning_rate": 0.001, "loss": 1.5297, "step": 180300 }, { "epoch": 58.30639948287007, "grad_norm": 1.793357491493225, "learning_rate": 0.001, "loss": 1.5384, "step": 180400 }, { "epoch": 58.338720103425985, "grad_norm": 2.0181198120117188, "learning_rate": 0.001, "loss": 1.5615, "step": 180500 }, { "epoch": 58.3710407239819, "grad_norm": 1.7327299118041992, "learning_rate": 0.001, "loss": 1.5527, "step": 180600 }, { "epoch": 58.403361344537814, "grad_norm": 2.45485782623291, "learning_rate": 0.001, "loss": 1.5408, "step": 180700 }, { "epoch": 58.43568196509373, "grad_norm": 1.6525640487670898, "learning_rate": 0.001, "loss": 1.5539, "step": 180800 }, { "epoch": 58.46800258564964, "grad_norm": 1.7305655479431152, "learning_rate": 0.001, "loss": 1.5571, "step": 180900 }, { "epoch": 58.50032320620556, "grad_norm": 2.0535521507263184, "learning_rate": 0.001, "loss": 1.5599, "step": 181000 }, { "epoch": 58.53264382676147, "grad_norm": 1.970602035522461, "learning_rate": 0.001, "loss": 1.5651, "step": 181100 }, { "epoch": 58.56496444731739, "grad_norm": 2.031547784805298, "learning_rate": 0.001, "loss": 1.562, "step": 181200 }, { "epoch": 58.5972850678733, "grad_norm": 1.9550069570541382, "learning_rate": 0.001, "loss": 1.5795, "step": 181300 }, { "epoch": 58.629605688429216, "grad_norm": 2.0826876163482666, "learning_rate": 0.001, "loss": 1.5786, "step": 181400 }, { "epoch": 58.66192630898513, "grad_norm": 2.283815860748291, "learning_rate": 0.001, "loss": 1.5809, "step": 181500 }, { "epoch": 58.694246929541045, "grad_norm": 2.006481170654297, "learning_rate": 0.001, "loss": 1.5804, "step": 181600 }, { "epoch": 58.72656755009696, "grad_norm": 1.988154411315918, "learning_rate": 0.001, "loss": 1.591, "step": 181700 }, { "epoch": 58.758888170652874, "grad_norm": 1.6653863191604614, "learning_rate": 0.001, "loss": 1.5923, "step": 181800 }, { "epoch": 58.79120879120879, "grad_norm": 1.7848979234695435, "learning_rate": 0.001, "loss": 1.6274, "step": 181900 }, { "epoch": 58.8235294117647, "grad_norm": 2.0813205242156982, "learning_rate": 0.001, "loss": 1.6043, "step": 182000 }, { "epoch": 58.85585003232062, "grad_norm": 2.4704935550689697, "learning_rate": 0.001, "loss": 1.5916, "step": 182100 }, { "epoch": 58.88817065287653, "grad_norm": 1.8839926719665527, "learning_rate": 0.001, "loss": 1.6067, "step": 182200 }, { "epoch": 58.92049127343245, "grad_norm": 2.0230321884155273, "learning_rate": 0.001, "loss": 1.6179, "step": 182300 }, { "epoch": 58.95281189398836, "grad_norm": 1.7509673833847046, "learning_rate": 0.001, "loss": 1.6122, "step": 182400 }, { "epoch": 58.985132514544276, "grad_norm": 1.4696741104125977, "learning_rate": 0.001, "loss": 1.6118, "step": 182500 }, { "epoch": 59.0174531351002, "grad_norm": 1.685705542564392, "learning_rate": 0.001, "loss": 1.5439, "step": 182600 }, { "epoch": 59.04977375565611, "grad_norm": 1.7140778303146362, "learning_rate": 0.001, "loss": 1.4896, "step": 182700 }, { "epoch": 59.08209437621203, "grad_norm": 1.9085273742675781, "learning_rate": 0.001, "loss": 1.4753, "step": 182800 }, { "epoch": 59.11441499676794, "grad_norm": 1.8200005292892456, "learning_rate": 0.001, "loss": 1.496, "step": 182900 }, { "epoch": 59.146735617323856, "grad_norm": 1.8019359111785889, "learning_rate": 0.001, "loss": 1.4856, "step": 183000 }, { "epoch": 59.17905623787977, "grad_norm": 1.8619072437286377, "learning_rate": 0.001, "loss": 1.5037, "step": 183100 }, { "epoch": 59.211376858435685, "grad_norm": 1.6231518983840942, "learning_rate": 0.001, "loss": 1.5317, "step": 183200 }, { "epoch": 59.2436974789916, "grad_norm": 1.4733821153640747, "learning_rate": 0.001, "loss": 1.524, "step": 183300 }, { "epoch": 59.276018099547514, "grad_norm": 1.8434406518936157, "learning_rate": 0.001, "loss": 1.5269, "step": 183400 }, { "epoch": 59.30833872010343, "grad_norm": 2.057749032974243, "learning_rate": 0.001, "loss": 1.5341, "step": 183500 }, { "epoch": 59.34065934065934, "grad_norm": 1.3642988204956055, "learning_rate": 0.001, "loss": 1.5506, "step": 183600 }, { "epoch": 59.37297996121526, "grad_norm": 1.9842780828475952, "learning_rate": 0.001, "loss": 1.5516, "step": 183700 }, { "epoch": 59.40530058177117, "grad_norm": 1.4964210987091064, "learning_rate": 0.001, "loss": 1.5571, "step": 183800 }, { "epoch": 59.43762120232709, "grad_norm": 1.7212563753128052, "learning_rate": 0.001, "loss": 1.5561, "step": 183900 }, { "epoch": 59.469941822883, "grad_norm": 1.554451823234558, "learning_rate": 0.001, "loss": 1.5543, "step": 184000 }, { "epoch": 59.502262443438916, "grad_norm": 1.716330647468567, "learning_rate": 0.001, "loss": 1.5572, "step": 184100 }, { "epoch": 59.53458306399483, "grad_norm": 1.8828051090240479, "learning_rate": 0.001, "loss": 1.5555, "step": 184200 }, { "epoch": 59.566903684550745, "grad_norm": 1.6634325981140137, "learning_rate": 0.001, "loss": 1.5683, "step": 184300 }, { "epoch": 59.59922430510666, "grad_norm": 1.8589503765106201, "learning_rate": 0.001, "loss": 1.5656, "step": 184400 }, { "epoch": 59.631544925662574, "grad_norm": 1.3951932191848755, "learning_rate": 0.001, "loss": 1.5538, "step": 184500 }, { "epoch": 59.66386554621849, "grad_norm": 1.4658318758010864, "learning_rate": 0.001, "loss": 1.5737, "step": 184600 }, { "epoch": 59.6961861667744, "grad_norm": 1.5063440799713135, "learning_rate": 0.001, "loss": 1.5675, "step": 184700 }, { "epoch": 59.72850678733032, "grad_norm": 1.8502857685089111, "learning_rate": 0.001, "loss": 1.5768, "step": 184800 }, { "epoch": 59.76082740788623, "grad_norm": 1.649667739868164, "learning_rate": 0.001, "loss": 1.5736, "step": 184900 }, { "epoch": 59.79314802844215, "grad_norm": 1.669318437576294, "learning_rate": 0.001, "loss": 1.5923, "step": 185000 }, { "epoch": 59.82546864899806, "grad_norm": 1.6056246757507324, "learning_rate": 0.001, "loss": 1.6, "step": 185100 }, { "epoch": 59.857789269553976, "grad_norm": 1.629918098449707, "learning_rate": 0.001, "loss": 1.5971, "step": 185200 }, { "epoch": 59.89010989010989, "grad_norm": 1.788417935371399, "learning_rate": 0.001, "loss": 1.5848, "step": 185300 }, { "epoch": 59.922430510665805, "grad_norm": 1.831823706626892, "learning_rate": 0.001, "loss": 1.5893, "step": 185400 }, { "epoch": 59.95475113122172, "grad_norm": 1.2721410989761353, "learning_rate": 0.001, "loss": 1.5994, "step": 185500 }, { "epoch": 59.987071751777634, "grad_norm": 1.7463867664337158, "learning_rate": 0.001, "loss": 1.6055, "step": 185600 }, { "epoch": 60.01939237233355, "grad_norm": 1.6742517948150635, "learning_rate": 0.001, "loss": 1.5455, "step": 185700 }, { "epoch": 60.05171299288946, "grad_norm": 1.5779138803482056, "learning_rate": 0.001, "loss": 1.4833, "step": 185800 }, { "epoch": 60.08403361344538, "grad_norm": 1.3153148889541626, "learning_rate": 0.001, "loss": 1.482, "step": 185900 }, { "epoch": 60.11635423400129, "grad_norm": 1.6626505851745605, "learning_rate": 0.001, "loss": 1.5004, "step": 186000 }, { "epoch": 60.14867485455721, "grad_norm": 1.7429184913635254, "learning_rate": 0.001, "loss": 1.4972, "step": 186100 }, { "epoch": 60.18099547511312, "grad_norm": 1.5026839971542358, "learning_rate": 0.001, "loss": 1.5169, "step": 186200 }, { "epoch": 60.213316095669036, "grad_norm": 1.4758402109146118, "learning_rate": 0.001, "loss": 1.511, "step": 186300 }, { "epoch": 60.24563671622495, "grad_norm": 1.134965419769287, "learning_rate": 0.001, "loss": 1.5019, "step": 186400 }, { "epoch": 60.277957336780865, "grad_norm": 1.6340572834014893, "learning_rate": 0.001, "loss": 1.5235, "step": 186500 }, { "epoch": 60.31027795733678, "grad_norm": 1.5940749645233154, "learning_rate": 0.001, "loss": 1.5112, "step": 186600 }, { "epoch": 60.342598577892694, "grad_norm": 1.5816295146942139, "learning_rate": 0.001, "loss": 1.5415, "step": 186700 }, { "epoch": 60.37491919844861, "grad_norm": 1.4606095552444458, "learning_rate": 0.001, "loss": 1.5149, "step": 186800 }, { "epoch": 60.40723981900452, "grad_norm": 1.4606409072875977, "learning_rate": 0.001, "loss": 1.5175, "step": 186900 }, { "epoch": 60.43956043956044, "grad_norm": 1.3545846939086914, "learning_rate": 0.001, "loss": 1.5255, "step": 187000 }, { "epoch": 60.47188106011635, "grad_norm": 1.2157231569290161, "learning_rate": 0.001, "loss": 1.5399, "step": 187100 }, { "epoch": 60.50420168067227, "grad_norm": 1.1669540405273438, "learning_rate": 0.001, "loss": 1.5439, "step": 187200 }, { "epoch": 60.53652230122818, "grad_norm": 1.5640186071395874, "learning_rate": 0.001, "loss": 1.5655, "step": 187300 }, { "epoch": 60.568842921784096, "grad_norm": 1.237733006477356, "learning_rate": 0.001, "loss": 1.5536, "step": 187400 }, { "epoch": 60.60116354234001, "grad_norm": 1.4821105003356934, "learning_rate": 0.001, "loss": 1.5481, "step": 187500 }, { "epoch": 60.633484162895925, "grad_norm": 1.6118173599243164, "learning_rate": 0.001, "loss": 1.5542, "step": 187600 }, { "epoch": 60.66580478345184, "grad_norm": 1.2756856679916382, "learning_rate": 0.001, "loss": 1.552, "step": 187700 }, { "epoch": 60.698125404007754, "grad_norm": 1.218289852142334, "learning_rate": 0.001, "loss": 1.5685, "step": 187800 }, { "epoch": 60.73044602456367, "grad_norm": 1.4309395551681519, "learning_rate": 0.001, "loss": 1.5637, "step": 187900 }, { "epoch": 60.762766645119584, "grad_norm": 1.3158926963806152, "learning_rate": 0.001, "loss": 1.5833, "step": 188000 }, { "epoch": 60.7950872656755, "grad_norm": 1.5380806922912598, "learning_rate": 0.001, "loss": 1.5753, "step": 188100 }, { "epoch": 60.82740788623141, "grad_norm": 1.1491740942001343, "learning_rate": 0.001, "loss": 1.5885, "step": 188200 }, { "epoch": 60.85972850678733, "grad_norm": 1.592486023902893, "learning_rate": 0.001, "loss": 1.5668, "step": 188300 }, { "epoch": 60.89204912734324, "grad_norm": 1.9322662353515625, "learning_rate": 0.001, "loss": 1.5971, "step": 188400 }, { "epoch": 60.924369747899156, "grad_norm": 1.2906183004379272, "learning_rate": 0.001, "loss": 1.5866, "step": 188500 }, { "epoch": 60.95669036845507, "grad_norm": 1.088273048400879, "learning_rate": 0.001, "loss": 1.5761, "step": 188600 }, { "epoch": 60.98901098901099, "grad_norm": 1.6351017951965332, "learning_rate": 0.001, "loss": 1.5897, "step": 188700 }, { "epoch": 61.02133160956691, "grad_norm": 1.1794394254684448, "learning_rate": 0.001, "loss": 1.5269, "step": 188800 }, { "epoch": 61.05365223012282, "grad_norm": 1.6122803688049316, "learning_rate": 0.001, "loss": 1.4612, "step": 188900 }, { "epoch": 61.085972850678736, "grad_norm": 1.1957712173461914, "learning_rate": 0.001, "loss": 1.479, "step": 189000 }, { "epoch": 61.11829347123465, "grad_norm": 1.351511836051941, "learning_rate": 0.001, "loss": 1.4906, "step": 189100 }, { "epoch": 61.150614091790565, "grad_norm": 1.3004039525985718, "learning_rate": 0.001, "loss": 1.4888, "step": 189200 }, { "epoch": 61.18293471234648, "grad_norm": 1.1758172512054443, "learning_rate": 0.001, "loss": 1.5036, "step": 189300 }, { "epoch": 61.215255332902395, "grad_norm": 1.198196291923523, "learning_rate": 0.001, "loss": 1.5142, "step": 189400 }, { "epoch": 61.24757595345831, "grad_norm": 1.2117112874984741, "learning_rate": 0.001, "loss": 1.4881, "step": 189500 }, { "epoch": 61.279896574014224, "grad_norm": 1.1988940238952637, "learning_rate": 0.001, "loss": 1.5021, "step": 189600 }, { "epoch": 61.31221719457014, "grad_norm": 1.9979110956192017, "learning_rate": 0.001, "loss": 1.5264, "step": 189700 }, { "epoch": 61.34453781512605, "grad_norm": 1.6691573858261108, "learning_rate": 0.001, "loss": 1.5069, "step": 189800 }, { "epoch": 61.37685843568197, "grad_norm": 1.5724232196807861, "learning_rate": 0.001, "loss": 1.5237, "step": 189900 }, { "epoch": 61.40917905623788, "grad_norm": 1.1864527463912964, "learning_rate": 0.001, "loss": 1.5269, "step": 190000 }, { "epoch": 61.441499676793796, "grad_norm": 1.6931548118591309, "learning_rate": 0.001, "loss": 1.5398, "step": 190100 }, { "epoch": 61.47382029734971, "grad_norm": 1.3704824447631836, "learning_rate": 0.001, "loss": 1.5117, "step": 190200 }, { "epoch": 61.506140917905626, "grad_norm": 1.9242581129074097, "learning_rate": 0.001, "loss": 1.5334, "step": 190300 }, { "epoch": 61.53846153846154, "grad_norm": 1.3759894371032715, "learning_rate": 0.001, "loss": 1.5466, "step": 190400 }, { "epoch": 61.570782159017455, "grad_norm": 1.9773072004318237, "learning_rate": 0.001, "loss": 1.5369, "step": 190500 }, { "epoch": 61.60310277957337, "grad_norm": 1.736612319946289, "learning_rate": 0.001, "loss": 1.5414, "step": 190600 }, { "epoch": 61.635423400129284, "grad_norm": 1.4435505867004395, "learning_rate": 0.001, "loss": 1.5401, "step": 190700 }, { "epoch": 61.6677440206852, "grad_norm": 1.3289904594421387, "learning_rate": 0.001, "loss": 1.5401, "step": 190800 }, { "epoch": 61.70006464124111, "grad_norm": 1.6186822652816772, "learning_rate": 0.001, "loss": 1.5489, "step": 190900 }, { "epoch": 61.73238526179703, "grad_norm": 1.3240885734558105, "learning_rate": 0.001, "loss": 1.5423, "step": 191000 }, { "epoch": 61.76470588235294, "grad_norm": 1.7645050287246704, "learning_rate": 0.001, "loss": 1.5673, "step": 191100 }, { "epoch": 61.79702650290886, "grad_norm": 1.4898263216018677, "learning_rate": 0.001, "loss": 1.5592, "step": 191200 }, { "epoch": 61.82934712346477, "grad_norm": 1.2853797674179077, "learning_rate": 0.001, "loss": 1.573, "step": 191300 }, { "epoch": 61.861667744020686, "grad_norm": 1.157740592956543, "learning_rate": 0.001, "loss": 1.5736, "step": 191400 }, { "epoch": 61.8939883645766, "grad_norm": 1.4297878742218018, "learning_rate": 0.001, "loss": 1.5674, "step": 191500 }, { "epoch": 61.926308985132515, "grad_norm": 1.212586760520935, "learning_rate": 0.001, "loss": 1.5619, "step": 191600 }, { "epoch": 61.95862960568843, "grad_norm": 1.6560161113739014, "learning_rate": 0.001, "loss": 1.5732, "step": 191700 }, { "epoch": 61.990950226244344, "grad_norm": 1.7148549556732178, "learning_rate": 0.001, "loss": 1.597, "step": 191800 }, { "epoch": 62.02327084680026, "grad_norm": 1.8989347219467163, "learning_rate": 0.001, "loss": 1.5026, "step": 191900 }, { "epoch": 62.05559146735617, "grad_norm": 1.1705671548843384, "learning_rate": 0.001, "loss": 1.4667, "step": 192000 }, { "epoch": 62.08791208791209, "grad_norm": 1.3613288402557373, "learning_rate": 0.001, "loss": 1.4673, "step": 192100 }, { "epoch": 62.120232708468, "grad_norm": 1.5595178604125977, "learning_rate": 0.001, "loss": 1.4802, "step": 192200 }, { "epoch": 62.15255332902392, "grad_norm": 1.4763859510421753, "learning_rate": 0.001, "loss": 1.4813, "step": 192300 }, { "epoch": 62.18487394957983, "grad_norm": 2.1364266872406006, "learning_rate": 0.001, "loss": 1.4854, "step": 192400 }, { "epoch": 62.217194570135746, "grad_norm": 1.6717660427093506, "learning_rate": 0.001, "loss": 1.4927, "step": 192500 }, { "epoch": 62.24951519069166, "grad_norm": 1.4224249124526978, "learning_rate": 0.001, "loss": 1.4813, "step": 192600 }, { "epoch": 62.281835811247575, "grad_norm": 1.2694761753082275, "learning_rate": 0.001, "loss": 1.4936, "step": 192700 }, { "epoch": 62.31415643180349, "grad_norm": 1.412690281867981, "learning_rate": 0.001, "loss": 1.4977, "step": 192800 }, { "epoch": 62.346477052359404, "grad_norm": 1.2019102573394775, "learning_rate": 0.001, "loss": 1.5026, "step": 192900 }, { "epoch": 62.37879767291532, "grad_norm": 1.1802592277526855, "learning_rate": 0.001, "loss": 1.5033, "step": 193000 }, { "epoch": 62.41111829347123, "grad_norm": 1.4369657039642334, "learning_rate": 0.001, "loss": 1.5047, "step": 193100 }, { "epoch": 62.44343891402715, "grad_norm": 2.012082576751709, "learning_rate": 0.001, "loss": 1.5172, "step": 193200 }, { "epoch": 62.47575953458306, "grad_norm": 1.4512064456939697, "learning_rate": 0.001, "loss": 1.531, "step": 193300 }, { "epoch": 62.50808015513898, "grad_norm": 1.473134160041809, "learning_rate": 0.001, "loss": 1.5123, "step": 193400 }, { "epoch": 62.54040077569489, "grad_norm": 1.535452961921692, "learning_rate": 0.001, "loss": 1.5278, "step": 193500 }, { "epoch": 62.572721396250806, "grad_norm": 1.4422191381454468, "learning_rate": 0.001, "loss": 1.5345, "step": 193600 }, { "epoch": 62.60504201680672, "grad_norm": 1.3806055784225464, "learning_rate": 0.001, "loss": 1.5424, "step": 193700 }, { "epoch": 62.637362637362635, "grad_norm": 1.1664576530456543, "learning_rate": 0.001, "loss": 1.5507, "step": 193800 }, { "epoch": 62.66968325791855, "grad_norm": 1.386029839515686, "learning_rate": 0.001, "loss": 1.5445, "step": 193900 }, { "epoch": 62.702003878474464, "grad_norm": 1.5522286891937256, "learning_rate": 0.001, "loss": 1.5355, "step": 194000 }, { "epoch": 62.73432449903038, "grad_norm": 1.3496739864349365, "learning_rate": 0.001, "loss": 1.5467, "step": 194100 }, { "epoch": 62.76664511958629, "grad_norm": 1.4825329780578613, "learning_rate": 0.001, "loss": 1.5519, "step": 194200 }, { "epoch": 62.79896574014221, "grad_norm": 1.8328793048858643, "learning_rate": 0.001, "loss": 1.5495, "step": 194300 }, { "epoch": 62.83128636069812, "grad_norm": 1.2069604396820068, "learning_rate": 0.001, "loss": 1.5528, "step": 194400 }, { "epoch": 62.86360698125404, "grad_norm": 1.3283594846725464, "learning_rate": 0.001, "loss": 1.5547, "step": 194500 }, { "epoch": 62.89592760180995, "grad_norm": 1.2219054698944092, "learning_rate": 0.001, "loss": 1.566, "step": 194600 }, { "epoch": 62.928248222365866, "grad_norm": 1.385223150253296, "learning_rate": 0.001, "loss": 1.5755, "step": 194700 }, { "epoch": 62.96056884292178, "grad_norm": 1.4032306671142578, "learning_rate": 0.001, "loss": 1.5737, "step": 194800 }, { "epoch": 62.992889463477695, "grad_norm": 1.1347252130508423, "learning_rate": 0.001, "loss": 1.5622, "step": 194900 }, { "epoch": 63.02521008403362, "grad_norm": 1.2946356534957886, "learning_rate": 0.001, "loss": 1.4943, "step": 195000 }, { "epoch": 63.05753070458953, "grad_norm": 1.4891058206558228, "learning_rate": 0.001, "loss": 1.4452, "step": 195100 }, { "epoch": 63.089851325145446, "grad_norm": 1.3868176937103271, "learning_rate": 0.001, "loss": 1.4695, "step": 195200 }, { "epoch": 63.12217194570136, "grad_norm": 1.577714443206787, "learning_rate": 0.001, "loss": 1.4669, "step": 195300 }, { "epoch": 63.154492566257275, "grad_norm": 1.3947536945343018, "learning_rate": 0.001, "loss": 1.4661, "step": 195400 }, { "epoch": 63.18681318681319, "grad_norm": 1.5214914083480835, "learning_rate": 0.001, "loss": 1.4743, "step": 195500 }, { "epoch": 63.219133807369104, "grad_norm": 1.5438205003738403, "learning_rate": 0.001, "loss": 1.4707, "step": 195600 }, { "epoch": 63.25145442792502, "grad_norm": 1.4677503108978271, "learning_rate": 0.001, "loss": 1.4798, "step": 195700 }, { "epoch": 63.28377504848093, "grad_norm": 1.7563285827636719, "learning_rate": 0.001, "loss": 1.4916, "step": 195800 }, { "epoch": 63.31609566903685, "grad_norm": 1.5936578512191772, "learning_rate": 0.001, "loss": 1.4947, "step": 195900 }, { "epoch": 63.34841628959276, "grad_norm": 2.0166072845458984, "learning_rate": 0.001, "loss": 1.482, "step": 196000 }, { "epoch": 63.38073691014868, "grad_norm": 2.286106824874878, "learning_rate": 0.001, "loss": 1.4874, "step": 196100 }, { "epoch": 63.41305753070459, "grad_norm": 1.1472430229187012, "learning_rate": 0.001, "loss": 1.5037, "step": 196200 }, { "epoch": 63.445378151260506, "grad_norm": 1.718253493309021, "learning_rate": 0.001, "loss": 1.5103, "step": 196300 }, { "epoch": 63.47769877181642, "grad_norm": 1.5824445486068726, "learning_rate": 0.001, "loss": 1.5, "step": 196400 }, { "epoch": 63.510019392372335, "grad_norm": 1.6907223463058472, "learning_rate": 0.001, "loss": 1.5095, "step": 196500 }, { "epoch": 63.54234001292825, "grad_norm": 1.6708906888961792, "learning_rate": 0.001, "loss": 1.5171, "step": 196600 }, { "epoch": 63.574660633484164, "grad_norm": 1.2911310195922852, "learning_rate": 0.001, "loss": 1.5061, "step": 196700 }, { "epoch": 63.60698125404008, "grad_norm": 1.8056236505508423, "learning_rate": 0.001, "loss": 1.5303, "step": 196800 }, { "epoch": 63.63930187459599, "grad_norm": 1.9205259084701538, "learning_rate": 0.001, "loss": 1.5448, "step": 196900 }, { "epoch": 63.67162249515191, "grad_norm": 1.5220394134521484, "learning_rate": 0.001, "loss": 1.5306, "step": 197000 }, { "epoch": 63.70394311570782, "grad_norm": 1.3217031955718994, "learning_rate": 0.001, "loss": 1.5296, "step": 197100 }, { "epoch": 63.73626373626374, "grad_norm": 1.3838924169540405, "learning_rate": 0.001, "loss": 1.5338, "step": 197200 }, { "epoch": 63.76858435681965, "grad_norm": 1.1662184000015259, "learning_rate": 0.001, "loss": 1.5573, "step": 197300 }, { "epoch": 63.800904977375566, "grad_norm": 1.1319751739501953, "learning_rate": 0.001, "loss": 1.5474, "step": 197400 }, { "epoch": 63.83322559793148, "grad_norm": 1.2447794675827026, "learning_rate": 0.001, "loss": 1.5355, "step": 197500 }, { "epoch": 63.865546218487395, "grad_norm": 1.6842137575149536, "learning_rate": 0.001, "loss": 1.5658, "step": 197600 }, { "epoch": 63.89786683904331, "grad_norm": 1.3559083938598633, "learning_rate": 0.001, "loss": 1.5476, "step": 197700 }, { "epoch": 63.930187459599225, "grad_norm": 1.6996843814849854, "learning_rate": 0.001, "loss": 1.553, "step": 197800 }, { "epoch": 63.96250808015514, "grad_norm": 1.1497136354446411, "learning_rate": 0.001, "loss": 1.5744, "step": 197900 }, { "epoch": 63.994828700711054, "grad_norm": 1.206533432006836, "learning_rate": 0.001, "loss": 1.5465, "step": 198000 }, { "epoch": 64.02714932126698, "grad_norm": 1.6470812559127808, "learning_rate": 0.001, "loss": 1.4777, "step": 198100 }, { "epoch": 64.05946994182288, "grad_norm": 1.5419559478759766, "learning_rate": 0.001, "loss": 1.4359, "step": 198200 }, { "epoch": 64.0917905623788, "grad_norm": 1.462594747543335, "learning_rate": 0.001, "loss": 1.4459, "step": 198300 }, { "epoch": 64.12411118293471, "grad_norm": 1.344820499420166, "learning_rate": 0.001, "loss": 1.4427, "step": 198400 }, { "epoch": 64.15643180349063, "grad_norm": 1.4749201536178589, "learning_rate": 0.001, "loss": 1.4529, "step": 198500 }, { "epoch": 64.18875242404654, "grad_norm": 1.2434271574020386, "learning_rate": 0.001, "loss": 1.4589, "step": 198600 }, { "epoch": 64.22107304460246, "grad_norm": 1.9539108276367188, "learning_rate": 0.001, "loss": 1.4697, "step": 198700 }, { "epoch": 64.25339366515837, "grad_norm": 1.7359395027160645, "learning_rate": 0.001, "loss": 1.4675, "step": 198800 }, { "epoch": 64.28571428571429, "grad_norm": 1.6121803522109985, "learning_rate": 0.001, "loss": 1.4893, "step": 198900 }, { "epoch": 64.3180349062702, "grad_norm": 1.3574376106262207, "learning_rate": 0.001, "loss": 1.4833, "step": 199000 }, { "epoch": 64.35035552682612, "grad_norm": 1.2601439952850342, "learning_rate": 0.001, "loss": 1.4871, "step": 199100 }, { "epoch": 64.38267614738203, "grad_norm": 1.9087001085281372, "learning_rate": 0.001, "loss": 1.4774, "step": 199200 }, { "epoch": 64.41499676793795, "grad_norm": 1.5292835235595703, "learning_rate": 0.001, "loss": 1.4792, "step": 199300 }, { "epoch": 64.44731738849386, "grad_norm": 1.6694928407669067, "learning_rate": 0.001, "loss": 1.4967, "step": 199400 }, { "epoch": 64.47963800904978, "grad_norm": 1.6116362810134888, "learning_rate": 0.001, "loss": 1.4972, "step": 199500 }, { "epoch": 64.51195862960569, "grad_norm": 1.9715216159820557, "learning_rate": 0.001, "loss": 1.5139, "step": 199600 }, { "epoch": 64.54427925016161, "grad_norm": 1.5410069227218628, "learning_rate": 0.001, "loss": 1.5203, "step": 199700 }, { "epoch": 64.57659987071752, "grad_norm": 1.3730778694152832, "learning_rate": 0.001, "loss": 1.5153, "step": 199800 }, { "epoch": 64.60892049127344, "grad_norm": 1.2727103233337402, "learning_rate": 0.001, "loss": 1.5312, "step": 199900 }, { "epoch": 64.64124111182934, "grad_norm": 1.5073052644729614, "learning_rate": 0.001, "loss": 1.5263, "step": 200000 }, { "epoch": 64.67356173238527, "grad_norm": 2.369584321975708, "learning_rate": 0.001, "loss": 1.5204, "step": 200100 }, { "epoch": 64.70588235294117, "grad_norm": 1.1945735216140747, "learning_rate": 0.001, "loss": 1.5412, "step": 200200 }, { "epoch": 64.7382029734971, "grad_norm": 1.331651210784912, "learning_rate": 0.001, "loss": 1.524, "step": 200300 }, { "epoch": 64.770523594053, "grad_norm": 1.285965919494629, "learning_rate": 0.001, "loss": 1.5177, "step": 200400 }, { "epoch": 64.80284421460892, "grad_norm": 1.4945372343063354, "learning_rate": 0.001, "loss": 1.5508, "step": 200500 }, { "epoch": 64.83516483516483, "grad_norm": 1.5643702745437622, "learning_rate": 0.001, "loss": 1.543, "step": 200600 }, { "epoch": 64.86748545572075, "grad_norm": 1.7975209951400757, "learning_rate": 0.001, "loss": 1.5458, "step": 200700 }, { "epoch": 64.89980607627666, "grad_norm": 1.5219054222106934, "learning_rate": 0.001, "loss": 1.5455, "step": 200800 }, { "epoch": 64.93212669683258, "grad_norm": 1.8230688571929932, "learning_rate": 0.001, "loss": 1.5479, "step": 200900 }, { "epoch": 64.96444731738849, "grad_norm": 1.3300844430923462, "learning_rate": 0.001, "loss": 1.5498, "step": 201000 }, { "epoch": 64.99676793794441, "grad_norm": 1.855148196220398, "learning_rate": 0.001, "loss": 1.5354, "step": 201100 }, { "epoch": 65.02908855850032, "grad_norm": 1.267386555671692, "learning_rate": 0.001, "loss": 1.4431, "step": 201200 }, { "epoch": 65.06140917905624, "grad_norm": 1.9030799865722656, "learning_rate": 0.001, "loss": 1.4201, "step": 201300 }, { "epoch": 65.09372979961215, "grad_norm": 1.768187165260315, "learning_rate": 0.001, "loss": 1.4284, "step": 201400 }, { "epoch": 65.12605042016807, "grad_norm": 1.4768826961517334, "learning_rate": 0.001, "loss": 1.4641, "step": 201500 }, { "epoch": 65.15837104072398, "grad_norm": 1.999047040939331, "learning_rate": 0.001, "loss": 1.4743, "step": 201600 }, { "epoch": 65.1906916612799, "grad_norm": 1.8538107872009277, "learning_rate": 0.001, "loss": 1.4551, "step": 201700 }, { "epoch": 65.2230122818358, "grad_norm": 1.830643892288208, "learning_rate": 0.001, "loss": 1.4559, "step": 201800 }, { "epoch": 65.25533290239173, "grad_norm": 1.6101281642913818, "learning_rate": 0.001, "loss": 1.4737, "step": 201900 }, { "epoch": 65.28765352294764, "grad_norm": 1.428889513015747, "learning_rate": 0.001, "loss": 1.4659, "step": 202000 }, { "epoch": 65.31997414350356, "grad_norm": 1.7600284814834595, "learning_rate": 0.001, "loss": 1.4773, "step": 202100 }, { "epoch": 65.35229476405947, "grad_norm": 1.3676204681396484, "learning_rate": 0.001, "loss": 1.4902, "step": 202200 }, { "epoch": 65.38461538461539, "grad_norm": 1.7085014581680298, "learning_rate": 0.001, "loss": 1.4743, "step": 202300 }, { "epoch": 65.4169360051713, "grad_norm": 1.8393412828445435, "learning_rate": 0.001, "loss": 1.4702, "step": 202400 }, { "epoch": 65.44925662572722, "grad_norm": 1.8859540224075317, "learning_rate": 0.001, "loss": 1.4855, "step": 202500 }, { "epoch": 65.48157724628312, "grad_norm": 1.6094419956207275, "learning_rate": 0.001, "loss": 1.4896, "step": 202600 }, { "epoch": 65.51389786683905, "grad_norm": 1.5427072048187256, "learning_rate": 0.001, "loss": 1.5094, "step": 202700 }, { "epoch": 65.54621848739495, "grad_norm": 1.7525871992111206, "learning_rate": 0.001, "loss": 1.502, "step": 202800 }, { "epoch": 65.57853910795087, "grad_norm": 1.5022687911987305, "learning_rate": 0.001, "loss": 1.5054, "step": 202900 }, { "epoch": 65.61085972850678, "grad_norm": 1.9936814308166504, "learning_rate": 0.001, "loss": 1.5191, "step": 203000 }, { "epoch": 65.6431803490627, "grad_norm": 1.6191157102584839, "learning_rate": 0.001, "loss": 1.5084, "step": 203100 }, { "epoch": 65.67550096961861, "grad_norm": 1.296644687652588, "learning_rate": 0.001, "loss": 1.5226, "step": 203200 }, { "epoch": 65.70782159017453, "grad_norm": 1.4507417678833008, "learning_rate": 0.001, "loss": 1.5152, "step": 203300 }, { "epoch": 65.74014221073044, "grad_norm": 1.6204619407653809, "learning_rate": 0.001, "loss": 1.5141, "step": 203400 }, { "epoch": 65.77246283128636, "grad_norm": 1.3483736515045166, "learning_rate": 0.001, "loss": 1.5228, "step": 203500 }, { "epoch": 65.80478345184227, "grad_norm": 1.2593847513198853, "learning_rate": 0.001, "loss": 1.5208, "step": 203600 }, { "epoch": 65.83710407239819, "grad_norm": 1.685653567314148, "learning_rate": 0.001, "loss": 1.5312, "step": 203700 }, { "epoch": 65.8694246929541, "grad_norm": 1.7216143608093262, "learning_rate": 0.001, "loss": 1.5274, "step": 203800 }, { "epoch": 65.90174531351002, "grad_norm": 2.0874483585357666, "learning_rate": 0.001, "loss": 1.5377, "step": 203900 }, { "epoch": 65.93406593406593, "grad_norm": 1.7108731269836426, "learning_rate": 0.001, "loss": 1.5434, "step": 204000 }, { "epoch": 65.96638655462185, "grad_norm": 1.8892107009887695, "learning_rate": 0.001, "loss": 1.5339, "step": 204100 }, { "epoch": 65.99870717517777, "grad_norm": 1.5250515937805176, "learning_rate": 0.001, "loss": 1.5228, "step": 204200 }, { "epoch": 66.03102779573368, "grad_norm": 2.090299129486084, "learning_rate": 0.001, "loss": 1.4188, "step": 204300 }, { "epoch": 66.0633484162896, "grad_norm": 1.6703615188598633, "learning_rate": 0.001, "loss": 1.4204, "step": 204400 }, { "epoch": 66.0956690368455, "grad_norm": 1.480782389640808, "learning_rate": 0.001, "loss": 1.4396, "step": 204500 }, { "epoch": 66.12798965740143, "grad_norm": 1.6650584936141968, "learning_rate": 0.001, "loss": 1.4409, "step": 204600 }, { "epoch": 66.16031027795734, "grad_norm": 2.058734178543091, "learning_rate": 0.001, "loss": 1.4552, "step": 204700 }, { "epoch": 66.19263089851326, "grad_norm": 1.3215047121047974, "learning_rate": 0.001, "loss": 1.4592, "step": 204800 }, { "epoch": 66.22495151906917, "grad_norm": 1.9270176887512207, "learning_rate": 0.001, "loss": 1.4427, "step": 204900 }, { "epoch": 66.25727213962509, "grad_norm": 1.6073015928268433, "learning_rate": 0.001, "loss": 1.4681, "step": 205000 }, { "epoch": 66.289592760181, "grad_norm": 1.4834781885147095, "learning_rate": 0.001, "loss": 1.4491, "step": 205100 }, { "epoch": 66.32191338073692, "grad_norm": 1.7395453453063965, "learning_rate": 0.001, "loss": 1.473, "step": 205200 }, { "epoch": 66.35423400129282, "grad_norm": 1.753955364227295, "learning_rate": 0.001, "loss": 1.4717, "step": 205300 }, { "epoch": 66.38655462184875, "grad_norm": 2.1493654251098633, "learning_rate": 0.001, "loss": 1.4602, "step": 205400 }, { "epoch": 66.41887524240465, "grad_norm": 1.885465145111084, "learning_rate": 0.001, "loss": 1.4788, "step": 205500 }, { "epoch": 66.45119586296057, "grad_norm": 1.9838215112686157, "learning_rate": 0.001, "loss": 1.492, "step": 205600 }, { "epoch": 66.48351648351648, "grad_norm": 1.5979394912719727, "learning_rate": 0.001, "loss": 1.4711, "step": 205700 }, { "epoch": 66.5158371040724, "grad_norm": 1.5088797807693481, "learning_rate": 0.001, "loss": 1.48, "step": 205800 }, { "epoch": 66.54815772462831, "grad_norm": 1.7536591291427612, "learning_rate": 0.001, "loss": 1.4945, "step": 205900 }, { "epoch": 66.58047834518423, "grad_norm": 1.5535506010055542, "learning_rate": 0.001, "loss": 1.4962, "step": 206000 }, { "epoch": 66.61279896574014, "grad_norm": 2.4775919914245605, "learning_rate": 0.001, "loss": 1.4974, "step": 206100 }, { "epoch": 66.64511958629606, "grad_norm": 2.5249624252319336, "learning_rate": 0.001, "loss": 1.4985, "step": 206200 }, { "epoch": 66.67744020685197, "grad_norm": 2.7047958374023438, "learning_rate": 0.001, "loss": 1.5138, "step": 206300 }, { "epoch": 66.70976082740789, "grad_norm": 2.077918291091919, "learning_rate": 0.001, "loss": 1.5047, "step": 206400 }, { "epoch": 66.7420814479638, "grad_norm": 1.75831937789917, "learning_rate": 0.001, "loss": 1.5122, "step": 206500 }, { "epoch": 66.77440206851972, "grad_norm": 1.9813868999481201, "learning_rate": 0.001, "loss": 1.5156, "step": 206600 }, { "epoch": 66.80672268907563, "grad_norm": 1.6824148893356323, "learning_rate": 0.001, "loss": 1.5052, "step": 206700 }, { "epoch": 66.83904330963155, "grad_norm": 1.811954379081726, "learning_rate": 0.001, "loss": 1.52, "step": 206800 }, { "epoch": 66.87136393018746, "grad_norm": 1.7890185117721558, "learning_rate": 0.001, "loss": 1.5331, "step": 206900 }, { "epoch": 66.90368455074338, "grad_norm": 1.6107674837112427, "learning_rate": 0.001, "loss": 1.5288, "step": 207000 }, { "epoch": 66.93600517129929, "grad_norm": 1.9740185737609863, "learning_rate": 0.001, "loss": 1.5201, "step": 207100 }, { "epoch": 66.96832579185521, "grad_norm": 1.9510257244110107, "learning_rate": 0.001, "loss": 1.5429, "step": 207200 }, { "epoch": 67.00064641241111, "grad_norm": 1.516554355621338, "learning_rate": 0.001, "loss": 1.516, "step": 207300 }, { "epoch": 67.03296703296704, "grad_norm": 1.8894391059875488, "learning_rate": 0.001, "loss": 1.4038, "step": 207400 }, { "epoch": 67.06528765352294, "grad_norm": 1.7522579431533813, "learning_rate": 0.001, "loss": 1.4134, "step": 207500 }, { "epoch": 67.09760827407887, "grad_norm": 2.453711748123169, "learning_rate": 0.001, "loss": 1.4311, "step": 207600 }, { "epoch": 67.12992889463477, "grad_norm": 1.8193609714508057, "learning_rate": 0.001, "loss": 1.4322, "step": 207700 }, { "epoch": 67.1622495151907, "grad_norm": 1.7121226787567139, "learning_rate": 0.001, "loss": 1.4251, "step": 207800 }, { "epoch": 67.1945701357466, "grad_norm": 2.3753931522369385, "learning_rate": 0.001, "loss": 1.4413, "step": 207900 }, { "epoch": 67.22689075630252, "grad_norm": 1.9751302003860474, "learning_rate": 0.001, "loss": 1.4579, "step": 208000 }, { "epoch": 67.25921137685843, "grad_norm": 1.4990696907043457, "learning_rate": 0.001, "loss": 1.4414, "step": 208100 }, { "epoch": 67.29153199741435, "grad_norm": 2.51631236076355, "learning_rate": 0.001, "loss": 1.4483, "step": 208200 }, { "epoch": 67.32385261797026, "grad_norm": 1.1820275783538818, "learning_rate": 0.001, "loss": 1.4559, "step": 208300 }, { "epoch": 67.35617323852618, "grad_norm": 1.8012865781784058, "learning_rate": 0.001, "loss": 1.4585, "step": 208400 }, { "epoch": 67.38849385908209, "grad_norm": 1.9256772994995117, "learning_rate": 0.001, "loss": 1.4568, "step": 208500 }, { "epoch": 67.42081447963801, "grad_norm": 2.4829864501953125, "learning_rate": 0.001, "loss": 1.4823, "step": 208600 }, { "epoch": 67.45313510019392, "grad_norm": 1.6891846656799316, "learning_rate": 0.001, "loss": 1.4668, "step": 208700 }, { "epoch": 67.48545572074984, "grad_norm": 2.690870523452759, "learning_rate": 0.001, "loss": 1.4826, "step": 208800 }, { "epoch": 67.51777634130575, "grad_norm": 2.1645565032958984, "learning_rate": 0.001, "loss": 1.4863, "step": 208900 }, { "epoch": 67.55009696186167, "grad_norm": 1.723306655883789, "learning_rate": 0.001, "loss": 1.4837, "step": 209000 }, { "epoch": 67.58241758241758, "grad_norm": 1.591162919998169, "learning_rate": 0.001, "loss": 1.4975, "step": 209100 }, { "epoch": 67.6147382029735, "grad_norm": 1.988744854927063, "learning_rate": 0.001, "loss": 1.4862, "step": 209200 }, { "epoch": 67.6470588235294, "grad_norm": 1.8148858547210693, "learning_rate": 0.001, "loss": 1.4787, "step": 209300 }, { "epoch": 67.67937944408533, "grad_norm": 1.5285876989364624, "learning_rate": 0.001, "loss": 1.4985, "step": 209400 }, { "epoch": 67.71170006464124, "grad_norm": 1.3206168413162231, "learning_rate": 0.001, "loss": 1.4879, "step": 209500 }, { "epoch": 67.74402068519716, "grad_norm": 1.4289054870605469, "learning_rate": 0.001, "loss": 1.4955, "step": 209600 }, { "epoch": 67.77634130575306, "grad_norm": 1.9192121028900146, "learning_rate": 0.001, "loss": 1.5137, "step": 209700 }, { "epoch": 67.80866192630899, "grad_norm": 2.1331663131713867, "learning_rate": 0.001, "loss": 1.4892, "step": 209800 }, { "epoch": 67.8409825468649, "grad_norm": 1.5964634418487549, "learning_rate": 0.001, "loss": 1.5037, "step": 209900 }, { "epoch": 67.87330316742081, "grad_norm": 2.2187821865081787, "learning_rate": 0.001, "loss": 1.5305, "step": 210000 }, { "epoch": 67.90562378797672, "grad_norm": 1.634127140045166, "learning_rate": 0.001, "loss": 1.5144, "step": 210100 }, { "epoch": 67.93794440853264, "grad_norm": 2.275919198989868, "learning_rate": 0.001, "loss": 1.5178, "step": 210200 }, { "epoch": 67.97026502908855, "grad_norm": 2.2488794326782227, "learning_rate": 0.001, "loss": 1.5312, "step": 210300 }, { "epoch": 68.00258564964447, "grad_norm": 1.4717329740524292, "learning_rate": 0.001, "loss": 1.5331, "step": 210400 }, { "epoch": 68.0349062702004, "grad_norm": 1.4725350141525269, "learning_rate": 0.001, "loss": 1.3944, "step": 210500 }, { "epoch": 68.0672268907563, "grad_norm": 1.347701907157898, "learning_rate": 0.001, "loss": 1.4202, "step": 210600 }, { "epoch": 68.09954751131222, "grad_norm": 1.8046602010726929, "learning_rate": 0.001, "loss": 1.4218, "step": 210700 }, { "epoch": 68.13186813186813, "grad_norm": 1.684139609336853, "learning_rate": 0.001, "loss": 1.4481, "step": 210800 }, { "epoch": 68.16418875242405, "grad_norm": 2.3283982276916504, "learning_rate": 0.001, "loss": 1.4293, "step": 210900 }, { "epoch": 68.19650937297996, "grad_norm": 1.7807434797286987, "learning_rate": 0.001, "loss": 1.4385, "step": 211000 }, { "epoch": 68.22882999353588, "grad_norm": 1.2383530139923096, "learning_rate": 0.001, "loss": 1.44, "step": 211100 }, { "epoch": 68.26115061409179, "grad_norm": 1.3734993934631348, "learning_rate": 0.001, "loss": 1.4455, "step": 211200 }, { "epoch": 68.29347123464771, "grad_norm": 1.8806383609771729, "learning_rate": 0.001, "loss": 1.4463, "step": 211300 }, { "epoch": 68.32579185520362, "grad_norm": 1.524986743927002, "learning_rate": 0.001, "loss": 1.4357, "step": 211400 }, { "epoch": 68.35811247575954, "grad_norm": 1.678237795829773, "learning_rate": 0.001, "loss": 1.4457, "step": 211500 }, { "epoch": 68.39043309631545, "grad_norm": 1.5592761039733887, "learning_rate": 0.001, "loss": 1.459, "step": 211600 }, { "epoch": 68.42275371687137, "grad_norm": 1.8970290422439575, "learning_rate": 0.001, "loss": 1.4517, "step": 211700 }, { "epoch": 68.45507433742728, "grad_norm": 1.3881733417510986, "learning_rate": 0.001, "loss": 1.4662, "step": 211800 }, { "epoch": 68.4873949579832, "grad_norm": 1.554467797279358, "learning_rate": 0.001, "loss": 1.4669, "step": 211900 }, { "epoch": 68.5197155785391, "grad_norm": 1.3873423337936401, "learning_rate": 0.001, "loss": 1.4684, "step": 212000 }, { "epoch": 68.55203619909503, "grad_norm": 1.2659697532653809, "learning_rate": 0.001, "loss": 1.4759, "step": 212100 }, { "epoch": 68.58435681965094, "grad_norm": 2.0712532997131348, "learning_rate": 0.001, "loss": 1.4665, "step": 212200 }, { "epoch": 68.61667744020686, "grad_norm": 1.8107807636260986, "learning_rate": 0.001, "loss": 1.481, "step": 212300 }, { "epoch": 68.64899806076276, "grad_norm": 1.7803953886032104, "learning_rate": 0.001, "loss": 1.4729, "step": 212400 }, { "epoch": 68.68131868131869, "grad_norm": 1.3027220964431763, "learning_rate": 0.001, "loss": 1.4872, "step": 212500 }, { "epoch": 68.7136393018746, "grad_norm": 1.3887300491333008, "learning_rate": 0.001, "loss": 1.4839, "step": 212600 }, { "epoch": 68.74595992243052, "grad_norm": 1.626133918762207, "learning_rate": 0.001, "loss": 1.493, "step": 212700 }, { "epoch": 68.77828054298642, "grad_norm": 1.6508300304412842, "learning_rate": 0.001, "loss": 1.4785, "step": 212800 }, { "epoch": 68.81060116354234, "grad_norm": 1.3312135934829712, "learning_rate": 0.001, "loss": 1.4954, "step": 212900 }, { "epoch": 68.84292178409825, "grad_norm": 2.1756434440612793, "learning_rate": 0.001, "loss": 1.5145, "step": 213000 }, { "epoch": 68.87524240465417, "grad_norm": 1.4839807748794556, "learning_rate": 0.001, "loss": 1.5196, "step": 213100 }, { "epoch": 68.90756302521008, "grad_norm": 1.1613637208938599, "learning_rate": 0.001, "loss": 1.5089, "step": 213200 }, { "epoch": 68.939883645766, "grad_norm": 1.2411142587661743, "learning_rate": 0.001, "loss": 1.5188, "step": 213300 }, { "epoch": 68.97220426632191, "grad_norm": 1.54549241065979, "learning_rate": 0.001, "loss": 1.5081, "step": 213400 }, { "epoch": 69.00452488687783, "grad_norm": 1.4730149507522583, "learning_rate": 0.001, "loss": 1.5034, "step": 213500 }, { "epoch": 69.03684550743374, "grad_norm": 1.7247742414474487, "learning_rate": 0.001, "loss": 1.4005, "step": 213600 }, { "epoch": 69.06916612798966, "grad_norm": 1.249778389930725, "learning_rate": 0.001, "loss": 1.4052, "step": 213700 }, { "epoch": 69.10148674854557, "grad_norm": 1.4992344379425049, "learning_rate": 0.001, "loss": 1.4217, "step": 213800 }, { "epoch": 69.13380736910149, "grad_norm": 1.7354190349578857, "learning_rate": 0.001, "loss": 1.417, "step": 213900 }, { "epoch": 69.1661279896574, "grad_norm": 1.257768154144287, "learning_rate": 0.001, "loss": 1.4199, "step": 214000 }, { "epoch": 69.19844861021332, "grad_norm": 1.6139253377914429, "learning_rate": 0.001, "loss": 1.4124, "step": 214100 }, { "epoch": 69.23076923076923, "grad_norm": 1.6175397634506226, "learning_rate": 0.001, "loss": 1.4277, "step": 214200 }, { "epoch": 69.26308985132515, "grad_norm": 1.3936841487884521, "learning_rate": 0.001, "loss": 1.4343, "step": 214300 }, { "epoch": 69.29541047188106, "grad_norm": 1.8072025775909424, "learning_rate": 0.001, "loss": 1.4298, "step": 214400 }, { "epoch": 69.32773109243698, "grad_norm": 1.3766324520111084, "learning_rate": 0.001, "loss": 1.4584, "step": 214500 }, { "epoch": 69.36005171299288, "grad_norm": 1.3047598600387573, "learning_rate": 0.001, "loss": 1.4385, "step": 214600 }, { "epoch": 69.3923723335488, "grad_norm": 1.836218237876892, "learning_rate": 0.001, "loss": 1.4313, "step": 214700 }, { "epoch": 69.42469295410471, "grad_norm": 1.3267902135849, "learning_rate": 0.001, "loss": 1.4397, "step": 214800 }, { "epoch": 69.45701357466064, "grad_norm": 1.6668914556503296, "learning_rate": 0.001, "loss": 1.454, "step": 214900 }, { "epoch": 69.48933419521654, "grad_norm": 1.2520886659622192, "learning_rate": 0.001, "loss": 1.4726, "step": 215000 }, { "epoch": 69.52165481577246, "grad_norm": 1.1094666719436646, "learning_rate": 0.001, "loss": 1.4574, "step": 215100 }, { "epoch": 69.55397543632837, "grad_norm": 1.2861254215240479, "learning_rate": 0.001, "loss": 1.4626, "step": 215200 }, { "epoch": 69.5862960568843, "grad_norm": 1.371955394744873, "learning_rate": 0.001, "loss": 1.4746, "step": 215300 }, { "epoch": 69.6186166774402, "grad_norm": 1.4750844240188599, "learning_rate": 0.001, "loss": 1.4771, "step": 215400 }, { "epoch": 69.65093729799612, "grad_norm": 1.984249472618103, "learning_rate": 0.001, "loss": 1.4874, "step": 215500 }, { "epoch": 69.68325791855203, "grad_norm": 1.4645411968231201, "learning_rate": 0.001, "loss": 1.4786, "step": 215600 }, { "epoch": 69.71557853910795, "grad_norm": 1.7948253154754639, "learning_rate": 0.001, "loss": 1.4895, "step": 215700 }, { "epoch": 69.74789915966386, "grad_norm": 1.4121185541152954, "learning_rate": 0.001, "loss": 1.482, "step": 215800 }, { "epoch": 69.78021978021978, "grad_norm": 2.0017154216766357, "learning_rate": 0.001, "loss": 1.4704, "step": 215900 }, { "epoch": 69.81254040077569, "grad_norm": 1.3346022367477417, "learning_rate": 0.001, "loss": 1.4875, "step": 216000 }, { "epoch": 69.84486102133161, "grad_norm": 1.9833992719650269, "learning_rate": 0.001, "loss": 1.4983, "step": 216100 }, { "epoch": 69.87718164188752, "grad_norm": 1.3632301092147827, "learning_rate": 0.001, "loss": 1.4786, "step": 216200 }, { "epoch": 69.90950226244344, "grad_norm": 1.5925140380859375, "learning_rate": 0.001, "loss": 1.4915, "step": 216300 }, { "epoch": 69.94182288299935, "grad_norm": 1.153775691986084, "learning_rate": 0.001, "loss": 1.5128, "step": 216400 }, { "epoch": 69.97414350355527, "grad_norm": 1.3893563747406006, "learning_rate": 0.001, "loss": 1.5076, "step": 216500 }, { "epoch": 70.00646412411119, "grad_norm": 1.3907309770584106, "learning_rate": 0.001, "loss": 1.4967, "step": 216600 }, { "epoch": 70.0387847446671, "grad_norm": 1.9319952726364136, "learning_rate": 0.001, "loss": 1.3789, "step": 216700 }, { "epoch": 70.07110536522302, "grad_norm": 0.9622918963432312, "learning_rate": 0.001, "loss": 1.3895, "step": 216800 }, { "epoch": 70.10342598577893, "grad_norm": 1.3357387781143188, "learning_rate": 0.001, "loss": 1.4076, "step": 216900 }, { "epoch": 70.13574660633485, "grad_norm": 1.3829296827316284, "learning_rate": 0.001, "loss": 1.4108, "step": 217000 }, { "epoch": 70.16806722689076, "grad_norm": 1.5978502035140991, "learning_rate": 0.001, "loss": 1.4024, "step": 217100 }, { "epoch": 70.20038784744668, "grad_norm": 1.342020034790039, "learning_rate": 0.001, "loss": 1.414, "step": 217200 }, { "epoch": 70.23270846800258, "grad_norm": 1.5343835353851318, "learning_rate": 0.001, "loss": 1.4224, "step": 217300 }, { "epoch": 70.2650290885585, "grad_norm": 1.2469536066055298, "learning_rate": 0.001, "loss": 1.428, "step": 217400 }, { "epoch": 70.29734970911441, "grad_norm": 1.3249125480651855, "learning_rate": 0.001, "loss": 1.4307, "step": 217500 }, { "epoch": 70.32967032967034, "grad_norm": 2.6667747497558594, "learning_rate": 0.001, "loss": 1.4408, "step": 217600 }, { "epoch": 70.36199095022624, "grad_norm": 1.1911907196044922, "learning_rate": 0.001, "loss": 1.4318, "step": 217700 }, { "epoch": 70.39431157078216, "grad_norm": 1.585929274559021, "learning_rate": 0.001, "loss": 1.4405, "step": 217800 }, { "epoch": 70.42663219133807, "grad_norm": 1.2287694215774536, "learning_rate": 0.001, "loss": 1.4416, "step": 217900 }, { "epoch": 70.458952811894, "grad_norm": 1.3678672313690186, "learning_rate": 0.001, "loss": 1.4487, "step": 218000 }, { "epoch": 70.4912734324499, "grad_norm": 1.9033615589141846, "learning_rate": 0.001, "loss": 1.4612, "step": 218100 }, { "epoch": 70.52359405300582, "grad_norm": 1.2729027271270752, "learning_rate": 0.001, "loss": 1.4598, "step": 218200 }, { "epoch": 70.55591467356173, "grad_norm": 1.6137685775756836, "learning_rate": 0.001, "loss": 1.4697, "step": 218300 }, { "epoch": 70.58823529411765, "grad_norm": 1.2115042209625244, "learning_rate": 0.001, "loss": 1.4754, "step": 218400 }, { "epoch": 70.62055591467356, "grad_norm": 1.4818336963653564, "learning_rate": 0.001, "loss": 1.4469, "step": 218500 }, { "epoch": 70.65287653522948, "grad_norm": 1.2290292978286743, "learning_rate": 0.001, "loss": 1.4621, "step": 218600 }, { "epoch": 70.68519715578539, "grad_norm": 1.5580615997314453, "learning_rate": 0.001, "loss": 1.4557, "step": 218700 }, { "epoch": 70.71751777634131, "grad_norm": 1.2235702276229858, "learning_rate": 0.001, "loss": 1.4739, "step": 218800 }, { "epoch": 70.74983839689722, "grad_norm": 1.3991035223007202, "learning_rate": 0.001, "loss": 1.4676, "step": 218900 }, { "epoch": 70.78215901745314, "grad_norm": 2.4721031188964844, "learning_rate": 0.001, "loss": 1.4849, "step": 219000 }, { "epoch": 70.81447963800905, "grad_norm": 1.246346354484558, "learning_rate": 0.001, "loss": 1.4824, "step": 219100 }, { "epoch": 70.84680025856497, "grad_norm": 1.8876065015792847, "learning_rate": 0.001, "loss": 1.5024, "step": 219200 }, { "epoch": 70.87912087912088, "grad_norm": 1.9916150569915771, "learning_rate": 0.001, "loss": 1.4854, "step": 219300 }, { "epoch": 70.9114414996768, "grad_norm": 1.6076717376708984, "learning_rate": 0.001, "loss": 1.4978, "step": 219400 }, { "epoch": 70.9437621202327, "grad_norm": 1.393532156944275, "learning_rate": 0.001, "loss": 1.4932, "step": 219500 }, { "epoch": 70.97608274078863, "grad_norm": 1.4620736837387085, "learning_rate": 0.001, "loss": 1.4862, "step": 219600 }, { "epoch": 71.00840336134453, "grad_norm": 1.4317773580551147, "learning_rate": 0.001, "loss": 1.486, "step": 219700 }, { "epoch": 71.04072398190046, "grad_norm": 1.4598522186279297, "learning_rate": 0.001, "loss": 1.3638, "step": 219800 }, { "epoch": 71.07304460245636, "grad_norm": 1.6084305047988892, "learning_rate": 0.001, "loss": 1.3949, "step": 219900 }, { "epoch": 71.10536522301229, "grad_norm": 1.2353745698928833, "learning_rate": 0.001, "loss": 1.3805, "step": 220000 }, { "epoch": 71.13768584356819, "grad_norm": 1.1433589458465576, "learning_rate": 0.001, "loss": 1.3986, "step": 220100 }, { "epoch": 71.17000646412411, "grad_norm": 1.6471236944198608, "learning_rate": 0.001, "loss": 1.3963, "step": 220200 }, { "epoch": 71.20232708468002, "grad_norm": 1.7072899341583252, "learning_rate": 0.001, "loss": 1.412, "step": 220300 }, { "epoch": 71.23464770523594, "grad_norm": 1.123854637145996, "learning_rate": 0.001, "loss": 1.422, "step": 220400 }, { "epoch": 71.26696832579185, "grad_norm": 1.6275721788406372, "learning_rate": 0.001, "loss": 1.3955, "step": 220500 }, { "epoch": 71.29928894634777, "grad_norm": 1.697174072265625, "learning_rate": 0.001, "loss": 1.4271, "step": 220600 }, { "epoch": 71.33160956690368, "grad_norm": 1.770520806312561, "learning_rate": 0.001, "loss": 1.4234, "step": 220700 }, { "epoch": 71.3639301874596, "grad_norm": 1.3543721437454224, "learning_rate": 0.001, "loss": 1.4203, "step": 220800 }, { "epoch": 71.39625080801551, "grad_norm": 1.134240984916687, "learning_rate": 0.001, "loss": 1.4453, "step": 220900 }, { "epoch": 71.42857142857143, "grad_norm": 1.407908320426941, "learning_rate": 0.001, "loss": 1.4446, "step": 221000 }, { "epoch": 71.46089204912734, "grad_norm": 2.0162112712860107, "learning_rate": 0.001, "loss": 1.4301, "step": 221100 }, { "epoch": 71.49321266968326, "grad_norm": 1.4100801944732666, "learning_rate": 0.001, "loss": 1.4479, "step": 221200 }, { "epoch": 71.52553329023917, "grad_norm": 1.562717318534851, "learning_rate": 0.001, "loss": 1.4426, "step": 221300 }, { "epoch": 71.55785391079509, "grad_norm": 1.5795971155166626, "learning_rate": 0.001, "loss": 1.4547, "step": 221400 }, { "epoch": 71.590174531351, "grad_norm": 1.2706962823867798, "learning_rate": 0.001, "loss": 1.4572, "step": 221500 }, { "epoch": 71.62249515190692, "grad_norm": 1.587480068206787, "learning_rate": 0.001, "loss": 1.4593, "step": 221600 }, { "epoch": 71.65481577246283, "grad_norm": 1.6591274738311768, "learning_rate": 0.001, "loss": 1.4629, "step": 221700 }, { "epoch": 71.68713639301875, "grad_norm": 1.419433355331421, "learning_rate": 0.001, "loss": 1.454, "step": 221800 }, { "epoch": 71.71945701357465, "grad_norm": 1.8765602111816406, "learning_rate": 0.001, "loss": 1.4504, "step": 221900 }, { "epoch": 71.75177763413058, "grad_norm": 1.2230541706085205, "learning_rate": 0.001, "loss": 1.4643, "step": 222000 }, { "epoch": 71.78409825468648, "grad_norm": 1.6914892196655273, "learning_rate": 0.001, "loss": 1.4887, "step": 222100 }, { "epoch": 71.8164188752424, "grad_norm": 1.3152620792388916, "learning_rate": 0.001, "loss": 1.4798, "step": 222200 }, { "epoch": 71.84873949579831, "grad_norm": 1.575087070465088, "learning_rate": 0.001, "loss": 1.4931, "step": 222300 }, { "epoch": 71.88106011635423, "grad_norm": 1.1092244386672974, "learning_rate": 0.001, "loss": 1.4747, "step": 222400 }, { "epoch": 71.91338073691014, "grad_norm": 2.302403450012207, "learning_rate": 0.001, "loss": 1.4785, "step": 222500 }, { "epoch": 71.94570135746606, "grad_norm": 1.587856650352478, "learning_rate": 0.001, "loss": 1.4847, "step": 222600 }, { "epoch": 71.97802197802197, "grad_norm": 1.3023202419281006, "learning_rate": 0.001, "loss": 1.5005, "step": 222700 }, { "epoch": 72.01034259857789, "grad_norm": 1.4531750679016113, "learning_rate": 0.001, "loss": 1.4403, "step": 222800 }, { "epoch": 72.04266321913381, "grad_norm": 1.6974835395812988, "learning_rate": 0.001, "loss": 1.3789, "step": 222900 }, { "epoch": 72.07498383968972, "grad_norm": 2.306978702545166, "learning_rate": 0.001, "loss": 1.3747, "step": 223000 }, { "epoch": 72.10730446024564, "grad_norm": 1.7029024362564087, "learning_rate": 0.001, "loss": 1.3762, "step": 223100 }, { "epoch": 72.13962508080155, "grad_norm": 1.4630860090255737, "learning_rate": 0.001, "loss": 1.3893, "step": 223200 }, { "epoch": 72.17194570135747, "grad_norm": 1.477061152458191, "learning_rate": 0.001, "loss": 1.3885, "step": 223300 }, { "epoch": 72.20426632191338, "grad_norm": 1.8601970672607422, "learning_rate": 0.001, "loss": 1.4036, "step": 223400 }, { "epoch": 72.2365869424693, "grad_norm": 1.5833276510238647, "learning_rate": 0.001, "loss": 1.4037, "step": 223500 }, { "epoch": 72.26890756302521, "grad_norm": 1.3761273622512817, "learning_rate": 0.001, "loss": 1.4053, "step": 223600 }, { "epoch": 72.30122818358113, "grad_norm": 1.475314736366272, "learning_rate": 0.001, "loss": 1.4223, "step": 223700 }, { "epoch": 72.33354880413704, "grad_norm": 1.369614601135254, "learning_rate": 0.001, "loss": 1.405, "step": 223800 }, { "epoch": 72.36586942469296, "grad_norm": 1.848175287246704, "learning_rate": 0.001, "loss": 1.425, "step": 223900 }, { "epoch": 72.39819004524887, "grad_norm": 1.6656006574630737, "learning_rate": 0.001, "loss": 1.424, "step": 224000 }, { "epoch": 72.43051066580479, "grad_norm": 1.6123826503753662, "learning_rate": 0.001, "loss": 1.4376, "step": 224100 }, { "epoch": 72.4628312863607, "grad_norm": 1.6487890481948853, "learning_rate": 0.001, "loss": 1.4475, "step": 224200 }, { "epoch": 72.49515190691662, "grad_norm": 1.5751174688339233, "learning_rate": 0.001, "loss": 1.4366, "step": 224300 }, { "epoch": 72.52747252747253, "grad_norm": 1.2488092184066772, "learning_rate": 0.001, "loss": 1.4392, "step": 224400 }, { "epoch": 72.55979314802845, "grad_norm": 1.2981414794921875, "learning_rate": 0.001, "loss": 1.4297, "step": 224500 }, { "epoch": 72.59211376858435, "grad_norm": 2.0852699279785156, "learning_rate": 0.001, "loss": 1.4447, "step": 224600 }, { "epoch": 72.62443438914028, "grad_norm": 1.5971038341522217, "learning_rate": 0.001, "loss": 1.4467, "step": 224700 }, { "epoch": 72.65675500969618, "grad_norm": 2.1098439693450928, "learning_rate": 0.001, "loss": 1.451, "step": 224800 }, { "epoch": 72.6890756302521, "grad_norm": 2.0493147373199463, "learning_rate": 0.001, "loss": 1.4557, "step": 224900 }, { "epoch": 72.72139625080801, "grad_norm": 1.8858246803283691, "learning_rate": 0.001, "loss": 1.4635, "step": 225000 }, { "epoch": 72.75371687136393, "grad_norm": 1.9451382160186768, "learning_rate": 0.001, "loss": 1.4601, "step": 225100 }, { "epoch": 72.78603749191984, "grad_norm": 1.993582844734192, "learning_rate": 0.001, "loss": 1.4711, "step": 225200 }, { "epoch": 72.81835811247576, "grad_norm": 1.8193069696426392, "learning_rate": 0.001, "loss": 1.481, "step": 225300 }, { "epoch": 72.85067873303167, "grad_norm": 1.814009189605713, "learning_rate": 0.001, "loss": 1.4684, "step": 225400 }, { "epoch": 72.88299935358759, "grad_norm": 1.269718885421753, "learning_rate": 0.001, "loss": 1.4661, "step": 225500 }, { "epoch": 72.9153199741435, "grad_norm": 1.751963496208191, "learning_rate": 0.001, "loss": 1.4599, "step": 225600 }, { "epoch": 72.94764059469942, "grad_norm": 1.4845296144485474, "learning_rate": 0.001, "loss": 1.48, "step": 225700 }, { "epoch": 72.97996121525533, "grad_norm": 1.3906903266906738, "learning_rate": 0.001, "loss": 1.492, "step": 225800 }, { "epoch": 73.01228183581125, "grad_norm": 1.5307003259658813, "learning_rate": 0.001, "loss": 1.4285, "step": 225900 }, { "epoch": 73.04460245636716, "grad_norm": 1.7780219316482544, "learning_rate": 0.001, "loss": 1.3561, "step": 226000 }, { "epoch": 73.07692307692308, "grad_norm": 2.0099196434020996, "learning_rate": 0.001, "loss": 1.3792, "step": 226100 }, { "epoch": 73.10924369747899, "grad_norm": 1.7311296463012695, "learning_rate": 0.001, "loss": 1.3763, "step": 226200 }, { "epoch": 73.14156431803491, "grad_norm": 1.5329862833023071, "learning_rate": 0.001, "loss": 1.3805, "step": 226300 }, { "epoch": 73.17388493859082, "grad_norm": 1.6151118278503418, "learning_rate": 0.001, "loss": 1.3825, "step": 226400 }, { "epoch": 73.20620555914674, "grad_norm": 2.4479856491088867, "learning_rate": 0.001, "loss": 1.3989, "step": 226500 }, { "epoch": 73.23852617970265, "grad_norm": 1.8405131101608276, "learning_rate": 0.001, "loss": 1.3833, "step": 226600 }, { "epoch": 73.27084680025857, "grad_norm": 1.5066527128219604, "learning_rate": 0.001, "loss": 1.3924, "step": 226700 }, { "epoch": 73.30316742081448, "grad_norm": 1.6452630758285522, "learning_rate": 0.001, "loss": 1.4048, "step": 226800 }, { "epoch": 73.3354880413704, "grad_norm": 1.8133480548858643, "learning_rate": 0.001, "loss": 1.4033, "step": 226900 }, { "epoch": 73.3678086619263, "grad_norm": 1.680242657661438, "learning_rate": 0.001, "loss": 1.4081, "step": 227000 }, { "epoch": 73.40012928248223, "grad_norm": 1.4392516613006592, "learning_rate": 0.001, "loss": 1.4146, "step": 227100 }, { "epoch": 73.43244990303813, "grad_norm": 1.518934965133667, "learning_rate": 0.001, "loss": 1.419, "step": 227200 }, { "epoch": 73.46477052359405, "grad_norm": 1.6375625133514404, "learning_rate": 0.001, "loss": 1.4169, "step": 227300 }, { "epoch": 73.49709114414996, "grad_norm": 1.5314610004425049, "learning_rate": 0.001, "loss": 1.4245, "step": 227400 }, { "epoch": 73.52941176470588, "grad_norm": 1.4845598936080933, "learning_rate": 0.001, "loss": 1.4381, "step": 227500 }, { "epoch": 73.56173238526179, "grad_norm": 1.3921239376068115, "learning_rate": 0.001, "loss": 1.4488, "step": 227600 }, { "epoch": 73.59405300581771, "grad_norm": 1.6126028299331665, "learning_rate": 0.001, "loss": 1.4386, "step": 227700 }, { "epoch": 73.62637362637362, "grad_norm": 1.7768104076385498, "learning_rate": 0.001, "loss": 1.4442, "step": 227800 }, { "epoch": 73.65869424692954, "grad_norm": 2.0095489025115967, "learning_rate": 0.001, "loss": 1.4483, "step": 227900 }, { "epoch": 73.69101486748545, "grad_norm": 1.9532029628753662, "learning_rate": 0.001, "loss": 1.446, "step": 228000 }, { "epoch": 73.72333548804137, "grad_norm": 1.411705732345581, "learning_rate": 0.001, "loss": 1.452, "step": 228100 }, { "epoch": 73.75565610859728, "grad_norm": 1.5540069341659546, "learning_rate": 0.001, "loss": 1.4569, "step": 228200 }, { "epoch": 73.7879767291532, "grad_norm": 1.84498929977417, "learning_rate": 0.001, "loss": 1.4545, "step": 228300 }, { "epoch": 73.82029734970911, "grad_norm": 1.4782354831695557, "learning_rate": 0.001, "loss": 1.4609, "step": 228400 }, { "epoch": 73.85261797026503, "grad_norm": 1.464579701423645, "learning_rate": 0.001, "loss": 1.4645, "step": 228500 }, { "epoch": 73.88493859082094, "grad_norm": 1.541219711303711, "learning_rate": 0.001, "loss": 1.467, "step": 228600 }, { "epoch": 73.91725921137686, "grad_norm": 1.4827616214752197, "learning_rate": 0.001, "loss": 1.4673, "step": 228700 }, { "epoch": 73.94957983193277, "grad_norm": 1.661637783050537, "learning_rate": 0.001, "loss": 1.4626, "step": 228800 }, { "epoch": 73.98190045248869, "grad_norm": 1.7068231105804443, "learning_rate": 0.001, "loss": 1.4766, "step": 228900 }, { "epoch": 74.01422107304461, "grad_norm": 2.049675226211548, "learning_rate": 0.001, "loss": 1.3957, "step": 229000 }, { "epoch": 74.04654169360052, "grad_norm": 1.8042556047439575, "learning_rate": 0.001, "loss": 1.35, "step": 229100 }, { "epoch": 74.07886231415644, "grad_norm": 1.642234206199646, "learning_rate": 0.001, "loss": 1.3546, "step": 229200 }, { "epoch": 74.11118293471235, "grad_norm": 1.505800485610962, "learning_rate": 0.001, "loss": 1.3772, "step": 229300 }, { "epoch": 74.14350355526827, "grad_norm": 1.6339256763458252, "learning_rate": 0.001, "loss": 1.3674, "step": 229400 }, { "epoch": 74.17582417582418, "grad_norm": 1.936946988105774, "learning_rate": 0.001, "loss": 1.3837, "step": 229500 }, { "epoch": 74.2081447963801, "grad_norm": 1.9458225965499878, "learning_rate": 0.001, "loss": 1.3847, "step": 229600 }, { "epoch": 74.240465416936, "grad_norm": 1.7668321132659912, "learning_rate": 0.001, "loss": 1.3762, "step": 229700 }, { "epoch": 74.27278603749193, "grad_norm": 2.21065354347229, "learning_rate": 0.001, "loss": 1.3969, "step": 229800 }, { "epoch": 74.30510665804783, "grad_norm": 2.361572504043579, "learning_rate": 0.001, "loss": 1.3836, "step": 229900 }, { "epoch": 74.33742727860376, "grad_norm": 2.1231894493103027, "learning_rate": 0.001, "loss": 1.4046, "step": 230000 }, { "epoch": 74.36974789915966, "grad_norm": 2.058026075363159, "learning_rate": 0.001, "loss": 1.3956, "step": 230100 }, { "epoch": 74.40206851971558, "grad_norm": 1.578091025352478, "learning_rate": 0.001, "loss": 1.4145, "step": 230200 }, { "epoch": 74.43438914027149, "grad_norm": 2.1628458499908447, "learning_rate": 0.001, "loss": 1.4188, "step": 230300 }, { "epoch": 74.46670976082741, "grad_norm": 1.9572947025299072, "learning_rate": 0.001, "loss": 1.4106, "step": 230400 }, { "epoch": 74.49903038138332, "grad_norm": 1.8140960931777954, "learning_rate": 0.001, "loss": 1.4309, "step": 230500 }, { "epoch": 74.53135100193924, "grad_norm": 2.127092123031616, "learning_rate": 0.001, "loss": 1.4259, "step": 230600 }, { "epoch": 74.56367162249515, "grad_norm": 1.8602912425994873, "learning_rate": 0.001, "loss": 1.4266, "step": 230700 }, { "epoch": 74.59599224305107, "grad_norm": 2.0972392559051514, "learning_rate": 0.001, "loss": 1.4374, "step": 230800 }, { "epoch": 74.62831286360698, "grad_norm": 2.137838125228882, "learning_rate": 0.001, "loss": 1.4413, "step": 230900 }, { "epoch": 74.6606334841629, "grad_norm": 1.6585432291030884, "learning_rate": 0.001, "loss": 1.4344, "step": 231000 }, { "epoch": 74.69295410471881, "grad_norm": 1.632261037826538, "learning_rate": 0.001, "loss": 1.4372, "step": 231100 }, { "epoch": 74.72527472527473, "grad_norm": 1.964889645576477, "learning_rate": 0.001, "loss": 1.4358, "step": 231200 }, { "epoch": 74.75759534583064, "grad_norm": 1.7561458349227905, "learning_rate": 0.001, "loss": 1.4463, "step": 231300 }, { "epoch": 74.78991596638656, "grad_norm": 1.3315393924713135, "learning_rate": 0.001, "loss": 1.4494, "step": 231400 }, { "epoch": 74.82223658694247, "grad_norm": 1.650225281715393, "learning_rate": 0.001, "loss": 1.4558, "step": 231500 }, { "epoch": 74.85455720749839, "grad_norm": 2.464940309524536, "learning_rate": 0.001, "loss": 1.4441, "step": 231600 }, { "epoch": 74.8868778280543, "grad_norm": 2.1637585163116455, "learning_rate": 0.001, "loss": 1.4602, "step": 231700 }, { "epoch": 74.91919844861022, "grad_norm": 1.736090064048767, "learning_rate": 0.001, "loss": 1.4664, "step": 231800 }, { "epoch": 74.95151906916612, "grad_norm": 1.712194800376892, "learning_rate": 0.001, "loss": 1.463, "step": 231900 }, { "epoch": 74.98383968972205, "grad_norm": 1.8816744089126587, "learning_rate": 0.001, "loss": 1.4736, "step": 232000 }, { "epoch": 75.01616031027795, "grad_norm": 1.9753756523132324, "learning_rate": 0.001, "loss": 1.3708, "step": 232100 }, { "epoch": 75.04848093083388, "grad_norm": 1.853826642036438, "learning_rate": 0.001, "loss": 1.355, "step": 232200 }, { "epoch": 75.08080155138978, "grad_norm": 2.6528196334838867, "learning_rate": 0.001, "loss": 1.3362, "step": 232300 }, { "epoch": 75.1131221719457, "grad_norm": 2.437607526779175, "learning_rate": 0.001, "loss": 1.3671, "step": 232400 }, { "epoch": 75.14544279250161, "grad_norm": 3.34757924079895, "learning_rate": 0.001, "loss": 1.379, "step": 232500 }, { "epoch": 75.17776341305753, "grad_norm": 2.346930742263794, "learning_rate": 0.001, "loss": 1.3635, "step": 232600 }, { "epoch": 75.21008403361344, "grad_norm": 1.8064361810684204, "learning_rate": 0.001, "loss": 1.3669, "step": 232700 }, { "epoch": 75.24240465416936, "grad_norm": 2.354738235473633, "learning_rate": 0.001, "loss": 1.3849, "step": 232800 }, { "epoch": 75.27472527472527, "grad_norm": 1.9407877922058105, "learning_rate": 0.001, "loss": 1.4039, "step": 232900 }, { "epoch": 75.30704589528119, "grad_norm": 3.2403366565704346, "learning_rate": 0.001, "loss": 1.3963, "step": 233000 }, { "epoch": 75.3393665158371, "grad_norm": 2.8879992961883545, "learning_rate": 0.001, "loss": 1.3914, "step": 233100 }, { "epoch": 75.37168713639302, "grad_norm": 2.049025774002075, "learning_rate": 0.001, "loss": 1.3899, "step": 233200 }, { "epoch": 75.40400775694893, "grad_norm": 2.4537274837493896, "learning_rate": 0.001, "loss": 1.4024, "step": 233300 }, { "epoch": 75.43632837750485, "grad_norm": 1.6752196550369263, "learning_rate": 0.001, "loss": 1.4074, "step": 233400 }, { "epoch": 75.46864899806076, "grad_norm": 1.8149118423461914, "learning_rate": 0.001, "loss": 1.4232, "step": 233500 }, { "epoch": 75.50096961861668, "grad_norm": 2.647524118423462, "learning_rate": 0.001, "loss": 1.3983, "step": 233600 }, { "epoch": 75.53329023917259, "grad_norm": 1.855002999305725, "learning_rate": 0.001, "loss": 1.4196, "step": 233700 }, { "epoch": 75.56561085972851, "grad_norm": 2.0383918285369873, "learning_rate": 0.001, "loss": 1.4305, "step": 233800 }, { "epoch": 75.59793148028442, "grad_norm": 2.192237377166748, "learning_rate": 0.001, "loss": 1.4212, "step": 233900 }, { "epoch": 75.63025210084034, "grad_norm": 3.2227871417999268, "learning_rate": 0.001, "loss": 1.4237, "step": 234000 }, { "epoch": 75.66257272139624, "grad_norm": 2.169130325317383, "learning_rate": 0.001, "loss": 1.4232, "step": 234100 }, { "epoch": 75.69489334195217, "grad_norm": 1.9388642311096191, "learning_rate": 0.001, "loss": 1.423, "step": 234200 }, { "epoch": 75.72721396250807, "grad_norm": 1.8224081993103027, "learning_rate": 0.001, "loss": 1.4394, "step": 234300 }, { "epoch": 75.759534583064, "grad_norm": 2.401827335357666, "learning_rate": 0.001, "loss": 1.4528, "step": 234400 }, { "epoch": 75.7918552036199, "grad_norm": 1.6107642650604248, "learning_rate": 0.001, "loss": 1.4336, "step": 234500 }, { "epoch": 75.82417582417582, "grad_norm": 2.166220188140869, "learning_rate": 0.001, "loss": 1.4319, "step": 234600 }, { "epoch": 75.85649644473173, "grad_norm": 2.4662373065948486, "learning_rate": 0.001, "loss": 1.4491, "step": 234700 }, { "epoch": 75.88881706528765, "grad_norm": 2.3787710666656494, "learning_rate": 0.001, "loss": 1.4581, "step": 234800 }, { "epoch": 75.92113768584356, "grad_norm": 1.996765375137329, "learning_rate": 0.001, "loss": 1.46, "step": 234900 }, { "epoch": 75.95345830639948, "grad_norm": 1.8469176292419434, "learning_rate": 0.001, "loss": 1.4564, "step": 235000 }, { "epoch": 75.98577892695539, "grad_norm": 2.4234673976898193, "learning_rate": 0.001, "loss": 1.4635, "step": 235100 }, { "epoch": 76.01809954751131, "grad_norm": 1.4825252294540405, "learning_rate": 0.001, "loss": 1.407, "step": 235200 }, { "epoch": 76.05042016806723, "grad_norm": 1.2833598852157593, "learning_rate": 0.001, "loss": 1.3349, "step": 235300 }, { "epoch": 76.08274078862314, "grad_norm": 1.134111762046814, "learning_rate": 0.001, "loss": 1.3442, "step": 235400 }, { "epoch": 76.11506140917906, "grad_norm": 1.416589617729187, "learning_rate": 0.001, "loss": 1.3481, "step": 235500 }, { "epoch": 76.14738202973497, "grad_norm": 1.3772751092910767, "learning_rate": 0.001, "loss": 1.3542, "step": 235600 }, { "epoch": 76.17970265029089, "grad_norm": 1.658582329750061, "learning_rate": 0.001, "loss": 1.3667, "step": 235700 }, { "epoch": 76.2120232708468, "grad_norm": 1.726936936378479, "learning_rate": 0.001, "loss": 1.3662, "step": 235800 }, { "epoch": 76.24434389140272, "grad_norm": 1.7430275678634644, "learning_rate": 0.001, "loss": 1.3698, "step": 235900 }, { "epoch": 76.27666451195863, "grad_norm": 1.8353071212768555, "learning_rate": 0.001, "loss": 1.3772, "step": 236000 }, { "epoch": 76.30898513251455, "grad_norm": 1.454705834388733, "learning_rate": 0.001, "loss": 1.3869, "step": 236100 }, { "epoch": 76.34130575307046, "grad_norm": 1.9173542261123657, "learning_rate": 0.001, "loss": 1.3844, "step": 236200 }, { "epoch": 76.37362637362638, "grad_norm": 1.5633434057235718, "learning_rate": 0.001, "loss": 1.3971, "step": 236300 }, { "epoch": 76.40594699418229, "grad_norm": 1.1957085132598877, "learning_rate": 0.001, "loss": 1.3983, "step": 236400 }, { "epoch": 76.43826761473821, "grad_norm": 1.5280741453170776, "learning_rate": 0.001, "loss": 1.4086, "step": 236500 }, { "epoch": 76.47058823529412, "grad_norm": 1.8125216960906982, "learning_rate": 0.001, "loss": 1.4004, "step": 236600 }, { "epoch": 76.50290885585004, "grad_norm": 1.591016173362732, "learning_rate": 0.001, "loss": 1.4034, "step": 236700 }, { "epoch": 76.53522947640595, "grad_norm": 1.4297205209732056, "learning_rate": 0.001, "loss": 1.4066, "step": 236800 }, { "epoch": 76.56755009696187, "grad_norm": 2.205801248550415, "learning_rate": 0.001, "loss": 1.4221, "step": 236900 }, { "epoch": 76.59987071751777, "grad_norm": 1.6374117136001587, "learning_rate": 0.001, "loss": 1.4368, "step": 237000 }, { "epoch": 76.6321913380737, "grad_norm": 1.2918282747268677, "learning_rate": 0.001, "loss": 1.4166, "step": 237100 }, { "epoch": 76.6645119586296, "grad_norm": 1.585101842880249, "learning_rate": 0.001, "loss": 1.4093, "step": 237200 }, { "epoch": 76.69683257918552, "grad_norm": 1.4704420566558838, "learning_rate": 0.001, "loss": 1.4287, "step": 237300 }, { "epoch": 76.72915319974143, "grad_norm": 1.5124256610870361, "learning_rate": 0.001, "loss": 1.4389, "step": 237400 }, { "epoch": 76.76147382029735, "grad_norm": 1.4330766201019287, "learning_rate": 0.001, "loss": 1.403, "step": 237500 }, { "epoch": 76.79379444085326, "grad_norm": 1.3854273557662964, "learning_rate": 0.001, "loss": 1.4406, "step": 237600 }, { "epoch": 76.82611506140918, "grad_norm": 1.5856598615646362, "learning_rate": 0.001, "loss": 1.426, "step": 237700 }, { "epoch": 76.85843568196509, "grad_norm": 2.090639114379883, "learning_rate": 0.001, "loss": 1.45, "step": 237800 }, { "epoch": 76.89075630252101, "grad_norm": 1.3450896739959717, "learning_rate": 0.001, "loss": 1.445, "step": 237900 }, { "epoch": 76.92307692307692, "grad_norm": 1.5369879007339478, "learning_rate": 0.001, "loss": 1.4475, "step": 238000 }, { "epoch": 76.95539754363284, "grad_norm": 1.3403236865997314, "learning_rate": 0.001, "loss": 1.4611, "step": 238100 }, { "epoch": 76.98771816418875, "grad_norm": 1.4753422737121582, "learning_rate": 0.001, "loss": 1.4393, "step": 238200 }, { "epoch": 77.02003878474467, "grad_norm": 1.7514846324920654, "learning_rate": 0.001, "loss": 1.3893, "step": 238300 }, { "epoch": 77.05235940530058, "grad_norm": 1.336416482925415, "learning_rate": 0.001, "loss": 1.3374, "step": 238400 }, { "epoch": 77.0846800258565, "grad_norm": 1.7057934999465942, "learning_rate": 0.001, "loss": 1.3572, "step": 238500 }, { "epoch": 77.11700064641241, "grad_norm": 1.643702745437622, "learning_rate": 0.001, "loss": 1.3525, "step": 238600 }, { "epoch": 77.14932126696833, "grad_norm": 1.624499797821045, "learning_rate": 0.001, "loss": 1.351, "step": 238700 }, { "epoch": 77.18164188752424, "grad_norm": 1.3781622648239136, "learning_rate": 0.001, "loss": 1.3642, "step": 238800 }, { "epoch": 77.21396250808016, "grad_norm": 1.6872320175170898, "learning_rate": 0.001, "loss": 1.3662, "step": 238900 }, { "epoch": 77.24628312863607, "grad_norm": 2.0610127449035645, "learning_rate": 0.001, "loss": 1.3575, "step": 239000 }, { "epoch": 77.27860374919199, "grad_norm": 2.052093029022217, "learning_rate": 0.001, "loss": 1.3666, "step": 239100 }, { "epoch": 77.3109243697479, "grad_norm": 1.6374685764312744, "learning_rate": 0.001, "loss": 1.3874, "step": 239200 }, { "epoch": 77.34324499030382, "grad_norm": 1.2537875175476074, "learning_rate": 0.001, "loss": 1.3643, "step": 239300 }, { "epoch": 77.37556561085972, "grad_norm": 1.6449648141860962, "learning_rate": 0.001, "loss": 1.3924, "step": 239400 }, { "epoch": 77.40788623141565, "grad_norm": 1.5691554546356201, "learning_rate": 0.001, "loss": 1.39, "step": 239500 }, { "epoch": 77.44020685197155, "grad_norm": 1.6928367614746094, "learning_rate": 0.001, "loss": 1.3914, "step": 239600 }, { "epoch": 77.47252747252747, "grad_norm": 1.9325333833694458, "learning_rate": 0.001, "loss": 1.3968, "step": 239700 }, { "epoch": 77.50484809308338, "grad_norm": 1.136733055114746, "learning_rate": 0.001, "loss": 1.3843, "step": 239800 }, { "epoch": 77.5371687136393, "grad_norm": 1.2559707164764404, "learning_rate": 0.001, "loss": 1.4021, "step": 239900 }, { "epoch": 77.56948933419521, "grad_norm": 1.187476634979248, "learning_rate": 0.001, "loss": 1.3994, "step": 240000 }, { "epoch": 77.60180995475113, "grad_norm": 2.1751227378845215, "learning_rate": 0.001, "loss": 1.4008, "step": 240100 }, { "epoch": 77.63413057530704, "grad_norm": 1.358201503753662, "learning_rate": 0.001, "loss": 1.418, "step": 240200 }, { "epoch": 77.66645119586296, "grad_norm": 2.1808340549468994, "learning_rate": 0.001, "loss": 1.4228, "step": 240300 }, { "epoch": 77.69877181641887, "grad_norm": 1.439213752746582, "learning_rate": 0.001, "loss": 1.4134, "step": 240400 }, { "epoch": 77.73109243697479, "grad_norm": 1.6052058935165405, "learning_rate": 0.001, "loss": 1.4239, "step": 240500 }, { "epoch": 77.7634130575307, "grad_norm": 1.9990873336791992, "learning_rate": 0.001, "loss": 1.4161, "step": 240600 }, { "epoch": 77.79573367808662, "grad_norm": 1.7052737474441528, "learning_rate": 0.001, "loss": 1.4266, "step": 240700 }, { "epoch": 77.82805429864253, "grad_norm": 1.595995306968689, "learning_rate": 0.001, "loss": 1.4319, "step": 240800 }, { "epoch": 77.86037491919845, "grad_norm": 1.4590080976486206, "learning_rate": 0.001, "loss": 1.4249, "step": 240900 }, { "epoch": 77.89269553975436, "grad_norm": 1.320956826210022, "learning_rate": 0.001, "loss": 1.4256, "step": 241000 }, { "epoch": 77.92501616031028, "grad_norm": 1.5802559852600098, "learning_rate": 0.001, "loss": 1.4435, "step": 241100 }, { "epoch": 77.95733678086619, "grad_norm": 1.6122311353683472, "learning_rate": 0.001, "loss": 1.4411, "step": 241200 }, { "epoch": 77.98965740142211, "grad_norm": 1.8389743566513062, "learning_rate": 0.001, "loss": 1.4368, "step": 241300 }, { "epoch": 78.02197802197803, "grad_norm": 1.7278374433517456, "learning_rate": 0.001, "loss": 1.3722, "step": 241400 }, { "epoch": 78.05429864253394, "grad_norm": 1.2872110605239868, "learning_rate": 0.001, "loss": 1.3232, "step": 241500 }, { "epoch": 78.08661926308986, "grad_norm": 1.199270486831665, "learning_rate": 0.001, "loss": 1.3319, "step": 241600 }, { "epoch": 78.11893988364577, "grad_norm": 2.1235804557800293, "learning_rate": 0.001, "loss": 1.3375, "step": 241700 }, { "epoch": 78.15126050420169, "grad_norm": 1.7035527229309082, "learning_rate": 0.001, "loss": 1.3381, "step": 241800 }, { "epoch": 78.1835811247576, "grad_norm": 1.2697619199752808, "learning_rate": 0.001, "loss": 1.3564, "step": 241900 }, { "epoch": 78.21590174531352, "grad_norm": 1.3939249515533447, "learning_rate": 0.001, "loss": 1.3411, "step": 242000 }, { "epoch": 78.24822236586942, "grad_norm": 1.811110019683838, "learning_rate": 0.001, "loss": 1.3553, "step": 242100 }, { "epoch": 78.28054298642535, "grad_norm": 1.4949676990509033, "learning_rate": 0.001, "loss": 1.3715, "step": 242200 }, { "epoch": 78.31286360698125, "grad_norm": 1.3313863277435303, "learning_rate": 0.001, "loss": 1.3589, "step": 242300 }, { "epoch": 78.34518422753717, "grad_norm": 1.227896809577942, "learning_rate": 0.001, "loss": 1.3671, "step": 242400 }, { "epoch": 78.37750484809308, "grad_norm": 1.4327553510665894, "learning_rate": 0.001, "loss": 1.3876, "step": 242500 }, { "epoch": 78.409825468649, "grad_norm": 1.4522910118103027, "learning_rate": 0.001, "loss": 1.3734, "step": 242600 }, { "epoch": 78.44214608920491, "grad_norm": 1.439487099647522, "learning_rate": 0.001, "loss": 1.3789, "step": 242700 }, { "epoch": 78.47446670976083, "grad_norm": 1.3468687534332275, "learning_rate": 0.001, "loss": 1.3858, "step": 242800 }, { "epoch": 78.50678733031674, "grad_norm": 1.7873753309249878, "learning_rate": 0.001, "loss": 1.3794, "step": 242900 }, { "epoch": 78.53910795087266, "grad_norm": 1.6426472663879395, "learning_rate": 0.001, "loss": 1.4109, "step": 243000 }, { "epoch": 78.57142857142857, "grad_norm": 1.3578706979751587, "learning_rate": 0.001, "loss": 1.4018, "step": 243100 }, { "epoch": 78.60374919198449, "grad_norm": 1.8222377300262451, "learning_rate": 0.001, "loss": 1.4009, "step": 243200 }, { "epoch": 78.6360698125404, "grad_norm": 1.3069651126861572, "learning_rate": 0.001, "loss": 1.4035, "step": 243300 }, { "epoch": 78.66839043309632, "grad_norm": 1.7161260843276978, "learning_rate": 0.001, "loss": 1.4051, "step": 243400 }, { "epoch": 78.70071105365223, "grad_norm": 1.2696171998977661, "learning_rate": 0.001, "loss": 1.406, "step": 243500 }, { "epoch": 78.73303167420815, "grad_norm": 1.432579517364502, "learning_rate": 0.001, "loss": 1.4206, "step": 243600 }, { "epoch": 78.76535229476406, "grad_norm": 1.4368444681167603, "learning_rate": 0.001, "loss": 1.4183, "step": 243700 }, { "epoch": 78.79767291531998, "grad_norm": 1.452947735786438, "learning_rate": 0.001, "loss": 1.4229, "step": 243800 }, { "epoch": 78.82999353587589, "grad_norm": 1.554884910583496, "learning_rate": 0.001, "loss": 1.4293, "step": 243900 }, { "epoch": 78.86231415643181, "grad_norm": 1.30029296875, "learning_rate": 0.001, "loss": 1.4314, "step": 244000 }, { "epoch": 78.89463477698771, "grad_norm": 1.4925910234451294, "learning_rate": 0.001, "loss": 1.4202, "step": 244100 }, { "epoch": 78.92695539754364, "grad_norm": 1.5145576000213623, "learning_rate": 0.001, "loss": 1.4292, "step": 244200 }, { "epoch": 78.95927601809954, "grad_norm": 1.426664113998413, "learning_rate": 0.001, "loss": 1.4427, "step": 244300 }, { "epoch": 78.99159663865547, "grad_norm": 1.0887047052383423, "learning_rate": 0.001, "loss": 1.4285, "step": 244400 }, { "epoch": 79.02391725921137, "grad_norm": 1.4953248500823975, "learning_rate": 0.001, "loss": 1.3586, "step": 244500 }, { "epoch": 79.0562378797673, "grad_norm": 1.7633830308914185, "learning_rate": 0.001, "loss": 1.3118, "step": 244600 }, { "epoch": 79.0885585003232, "grad_norm": 1.8045114278793335, "learning_rate": 0.001, "loss": 1.3062, "step": 244700 }, { "epoch": 79.12087912087912, "grad_norm": 1.6315350532531738, "learning_rate": 0.001, "loss": 1.3279, "step": 244800 }, { "epoch": 79.15319974143503, "grad_norm": 1.222394347190857, "learning_rate": 0.001, "loss": 1.3467, "step": 244900 }, { "epoch": 79.18552036199095, "grad_norm": 1.5777305364608765, "learning_rate": 0.001, "loss": 1.3519, "step": 245000 }, { "epoch": 79.21784098254686, "grad_norm": 1.2325632572174072, "learning_rate": 0.001, "loss": 1.3639, "step": 245100 }, { "epoch": 79.25016160310278, "grad_norm": 1.2535754442214966, "learning_rate": 0.001, "loss": 1.3473, "step": 245200 }, { "epoch": 79.28248222365869, "grad_norm": 1.4184768199920654, "learning_rate": 0.001, "loss": 1.3584, "step": 245300 }, { "epoch": 79.31480284421461, "grad_norm": 1.9302608966827393, "learning_rate": 0.001, "loss": 1.3527, "step": 245400 }, { "epoch": 79.34712346477052, "grad_norm": 1.5769174098968506, "learning_rate": 0.001, "loss": 1.3684, "step": 245500 }, { "epoch": 79.37944408532644, "grad_norm": 1.8105542659759521, "learning_rate": 0.001, "loss": 1.3696, "step": 245600 }, { "epoch": 79.41176470588235, "grad_norm": 1.3575557470321655, "learning_rate": 0.001, "loss": 1.3777, "step": 245700 }, { "epoch": 79.44408532643827, "grad_norm": 1.6130093336105347, "learning_rate": 0.001, "loss": 1.3734, "step": 245800 }, { "epoch": 79.47640594699418, "grad_norm": 1.426985740661621, "learning_rate": 0.001, "loss": 1.3688, "step": 245900 }, { "epoch": 79.5087265675501, "grad_norm": 1.4327107667922974, "learning_rate": 0.001, "loss": 1.3744, "step": 246000 }, { "epoch": 79.541047188106, "grad_norm": 1.8427186012268066, "learning_rate": 0.001, "loss": 1.3844, "step": 246100 }, { "epoch": 79.57336780866193, "grad_norm": 1.5242940187454224, "learning_rate": 0.001, "loss": 1.3893, "step": 246200 }, { "epoch": 79.60568842921784, "grad_norm": 1.4538577795028687, "learning_rate": 0.001, "loss": 1.3919, "step": 246300 }, { "epoch": 79.63800904977376, "grad_norm": 1.25160551071167, "learning_rate": 0.001, "loss": 1.4005, "step": 246400 }, { "epoch": 79.67032967032966, "grad_norm": 1.2579987049102783, "learning_rate": 0.001, "loss": 1.3975, "step": 246500 }, { "epoch": 79.70265029088559, "grad_norm": 2.07604718208313, "learning_rate": 0.001, "loss": 1.3984, "step": 246600 }, { "epoch": 79.7349709114415, "grad_norm": 1.2995349168777466, "learning_rate": 0.001, "loss": 1.3993, "step": 246700 }, { "epoch": 79.76729153199742, "grad_norm": 1.375459909439087, "learning_rate": 0.001, "loss": 1.3973, "step": 246800 }, { "epoch": 79.79961215255332, "grad_norm": 1.4694504737854004, "learning_rate": 0.001, "loss": 1.4256, "step": 246900 }, { "epoch": 79.83193277310924, "grad_norm": 1.3293267488479614, "learning_rate": 0.001, "loss": 1.4212, "step": 247000 }, { "epoch": 79.86425339366515, "grad_norm": 1.365396499633789, "learning_rate": 0.001, "loss": 1.4241, "step": 247100 }, { "epoch": 79.89657401422107, "grad_norm": 1.585095763206482, "learning_rate": 0.001, "loss": 1.4243, "step": 247200 }, { "epoch": 79.92889463477698, "grad_norm": 1.4808619022369385, "learning_rate": 0.001, "loss": 1.4353, "step": 247300 }, { "epoch": 79.9612152553329, "grad_norm": 1.486207365989685, "learning_rate": 0.001, "loss": 1.4415, "step": 247400 }, { "epoch": 79.99353587588882, "grad_norm": 1.218306303024292, "learning_rate": 0.001, "loss": 1.4317, "step": 247500 }, { "epoch": 80.02585649644473, "grad_norm": 1.3337205648422241, "learning_rate": 0.001, "loss": 1.3458, "step": 247600 }, { "epoch": 80.05817711700065, "grad_norm": 1.48308527469635, "learning_rate": 0.001, "loss": 1.3185, "step": 247700 }, { "epoch": 80.09049773755656, "grad_norm": 1.4150961637496948, "learning_rate": 0.001, "loss": 1.307, "step": 247800 }, { "epoch": 80.12281835811248, "grad_norm": 1.5947446823120117, "learning_rate": 0.001, "loss": 1.3251, "step": 247900 }, { "epoch": 80.15513897866839, "grad_norm": 1.572667121887207, "learning_rate": 0.001, "loss": 1.3371, "step": 248000 }, { "epoch": 80.18745959922431, "grad_norm": 1.3705382347106934, "learning_rate": 0.001, "loss": 1.3321, "step": 248100 }, { "epoch": 80.21978021978022, "grad_norm": 1.3644392490386963, "learning_rate": 0.001, "loss": 1.3476, "step": 248200 }, { "epoch": 80.25210084033614, "grad_norm": 1.9536592960357666, "learning_rate": 0.001, "loss": 1.341, "step": 248300 }, { "epoch": 80.28442146089205, "grad_norm": 1.5622602701187134, "learning_rate": 0.001, "loss": 1.3687, "step": 248400 }, { "epoch": 80.31674208144797, "grad_norm": 1.4836679697036743, "learning_rate": 0.001, "loss": 1.3541, "step": 248500 }, { "epoch": 80.34906270200388, "grad_norm": 1.6780399084091187, "learning_rate": 0.001, "loss": 1.3646, "step": 248600 }, { "epoch": 80.3813833225598, "grad_norm": 1.3770062923431396, "learning_rate": 0.001, "loss": 1.3766, "step": 248700 }, { "epoch": 80.4137039431157, "grad_norm": 1.387833595275879, "learning_rate": 0.001, "loss": 1.3698, "step": 248800 }, { "epoch": 80.44602456367163, "grad_norm": 1.4488047361373901, "learning_rate": 0.001, "loss": 1.3665, "step": 248900 }, { "epoch": 80.47834518422754, "grad_norm": 1.252181053161621, "learning_rate": 0.001, "loss": 1.3767, "step": 249000 }, { "epoch": 80.51066580478346, "grad_norm": 1.396899938583374, "learning_rate": 0.001, "loss": 1.3709, "step": 249100 }, { "epoch": 80.54298642533936, "grad_norm": 1.8860456943511963, "learning_rate": 0.001, "loss": 1.3729, "step": 249200 }, { "epoch": 80.57530704589529, "grad_norm": 1.6136033535003662, "learning_rate": 0.001, "loss": 1.392, "step": 249300 }, { "epoch": 80.6076276664512, "grad_norm": 1.41468346118927, "learning_rate": 0.001, "loss": 1.3889, "step": 249400 }, { "epoch": 80.63994828700712, "grad_norm": 1.5444872379302979, "learning_rate": 0.001, "loss": 1.3756, "step": 249500 }, { "epoch": 80.67226890756302, "grad_norm": 1.5710728168487549, "learning_rate": 0.001, "loss": 1.3886, "step": 249600 }, { "epoch": 80.70458952811894, "grad_norm": 1.7316027879714966, "learning_rate": 0.001, "loss": 1.3814, "step": 249700 }, { "epoch": 80.73691014867485, "grad_norm": 1.8401299715042114, "learning_rate": 0.001, "loss": 1.3982, "step": 249800 }, { "epoch": 80.76923076923077, "grad_norm": 1.5051045417785645, "learning_rate": 0.001, "loss": 1.3961, "step": 249900 }, { "epoch": 80.80155138978668, "grad_norm": 1.2414015531539917, "learning_rate": 0.001, "loss": 1.4, "step": 250000 }, { "epoch": 80.8338720103426, "grad_norm": 1.8896925449371338, "learning_rate": 0.001, "loss": 1.4005, "step": 250100 }, { "epoch": 80.86619263089851, "grad_norm": 1.4821033477783203, "learning_rate": 0.001, "loss": 1.414, "step": 250200 }, { "epoch": 80.89851325145443, "grad_norm": 1.7811696529388428, "learning_rate": 0.001, "loss": 1.4248, "step": 250300 }, { "epoch": 80.93083387201034, "grad_norm": 1.293363332748413, "learning_rate": 0.001, "loss": 1.4217, "step": 250400 }, { "epoch": 80.96315449256626, "grad_norm": 1.5930306911468506, "learning_rate": 0.001, "loss": 1.4193, "step": 250500 }, { "epoch": 80.99547511312217, "grad_norm": 1.5660650730133057, "learning_rate": 0.001, "loss": 1.4292, "step": 250600 }, { "epoch": 81.02779573367809, "grad_norm": 1.3551260232925415, "learning_rate": 0.001, "loss": 1.3123, "step": 250700 }, { "epoch": 81.060116354234, "grad_norm": 1.4512505531311035, "learning_rate": 0.001, "loss": 1.3193, "step": 250800 }, { "epoch": 81.09243697478992, "grad_norm": 1.7034558057785034, "learning_rate": 0.001, "loss": 1.3068, "step": 250900 }, { "epoch": 81.12475759534583, "grad_norm": 1.5442638397216797, "learning_rate": 0.001, "loss": 1.3242, "step": 251000 }, { "epoch": 81.15707821590175, "grad_norm": 1.3783601522445679, "learning_rate": 0.001, "loss": 1.3209, "step": 251100 }, { "epoch": 81.18939883645766, "grad_norm": 1.6340938806533813, "learning_rate": 0.001, "loss": 1.3332, "step": 251200 }, { "epoch": 81.22171945701358, "grad_norm": 1.429200291633606, "learning_rate": 0.001, "loss": 1.3159, "step": 251300 }, { "epoch": 81.25404007756948, "grad_norm": 1.5749098062515259, "learning_rate": 0.001, "loss": 1.3318, "step": 251400 }, { "epoch": 81.2863606981254, "grad_norm": 1.5038208961486816, "learning_rate": 0.001, "loss": 1.3545, "step": 251500 }, { "epoch": 81.31868131868131, "grad_norm": 1.7742317914962769, "learning_rate": 0.001, "loss": 1.3519, "step": 251600 }, { "epoch": 81.35100193923724, "grad_norm": 1.8789876699447632, "learning_rate": 0.001, "loss": 1.3502, "step": 251700 }, { "epoch": 81.38332255979314, "grad_norm": 1.5382606983184814, "learning_rate": 0.001, "loss": 1.356, "step": 251800 }, { "epoch": 81.41564318034906, "grad_norm": 1.2025485038757324, "learning_rate": 0.001, "loss": 1.3393, "step": 251900 }, { "epoch": 81.44796380090497, "grad_norm": 1.3355134725570679, "learning_rate": 0.001, "loss": 1.3805, "step": 252000 }, { "epoch": 81.4802844214609, "grad_norm": 1.201806664466858, "learning_rate": 0.001, "loss": 1.3651, "step": 252100 }, { "epoch": 81.5126050420168, "grad_norm": 1.2222882509231567, "learning_rate": 0.001, "loss": 1.3714, "step": 252200 }, { "epoch": 81.54492566257272, "grad_norm": 1.7975983619689941, "learning_rate": 0.001, "loss": 1.3684, "step": 252300 }, { "epoch": 81.57724628312863, "grad_norm": 1.6185847520828247, "learning_rate": 0.001, "loss": 1.3823, "step": 252400 }, { "epoch": 81.60956690368455, "grad_norm": 2.2196691036224365, "learning_rate": 0.001, "loss": 1.3948, "step": 252500 }, { "epoch": 81.64188752424046, "grad_norm": 1.5387237071990967, "learning_rate": 0.001, "loss": 1.3736, "step": 252600 }, { "epoch": 81.67420814479638, "grad_norm": 1.4336892366409302, "learning_rate": 0.001, "loss": 1.3801, "step": 252700 }, { "epoch": 81.70652876535229, "grad_norm": 2.1763482093811035, "learning_rate": 0.001, "loss": 1.3884, "step": 252800 }, { "epoch": 81.73884938590821, "grad_norm": 1.289603352546692, "learning_rate": 0.001, "loss": 1.4009, "step": 252900 }, { "epoch": 81.77117000646412, "grad_norm": 2.088289499282837, "learning_rate": 0.001, "loss": 1.386, "step": 253000 }, { "epoch": 81.80349062702004, "grad_norm": 1.4783682823181152, "learning_rate": 0.001, "loss": 1.4053, "step": 253100 }, { "epoch": 81.83581124757595, "grad_norm": 1.3838250637054443, "learning_rate": 0.001, "loss": 1.4138, "step": 253200 }, { "epoch": 81.86813186813187, "grad_norm": 1.427379846572876, "learning_rate": 0.001, "loss": 1.4119, "step": 253300 }, { "epoch": 81.90045248868778, "grad_norm": 1.6082667112350464, "learning_rate": 0.001, "loss": 1.403, "step": 253400 }, { "epoch": 81.9327731092437, "grad_norm": 1.5411512851715088, "learning_rate": 0.001, "loss": 1.4213, "step": 253500 }, { "epoch": 81.9650937297996, "grad_norm": 1.5107688903808594, "learning_rate": 0.001, "loss": 1.4177, "step": 253600 }, { "epoch": 81.99741435035553, "grad_norm": 1.9440659284591675, "learning_rate": 0.001, "loss": 1.4189, "step": 253700 }, { "epoch": 82.02973497091145, "grad_norm": 1.7341686487197876, "learning_rate": 0.001, "loss": 1.3041, "step": 253800 }, { "epoch": 82.06205559146736, "grad_norm": 1.5955758094787598, "learning_rate": 0.001, "loss": 1.3014, "step": 253900 }, { "epoch": 82.09437621202328, "grad_norm": 1.4994001388549805, "learning_rate": 0.001, "loss": 1.3156, "step": 254000 }, { "epoch": 82.12669683257919, "grad_norm": 1.6090569496154785, "learning_rate": 0.001, "loss": 1.3052, "step": 254100 }, { "epoch": 82.1590174531351, "grad_norm": 2.008561849594116, "learning_rate": 0.001, "loss": 1.3321, "step": 254200 }, { "epoch": 82.19133807369101, "grad_norm": 1.6287059783935547, "learning_rate": 0.001, "loss": 1.3321, "step": 254300 }, { "epoch": 82.22365869424694, "grad_norm": 1.3676732778549194, "learning_rate": 0.001, "loss": 1.3275, "step": 254400 }, { "epoch": 82.25597931480284, "grad_norm": 2.070295572280884, "learning_rate": 0.001, "loss": 1.3326, "step": 254500 }, { "epoch": 82.28829993535876, "grad_norm": 1.616440773010254, "learning_rate": 0.001, "loss": 1.3309, "step": 254600 }, { "epoch": 82.32062055591467, "grad_norm": 1.398174524307251, "learning_rate": 0.001, "loss": 1.3364, "step": 254700 }, { "epoch": 82.3529411764706, "grad_norm": 1.4400818347930908, "learning_rate": 0.001, "loss": 1.3502, "step": 254800 }, { "epoch": 82.3852617970265, "grad_norm": 1.355803370475769, "learning_rate": 0.001, "loss": 1.3546, "step": 254900 }, { "epoch": 82.41758241758242, "grad_norm": 1.6952476501464844, "learning_rate": 0.001, "loss": 1.3459, "step": 255000 }, { "epoch": 82.44990303813833, "grad_norm": 1.9007939100265503, "learning_rate": 0.001, "loss": 1.3412, "step": 255100 }, { "epoch": 82.48222365869425, "grad_norm": 1.2914032936096191, "learning_rate": 0.001, "loss": 1.3539, "step": 255200 }, { "epoch": 82.51454427925016, "grad_norm": 1.625333309173584, "learning_rate": 0.001, "loss": 1.3661, "step": 255300 }, { "epoch": 82.54686489980608, "grad_norm": 1.394900918006897, "learning_rate": 0.001, "loss": 1.3736, "step": 255400 }, { "epoch": 82.57918552036199, "grad_norm": 1.809337854385376, "learning_rate": 0.001, "loss": 1.3596, "step": 255500 }, { "epoch": 82.61150614091791, "grad_norm": 1.5900062322616577, "learning_rate": 0.001, "loss": 1.3831, "step": 255600 }, { "epoch": 82.64382676147382, "grad_norm": 1.26161789894104, "learning_rate": 0.001, "loss": 1.3672, "step": 255700 }, { "epoch": 82.67614738202974, "grad_norm": 1.521141767501831, "learning_rate": 0.001, "loss": 1.3775, "step": 255800 }, { "epoch": 82.70846800258565, "grad_norm": 1.447628140449524, "learning_rate": 0.001, "loss": 1.3968, "step": 255900 }, { "epoch": 82.74078862314157, "grad_norm": 1.5999592542648315, "learning_rate": 0.001, "loss": 1.3887, "step": 256000 }, { "epoch": 82.77310924369748, "grad_norm": 2.224372386932373, "learning_rate": 0.001, "loss": 1.3877, "step": 256100 }, { "epoch": 82.8054298642534, "grad_norm": 1.854551911354065, "learning_rate": 0.001, "loss": 1.3854, "step": 256200 }, { "epoch": 82.8377504848093, "grad_norm": 1.9979218244552612, "learning_rate": 0.001, "loss": 1.3951, "step": 256300 }, { "epoch": 82.87007110536523, "grad_norm": 1.5547971725463867, "learning_rate": 0.001, "loss": 1.3918, "step": 256400 }, { "epoch": 82.90239172592113, "grad_norm": 1.7112253904342651, "learning_rate": 0.001, "loss": 1.4129, "step": 256500 }, { "epoch": 82.93471234647706, "grad_norm": 1.6911065578460693, "learning_rate": 0.001, "loss": 1.4009, "step": 256600 }, { "epoch": 82.96703296703296, "grad_norm": 1.5267651081085205, "learning_rate": 0.001, "loss": 1.4078, "step": 256700 }, { "epoch": 82.99935358758889, "grad_norm": 1.7838150262832642, "learning_rate": 0.001, "loss": 1.3939, "step": 256800 }, { "epoch": 83.03167420814479, "grad_norm": 1.4784297943115234, "learning_rate": 0.001, "loss": 1.2894, "step": 256900 }, { "epoch": 83.06399482870071, "grad_norm": 2.03774094581604, "learning_rate": 0.001, "loss": 1.2907, "step": 257000 }, { "epoch": 83.09631544925662, "grad_norm": 1.9565315246582031, "learning_rate": 0.001, "loss": 1.3025, "step": 257100 }, { "epoch": 83.12863606981254, "grad_norm": 1.5921435356140137, "learning_rate": 0.001, "loss": 1.3043, "step": 257200 }, { "epoch": 83.16095669036845, "grad_norm": 1.700293779373169, "learning_rate": 0.001, "loss": 1.3172, "step": 257300 }, { "epoch": 83.19327731092437, "grad_norm": 1.9305436611175537, "learning_rate": 0.001, "loss": 1.3173, "step": 257400 }, { "epoch": 83.22559793148028, "grad_norm": 1.34230637550354, "learning_rate": 0.001, "loss": 1.3124, "step": 257500 }, { "epoch": 83.2579185520362, "grad_norm": 1.5128107070922852, "learning_rate": 0.001, "loss": 1.3324, "step": 257600 }, { "epoch": 83.29023917259211, "grad_norm": 1.4874143600463867, "learning_rate": 0.001, "loss": 1.3352, "step": 257700 }, { "epoch": 83.32255979314803, "grad_norm": 1.7499667406082153, "learning_rate": 0.001, "loss": 1.3258, "step": 257800 }, { "epoch": 83.35488041370394, "grad_norm": 2.480471611022949, "learning_rate": 0.001, "loss": 1.3301, "step": 257900 }, { "epoch": 83.38720103425986, "grad_norm": 1.7621139287948608, "learning_rate": 0.001, "loss": 1.3329, "step": 258000 }, { "epoch": 83.41952165481577, "grad_norm": 1.40170156955719, "learning_rate": 0.001, "loss": 1.3384, "step": 258100 }, { "epoch": 83.45184227537169, "grad_norm": 1.850550889968872, "learning_rate": 0.001, "loss": 1.3502, "step": 258200 }, { "epoch": 83.4841628959276, "grad_norm": 2.19671630859375, "learning_rate": 0.001, "loss": 1.3566, "step": 258300 }, { "epoch": 83.51648351648352, "grad_norm": 2.058199405670166, "learning_rate": 0.001, "loss": 1.3641, "step": 258400 }, { "epoch": 83.54880413703943, "grad_norm": 2.0466177463531494, "learning_rate": 0.001, "loss": 1.3537, "step": 258500 }, { "epoch": 83.58112475759535, "grad_norm": 1.7297381162643433, "learning_rate": 0.001, "loss": 1.362, "step": 258600 }, { "epoch": 83.61344537815125, "grad_norm": 2.032058000564575, "learning_rate": 0.001, "loss": 1.378, "step": 258700 }, { "epoch": 83.64576599870718, "grad_norm": 1.6873259544372559, "learning_rate": 0.001, "loss": 1.3815, "step": 258800 }, { "epoch": 83.67808661926308, "grad_norm": 2.212801456451416, "learning_rate": 0.001, "loss": 1.3635, "step": 258900 }, { "epoch": 83.710407239819, "grad_norm": 1.6788281202316284, "learning_rate": 0.001, "loss": 1.3743, "step": 259000 }, { "epoch": 83.74272786037491, "grad_norm": 1.7809021472930908, "learning_rate": 0.001, "loss": 1.3808, "step": 259100 }, { "epoch": 83.77504848093083, "grad_norm": 2.2536916732788086, "learning_rate": 0.001, "loss": 1.3789, "step": 259200 }, { "epoch": 83.80736910148674, "grad_norm": 1.964429259300232, "learning_rate": 0.001, "loss": 1.3914, "step": 259300 }, { "epoch": 83.83968972204266, "grad_norm": 1.6335227489471436, "learning_rate": 0.001, "loss": 1.3955, "step": 259400 }, { "epoch": 83.87201034259857, "grad_norm": 2.0025200843811035, "learning_rate": 0.001, "loss": 1.3997, "step": 259500 }, { "epoch": 83.9043309631545, "grad_norm": 1.9986916780471802, "learning_rate": 0.001, "loss": 1.4021, "step": 259600 }, { "epoch": 83.9366515837104, "grad_norm": 1.4429872035980225, "learning_rate": 0.001, "loss": 1.4065, "step": 259700 }, { "epoch": 83.96897220426632, "grad_norm": 2.2495028972625732, "learning_rate": 0.001, "loss": 1.4194, "step": 259800 }, { "epoch": 84.00129282482224, "grad_norm": 1.3998140096664429, "learning_rate": 0.001, "loss": 1.3954, "step": 259900 }, { "epoch": 84.03361344537815, "grad_norm": 2.64650559425354, "learning_rate": 0.001, "loss": 1.2849, "step": 260000 }, { "epoch": 84.06593406593407, "grad_norm": 1.8668577671051025, "learning_rate": 0.001, "loss": 1.2997, "step": 260100 }, { "epoch": 84.09825468648998, "grad_norm": 2.370088577270508, "learning_rate": 0.001, "loss": 1.2801, "step": 260200 }, { "epoch": 84.1305753070459, "grad_norm": 1.514963984489441, "learning_rate": 0.001, "loss": 1.3137, "step": 260300 }, { "epoch": 84.16289592760181, "grad_norm": 1.824122667312622, "learning_rate": 0.001, "loss": 1.2998, "step": 260400 }, { "epoch": 84.19521654815773, "grad_norm": 2.221118450164795, "learning_rate": 0.001, "loss": 1.3128, "step": 260500 }, { "epoch": 84.22753716871364, "grad_norm": 1.5105535984039307, "learning_rate": 0.001, "loss": 1.3176, "step": 260600 }, { "epoch": 84.25985778926956, "grad_norm": 1.4515846967697144, "learning_rate": 0.001, "loss": 1.3114, "step": 260700 }, { "epoch": 84.29217840982547, "grad_norm": 2.0730106830596924, "learning_rate": 0.001, "loss": 1.3138, "step": 260800 }, { "epoch": 84.32449903038139, "grad_norm": 1.670240044593811, "learning_rate": 0.001, "loss": 1.3265, "step": 260900 }, { "epoch": 84.3568196509373, "grad_norm": 1.5092387199401855, "learning_rate": 0.001, "loss": 1.349, "step": 261000 }, { "epoch": 84.38914027149322, "grad_norm": 1.756965160369873, "learning_rate": 0.001, "loss": 1.3311, "step": 261100 }, { "epoch": 84.42146089204913, "grad_norm": 1.6709240674972534, "learning_rate": 0.001, "loss": 1.3532, "step": 261200 }, { "epoch": 84.45378151260505, "grad_norm": 1.6208750009536743, "learning_rate": 0.001, "loss": 1.3453, "step": 261300 }, { "epoch": 84.48610213316095, "grad_norm": 1.4181149005889893, "learning_rate": 0.001, "loss": 1.3424, "step": 261400 }, { "epoch": 84.51842275371688, "grad_norm": 1.5968010425567627, "learning_rate": 0.001, "loss": 1.3502, "step": 261500 }, { "epoch": 84.55074337427278, "grad_norm": 1.5664920806884766, "learning_rate": 0.001, "loss": 1.3603, "step": 261600 }, { "epoch": 84.5830639948287, "grad_norm": 1.6634669303894043, "learning_rate": 0.001, "loss": 1.3668, "step": 261700 }, { "epoch": 84.61538461538461, "grad_norm": 1.7504676580429077, "learning_rate": 0.001, "loss": 1.356, "step": 261800 }, { "epoch": 84.64770523594053, "grad_norm": 1.6489694118499756, "learning_rate": 0.001, "loss": 1.351, "step": 261900 }, { "epoch": 84.68002585649644, "grad_norm": 1.3372224569320679, "learning_rate": 0.001, "loss": 1.3626, "step": 262000 }, { "epoch": 84.71234647705236, "grad_norm": 1.7510300874710083, "learning_rate": 0.001, "loss": 1.3782, "step": 262100 }, { "epoch": 84.74466709760827, "grad_norm": 2.0836756229400635, "learning_rate": 0.001, "loss": 1.3742, "step": 262200 }, { "epoch": 84.7769877181642, "grad_norm": 1.98588228225708, "learning_rate": 0.001, "loss": 1.3877, "step": 262300 }, { "epoch": 84.8093083387201, "grad_norm": 1.7963939905166626, "learning_rate": 0.001, "loss": 1.389, "step": 262400 }, { "epoch": 84.84162895927602, "grad_norm": 1.6637582778930664, "learning_rate": 0.001, "loss": 1.3757, "step": 262500 }, { "epoch": 84.87394957983193, "grad_norm": 1.9273699522018433, "learning_rate": 0.001, "loss": 1.3857, "step": 262600 }, { "epoch": 84.90627020038785, "grad_norm": 1.724212884902954, "learning_rate": 0.001, "loss": 1.3878, "step": 262700 }, { "epoch": 84.93859082094376, "grad_norm": 1.5962377786636353, "learning_rate": 0.001, "loss": 1.4054, "step": 262800 }, { "epoch": 84.97091144149968, "grad_norm": 1.804415225982666, "learning_rate": 0.001, "loss": 1.4049, "step": 262900 }, { "epoch": 85.00323206205559, "grad_norm": 1.5397160053253174, "learning_rate": 0.001, "loss": 1.3863, "step": 263000 }, { "epoch": 85.03555268261151, "grad_norm": 1.7321134805679321, "learning_rate": 0.001, "loss": 1.2727, "step": 263100 }, { "epoch": 85.06787330316742, "grad_norm": 1.2003074884414673, "learning_rate": 0.001, "loss": 1.2812, "step": 263200 }, { "epoch": 85.10019392372334, "grad_norm": 1.9206042289733887, "learning_rate": 0.001, "loss": 1.294, "step": 263300 }, { "epoch": 85.13251454427925, "grad_norm": 1.7302062511444092, "learning_rate": 0.001, "loss": 1.3028, "step": 263400 }, { "epoch": 85.16483516483517, "grad_norm": 1.5176318883895874, "learning_rate": 0.001, "loss": 1.305, "step": 263500 }, { "epoch": 85.19715578539108, "grad_norm": 1.221076488494873, "learning_rate": 0.001, "loss": 1.2955, "step": 263600 }, { "epoch": 85.229476405947, "grad_norm": 1.2028017044067383, "learning_rate": 0.001, "loss": 1.2969, "step": 263700 }, { "epoch": 85.2617970265029, "grad_norm": 1.3734420537948608, "learning_rate": 0.001, "loss": 1.3183, "step": 263800 }, { "epoch": 85.29411764705883, "grad_norm": 1.6204302310943604, "learning_rate": 0.001, "loss": 1.3096, "step": 263900 }, { "epoch": 85.32643826761473, "grad_norm": 1.4113692045211792, "learning_rate": 0.001, "loss": 1.324, "step": 264000 }, { "epoch": 85.35875888817066, "grad_norm": 1.7580782175064087, "learning_rate": 0.001, "loss": 1.3349, "step": 264100 }, { "epoch": 85.39107950872656, "grad_norm": 1.8114993572235107, "learning_rate": 0.001, "loss": 1.3232, "step": 264200 }, { "epoch": 85.42340012928248, "grad_norm": 1.8745627403259277, "learning_rate": 0.001, "loss": 1.3376, "step": 264300 }, { "epoch": 85.45572074983839, "grad_norm": 1.2741578817367554, "learning_rate": 0.001, "loss": 1.335, "step": 264400 }, { "epoch": 85.48804137039431, "grad_norm": 1.4233325719833374, "learning_rate": 0.001, "loss": 1.3441, "step": 264500 }, { "epoch": 85.52036199095022, "grad_norm": 1.3417024612426758, "learning_rate": 0.001, "loss": 1.3323, "step": 264600 }, { "epoch": 85.55268261150614, "grad_norm": 1.574867844581604, "learning_rate": 0.001, "loss": 1.3619, "step": 264700 }, { "epoch": 85.58500323206205, "grad_norm": 1.5219658613204956, "learning_rate": 0.001, "loss": 1.3517, "step": 264800 }, { "epoch": 85.61732385261797, "grad_norm": 1.6881141662597656, "learning_rate": 0.001, "loss": 1.3683, "step": 264900 }, { "epoch": 85.64964447317388, "grad_norm": 1.40277099609375, "learning_rate": 0.001, "loss": 1.3682, "step": 265000 }, { "epoch": 85.6819650937298, "grad_norm": 1.6496849060058594, "learning_rate": 0.001, "loss": 1.3553, "step": 265100 }, { "epoch": 85.71428571428571, "grad_norm": 1.9583226442337036, "learning_rate": 0.001, "loss": 1.3658, "step": 265200 }, { "epoch": 85.74660633484163, "grad_norm": 1.6086351871490479, "learning_rate": 0.001, "loss": 1.3623, "step": 265300 }, { "epoch": 85.77892695539754, "grad_norm": 1.4960424900054932, "learning_rate": 0.001, "loss": 1.3561, "step": 265400 }, { "epoch": 85.81124757595346, "grad_norm": 1.6357324123382568, "learning_rate": 0.001, "loss": 1.3876, "step": 265500 }, { "epoch": 85.84356819650937, "grad_norm": 1.3379572629928589, "learning_rate": 0.001, "loss": 1.3576, "step": 265600 }, { "epoch": 85.87588881706529, "grad_norm": 1.2775260210037231, "learning_rate": 0.001, "loss": 1.3735, "step": 265700 }, { "epoch": 85.9082094376212, "grad_norm": 1.6418566703796387, "learning_rate": 0.001, "loss": 1.384, "step": 265800 }, { "epoch": 85.94053005817712, "grad_norm": 1.4974275827407837, "learning_rate": 0.001, "loss": 1.39, "step": 265900 }, { "epoch": 85.97285067873302, "grad_norm": 1.4520913362503052, "learning_rate": 0.001, "loss": 1.3886, "step": 266000 }, { "epoch": 86.00517129928895, "grad_norm": 1.6784367561340332, "learning_rate": 0.001, "loss": 1.3867, "step": 266100 }, { "epoch": 86.03749191984487, "grad_norm": 1.85789155960083, "learning_rate": 0.001, "loss": 1.2681, "step": 266200 }, { "epoch": 86.06981254040078, "grad_norm": 2.0245468616485596, "learning_rate": 0.001, "loss": 1.2846, "step": 266300 }, { "epoch": 86.1021331609567, "grad_norm": 1.0982847213745117, "learning_rate": 0.001, "loss": 1.2784, "step": 266400 }, { "epoch": 86.1344537815126, "grad_norm": 1.1000640392303467, "learning_rate": 0.001, "loss": 1.296, "step": 266500 }, { "epoch": 86.16677440206853, "grad_norm": 2.0382845401763916, "learning_rate": 0.001, "loss": 1.2977, "step": 266600 }, { "epoch": 86.19909502262443, "grad_norm": 2.8659310340881348, "learning_rate": 0.001, "loss": 1.2818, "step": 266700 }, { "epoch": 86.23141564318036, "grad_norm": 1.5173022747039795, "learning_rate": 0.001, "loss": 1.3047, "step": 266800 }, { "epoch": 86.26373626373626, "grad_norm": 1.183382511138916, "learning_rate": 0.001, "loss": 1.3167, "step": 266900 }, { "epoch": 86.29605688429218, "grad_norm": 1.4100364446640015, "learning_rate": 0.001, "loss": 1.3181, "step": 267000 }, { "epoch": 86.32837750484809, "grad_norm": 1.3291791677474976, "learning_rate": 0.001, "loss": 1.3172, "step": 267100 }, { "epoch": 86.36069812540401, "grad_norm": 1.5253900289535522, "learning_rate": 0.001, "loss": 1.3138, "step": 267200 }, { "epoch": 86.39301874595992, "grad_norm": 1.2333720922470093, "learning_rate": 0.001, "loss": 1.3248, "step": 267300 }, { "epoch": 86.42533936651584, "grad_norm": 1.2060329914093018, "learning_rate": 0.001, "loss": 1.3278, "step": 267400 }, { "epoch": 86.45765998707175, "grad_norm": 1.940195918083191, "learning_rate": 0.001, "loss": 1.3232, "step": 267500 }, { "epoch": 86.48998060762767, "grad_norm": 1.408094882965088, "learning_rate": 0.001, "loss": 1.339, "step": 267600 }, { "epoch": 86.52230122818358, "grad_norm": 1.3757351636886597, "learning_rate": 0.001, "loss": 1.3378, "step": 267700 }, { "epoch": 86.5546218487395, "grad_norm": 1.6911505460739136, "learning_rate": 0.001, "loss": 1.3426, "step": 267800 }, { "epoch": 86.58694246929541, "grad_norm": 1.095239520072937, "learning_rate": 0.001, "loss": 1.3479, "step": 267900 }, { "epoch": 86.61926308985133, "grad_norm": 1.3571579456329346, "learning_rate": 0.001, "loss": 1.3418, "step": 268000 }, { "epoch": 86.65158371040724, "grad_norm": 1.5289149284362793, "learning_rate": 0.001, "loss": 1.3636, "step": 268100 }, { "epoch": 86.68390433096316, "grad_norm": 2.3517847061157227, "learning_rate": 0.001, "loss": 1.3577, "step": 268200 }, { "epoch": 86.71622495151907, "grad_norm": 1.7623791694641113, "learning_rate": 0.001, "loss": 1.3621, "step": 268300 }, { "epoch": 86.74854557207499, "grad_norm": 1.7347806692123413, "learning_rate": 0.001, "loss": 1.3591, "step": 268400 }, { "epoch": 86.7808661926309, "grad_norm": 1.9795016050338745, "learning_rate": 0.001, "loss": 1.3551, "step": 268500 }, { "epoch": 86.81318681318682, "grad_norm": 1.7498890161514282, "learning_rate": 0.001, "loss": 1.3696, "step": 268600 }, { "epoch": 86.84550743374272, "grad_norm": 1.118046522140503, "learning_rate": 0.001, "loss": 1.3541, "step": 268700 }, { "epoch": 86.87782805429865, "grad_norm": 1.7081708908081055, "learning_rate": 0.001, "loss": 1.3692, "step": 268800 }, { "epoch": 86.91014867485455, "grad_norm": 1.2541394233703613, "learning_rate": 0.001, "loss": 1.3738, "step": 268900 }, { "epoch": 86.94246929541048, "grad_norm": 1.55007803440094, "learning_rate": 0.001, "loss": 1.3728, "step": 269000 }, { "epoch": 86.97478991596638, "grad_norm": 1.1029446125030518, "learning_rate": 0.001, "loss": 1.3822, "step": 269100 }, { "epoch": 87.0071105365223, "grad_norm": 1.4424092769622803, "learning_rate": 0.001, "loss": 1.364, "step": 269200 }, { "epoch": 87.03943115707821, "grad_norm": 1.4126336574554443, "learning_rate": 0.001, "loss": 1.2606, "step": 269300 }, { "epoch": 87.07175177763413, "grad_norm": 1.1296864748001099, "learning_rate": 0.001, "loss": 1.2772, "step": 269400 }, { "epoch": 87.10407239819004, "grad_norm": 1.0883948802947998, "learning_rate": 0.001, "loss": 1.287, "step": 269500 }, { "epoch": 87.13639301874596, "grad_norm": 2.3048338890075684, "learning_rate": 0.001, "loss": 1.2915, "step": 269600 }, { "epoch": 87.16871363930187, "grad_norm": 1.4624378681182861, "learning_rate": 0.001, "loss": 1.2696, "step": 269700 }, { "epoch": 87.20103425985779, "grad_norm": 2.080893039703369, "learning_rate": 0.001, "loss": 1.2922, "step": 269800 }, { "epoch": 87.2333548804137, "grad_norm": 1.1155803203582764, "learning_rate": 0.001, "loss": 1.2881, "step": 269900 }, { "epoch": 87.26567550096962, "grad_norm": 2.0003838539123535, "learning_rate": 0.001, "loss": 1.3031, "step": 270000 }, { "epoch": 87.29799612152553, "grad_norm": 1.1450769901275635, "learning_rate": 0.001, "loss": 1.3158, "step": 270100 }, { "epoch": 87.33031674208145, "grad_norm": 1.6422492265701294, "learning_rate": 0.001, "loss": 1.3028, "step": 270200 }, { "epoch": 87.36263736263736, "grad_norm": 1.5813195705413818, "learning_rate": 0.001, "loss": 1.3043, "step": 270300 }, { "epoch": 87.39495798319328, "grad_norm": 2.072030782699585, "learning_rate": 0.001, "loss": 1.3095, "step": 270400 }, { "epoch": 87.42727860374919, "grad_norm": 1.496835470199585, "learning_rate": 0.001, "loss": 1.321, "step": 270500 }, { "epoch": 87.45959922430511, "grad_norm": 1.9745844602584839, "learning_rate": 0.001, "loss": 1.3344, "step": 270600 }, { "epoch": 87.49191984486102, "grad_norm": 1.3407188653945923, "learning_rate": 0.001, "loss": 1.3394, "step": 270700 }, { "epoch": 87.52424046541694, "grad_norm": 1.3977127075195312, "learning_rate": 0.001, "loss": 1.3158, "step": 270800 }, { "epoch": 87.55656108597285, "grad_norm": 1.1998000144958496, "learning_rate": 0.001, "loss": 1.3412, "step": 270900 }, { "epoch": 87.58888170652877, "grad_norm": 1.5857043266296387, "learning_rate": 0.001, "loss": 1.3398, "step": 271000 }, { "epoch": 87.62120232708467, "grad_norm": 1.3662618398666382, "learning_rate": 0.001, "loss": 1.3452, "step": 271100 }, { "epoch": 87.6535229476406, "grad_norm": 1.5763628482818604, "learning_rate": 0.001, "loss": 1.3347, "step": 271200 }, { "epoch": 87.6858435681965, "grad_norm": 1.1839090585708618, "learning_rate": 0.001, "loss": 1.354, "step": 271300 }, { "epoch": 87.71816418875243, "grad_norm": 1.1503945589065552, "learning_rate": 0.001, "loss": 1.351, "step": 271400 }, { "epoch": 87.75048480930833, "grad_norm": 1.364113688468933, "learning_rate": 0.001, "loss": 1.3464, "step": 271500 }, { "epoch": 87.78280542986425, "grad_norm": 1.640329122543335, "learning_rate": 0.001, "loss": 1.3523, "step": 271600 }, { "epoch": 87.81512605042016, "grad_norm": 1.3196192979812622, "learning_rate": 0.001, "loss": 1.3648, "step": 271700 }, { "epoch": 87.84744667097608, "grad_norm": 1.7141294479370117, "learning_rate": 0.001, "loss": 1.366, "step": 271800 }, { "epoch": 87.87976729153199, "grad_norm": 1.4717209339141846, "learning_rate": 0.001, "loss": 1.3733, "step": 271900 }, { "epoch": 87.91208791208791, "grad_norm": 1.763951063156128, "learning_rate": 0.001, "loss": 1.3805, "step": 272000 }, { "epoch": 87.94440853264382, "grad_norm": 2.454180955886841, "learning_rate": 0.001, "loss": 1.3777, "step": 272100 }, { "epoch": 87.97672915319974, "grad_norm": 1.2732707262039185, "learning_rate": 0.001, "loss": 1.3935, "step": 272200 }, { "epoch": 88.00904977375566, "grad_norm": 1.2473344802856445, "learning_rate": 0.001, "loss": 1.3515, "step": 272300 }, { "epoch": 88.04137039431157, "grad_norm": 1.722395896911621, "learning_rate": 0.001, "loss": 1.262, "step": 272400 }, { "epoch": 88.07369101486749, "grad_norm": 1.162625789642334, "learning_rate": 0.001, "loss": 1.2573, "step": 272500 }, { "epoch": 88.1060116354234, "grad_norm": 1.3701404333114624, "learning_rate": 0.001, "loss": 1.2692, "step": 272600 }, { "epoch": 88.13833225597932, "grad_norm": 1.517385721206665, "learning_rate": 0.001, "loss": 1.274, "step": 272700 }, { "epoch": 88.17065287653523, "grad_norm": 1.3972104787826538, "learning_rate": 0.001, "loss": 1.2774, "step": 272800 }, { "epoch": 88.20297349709115, "grad_norm": 1.48856520652771, "learning_rate": 0.001, "loss": 1.2776, "step": 272900 }, { "epoch": 88.23529411764706, "grad_norm": 1.5007410049438477, "learning_rate": 0.001, "loss": 1.289, "step": 273000 }, { "epoch": 88.26761473820298, "grad_norm": 1.6463698148727417, "learning_rate": 0.001, "loss": 1.2989, "step": 273100 }, { "epoch": 88.29993535875889, "grad_norm": 1.3843050003051758, "learning_rate": 0.001, "loss": 1.3107, "step": 273200 }, { "epoch": 88.33225597931481, "grad_norm": 1.6429800987243652, "learning_rate": 0.001, "loss": 1.3062, "step": 273300 }, { "epoch": 88.36457659987072, "grad_norm": 1.243739128112793, "learning_rate": 0.001, "loss": 1.3073, "step": 273400 }, { "epoch": 88.39689722042664, "grad_norm": 1.6305221319198608, "learning_rate": 0.001, "loss": 1.3134, "step": 273500 }, { "epoch": 88.42921784098255, "grad_norm": 1.9867161512374878, "learning_rate": 0.001, "loss": 1.3087, "step": 273600 }, { "epoch": 88.46153846153847, "grad_norm": 1.5061590671539307, "learning_rate": 0.001, "loss": 1.3178, "step": 273700 }, { "epoch": 88.49385908209437, "grad_norm": 1.2209558486938477, "learning_rate": 0.001, "loss": 1.3232, "step": 273800 }, { "epoch": 88.5261797026503, "grad_norm": 1.838896632194519, "learning_rate": 0.001, "loss": 1.3391, "step": 273900 }, { "epoch": 88.5585003232062, "grad_norm": 1.5392895936965942, "learning_rate": 0.001, "loss": 1.3292, "step": 274000 }, { "epoch": 88.59082094376213, "grad_norm": 1.872672438621521, "learning_rate": 0.001, "loss": 1.3353, "step": 274100 }, { "epoch": 88.62314156431803, "grad_norm": 1.6945205926895142, "learning_rate": 0.001, "loss": 1.3424, "step": 274200 }, { "epoch": 88.65546218487395, "grad_norm": 1.7650153636932373, "learning_rate": 0.001, "loss": 1.3442, "step": 274300 }, { "epoch": 88.68778280542986, "grad_norm": 1.4047414064407349, "learning_rate": 0.001, "loss": 1.3518, "step": 274400 }, { "epoch": 88.72010342598578, "grad_norm": 1.1727465391159058, "learning_rate": 0.001, "loss": 1.3535, "step": 274500 }, { "epoch": 88.75242404654169, "grad_norm": 1.6336814165115356, "learning_rate": 0.001, "loss": 1.3382, "step": 274600 }, { "epoch": 88.78474466709761, "grad_norm": 1.5113426446914673, "learning_rate": 0.001, "loss": 1.3513, "step": 274700 }, { "epoch": 88.81706528765352, "grad_norm": 1.4181830883026123, "learning_rate": 0.001, "loss": 1.3532, "step": 274800 }, { "epoch": 88.84938590820944, "grad_norm": 1.3083226680755615, "learning_rate": 0.001, "loss": 1.3592, "step": 274900 }, { "epoch": 88.88170652876535, "grad_norm": 1.3872716426849365, "learning_rate": 0.001, "loss": 1.3664, "step": 275000 }, { "epoch": 88.91402714932127, "grad_norm": 1.555129051208496, "learning_rate": 0.001, "loss": 1.364, "step": 275100 }, { "epoch": 88.94634776987718, "grad_norm": 2.2075741291046143, "learning_rate": 0.001, "loss": 1.3506, "step": 275200 }, { "epoch": 88.9786683904331, "grad_norm": 2.4672770500183105, "learning_rate": 0.001, "loss": 1.3606, "step": 275300 }, { "epoch": 89.01098901098901, "grad_norm": 1.3792235851287842, "learning_rate": 0.001, "loss": 1.3278, "step": 275400 }, { "epoch": 89.04330963154493, "grad_norm": 1.343404769897461, "learning_rate": 0.001, "loss": 1.2656, "step": 275500 }, { "epoch": 89.07563025210084, "grad_norm": 1.621608018875122, "learning_rate": 0.001, "loss": 1.2515, "step": 275600 }, { "epoch": 89.10795087265676, "grad_norm": 1.3299224376678467, "learning_rate": 0.001, "loss": 1.2787, "step": 275700 }, { "epoch": 89.14027149321267, "grad_norm": 1.6409183740615845, "learning_rate": 0.001, "loss": 1.2737, "step": 275800 }, { "epoch": 89.17259211376859, "grad_norm": 1.5294100046157837, "learning_rate": 0.001, "loss": 1.2814, "step": 275900 }, { "epoch": 89.2049127343245, "grad_norm": 1.9135456085205078, "learning_rate": 0.001, "loss": 1.2889, "step": 276000 }, { "epoch": 89.23723335488042, "grad_norm": 1.2390012741088867, "learning_rate": 0.001, "loss": 1.2958, "step": 276100 }, { "epoch": 89.26955397543632, "grad_norm": 1.4384171962738037, "learning_rate": 0.001, "loss": 1.288, "step": 276200 }, { "epoch": 89.30187459599225, "grad_norm": 1.1214848756790161, "learning_rate": 0.001, "loss": 1.28, "step": 276300 }, { "epoch": 89.33419521654815, "grad_norm": 1.3984450101852417, "learning_rate": 0.001, "loss": 1.2968, "step": 276400 }, { "epoch": 89.36651583710407, "grad_norm": 1.6903371810913086, "learning_rate": 0.001, "loss": 1.3195, "step": 276500 }, { "epoch": 89.39883645765998, "grad_norm": 1.9482558965682983, "learning_rate": 0.001, "loss": 1.3102, "step": 276600 }, { "epoch": 89.4311570782159, "grad_norm": 1.6029472351074219, "learning_rate": 0.001, "loss": 1.298, "step": 276700 }, { "epoch": 89.46347769877181, "grad_norm": 1.807937741279602, "learning_rate": 0.001, "loss": 1.311, "step": 276800 }, { "epoch": 89.49579831932773, "grad_norm": 1.967272400856018, "learning_rate": 0.001, "loss": 1.3206, "step": 276900 }, { "epoch": 89.52811893988364, "grad_norm": 1.4797887802124023, "learning_rate": 0.001, "loss": 1.325, "step": 277000 }, { "epoch": 89.56043956043956, "grad_norm": 1.414186954498291, "learning_rate": 0.001, "loss": 1.3176, "step": 277100 }, { "epoch": 89.59276018099547, "grad_norm": 1.4756308794021606, "learning_rate": 0.001, "loss": 1.322, "step": 277200 }, { "epoch": 89.62508080155139, "grad_norm": 1.391809105873108, "learning_rate": 0.001, "loss": 1.3396, "step": 277300 }, { "epoch": 89.6574014221073, "grad_norm": 1.6174664497375488, "learning_rate": 0.001, "loss": 1.3329, "step": 277400 }, { "epoch": 89.68972204266322, "grad_norm": 1.5664608478546143, "learning_rate": 0.001, "loss": 1.3236, "step": 277500 }, { "epoch": 89.72204266321913, "grad_norm": 1.3522831201553345, "learning_rate": 0.001, "loss": 1.346, "step": 277600 }, { "epoch": 89.75436328377505, "grad_norm": 1.4911127090454102, "learning_rate": 0.001, "loss": 1.3262, "step": 277700 }, { "epoch": 89.78668390433096, "grad_norm": 1.5883923768997192, "learning_rate": 0.001, "loss": 1.3321, "step": 277800 }, { "epoch": 89.81900452488688, "grad_norm": 1.4092780351638794, "learning_rate": 0.001, "loss": 1.3624, "step": 277900 }, { "epoch": 89.85132514544279, "grad_norm": 1.4824280738830566, "learning_rate": 0.001, "loss": 1.3345, "step": 278000 }, { "epoch": 89.88364576599871, "grad_norm": 1.2456746101379395, "learning_rate": 0.001, "loss": 1.3524, "step": 278100 }, { "epoch": 89.91596638655462, "grad_norm": 1.416329264640808, "learning_rate": 0.001, "loss": 1.3446, "step": 278200 }, { "epoch": 89.94828700711054, "grad_norm": 1.303393006324768, "learning_rate": 0.001, "loss": 1.3649, "step": 278300 }, { "epoch": 89.98060762766644, "grad_norm": 1.396873950958252, "learning_rate": 0.001, "loss": 1.3544, "step": 278400 }, { "epoch": 90.01292824822237, "grad_norm": 1.4742708206176758, "learning_rate": 0.001, "loss": 1.3035, "step": 278500 }, { "epoch": 90.04524886877829, "grad_norm": 1.7946443557739258, "learning_rate": 0.001, "loss": 1.2522, "step": 278600 }, { "epoch": 90.0775694893342, "grad_norm": 1.8204759359359741, "learning_rate": 0.001, "loss": 1.2583, "step": 278700 }, { "epoch": 90.10989010989012, "grad_norm": 1.6528264284133911, "learning_rate": 0.001, "loss": 1.2519, "step": 278800 }, { "epoch": 90.14221073044602, "grad_norm": 1.623073697090149, "learning_rate": 0.001, "loss": 1.2672, "step": 278900 }, { "epoch": 90.17453135100195, "grad_norm": 1.9576362371444702, "learning_rate": 0.001, "loss": 1.2641, "step": 279000 }, { "epoch": 90.20685197155785, "grad_norm": 1.9356460571289062, "learning_rate": 0.001, "loss": 1.2575, "step": 279100 }, { "epoch": 90.23917259211377, "grad_norm": 1.4619381427764893, "learning_rate": 0.001, "loss": 1.2932, "step": 279200 }, { "epoch": 90.27149321266968, "grad_norm": 1.5990549325942993, "learning_rate": 0.001, "loss": 1.2894, "step": 279300 }, { "epoch": 90.3038138332256, "grad_norm": 1.3064255714416504, "learning_rate": 0.001, "loss": 1.2767, "step": 279400 }, { "epoch": 90.33613445378151, "grad_norm": 1.3489830493927002, "learning_rate": 0.001, "loss": 1.289, "step": 279500 }, { "epoch": 90.36845507433743, "grad_norm": 1.441953420639038, "learning_rate": 0.001, "loss": 1.2965, "step": 279600 }, { "epoch": 90.40077569489334, "grad_norm": 1.8671915531158447, "learning_rate": 0.001, "loss": 1.3034, "step": 279700 }, { "epoch": 90.43309631544926, "grad_norm": 1.6628390550613403, "learning_rate": 0.001, "loss": 1.2943, "step": 279800 }, { "epoch": 90.46541693600517, "grad_norm": 1.5824347734451294, "learning_rate": 0.001, "loss": 1.3141, "step": 279900 }, { "epoch": 90.49773755656109, "grad_norm": 1.70791494846344, "learning_rate": 0.001, "loss": 1.3112, "step": 280000 }, { "epoch": 90.530058177117, "grad_norm": 2.0625507831573486, "learning_rate": 0.001, "loss": 1.3166, "step": 280100 }, { "epoch": 90.56237879767292, "grad_norm": 2.4934349060058594, "learning_rate": 0.001, "loss": 1.3166, "step": 280200 }, { "epoch": 90.59469941822883, "grad_norm": 1.3685812950134277, "learning_rate": 0.001, "loss": 1.317, "step": 280300 }, { "epoch": 90.62702003878475, "grad_norm": 1.607825517654419, "learning_rate": 0.001, "loss": 1.3241, "step": 280400 }, { "epoch": 90.65934065934066, "grad_norm": 1.5159963369369507, "learning_rate": 0.001, "loss": 1.3092, "step": 280500 }, { "epoch": 90.69166127989658, "grad_norm": 1.5785311460494995, "learning_rate": 0.001, "loss": 1.3321, "step": 280600 }, { "epoch": 90.72398190045249, "grad_norm": 1.2325959205627441, "learning_rate": 0.001, "loss": 1.3318, "step": 280700 }, { "epoch": 90.75630252100841, "grad_norm": 1.4121829271316528, "learning_rate": 0.001, "loss": 1.3351, "step": 280800 }, { "epoch": 90.78862314156432, "grad_norm": 1.5694589614868164, "learning_rate": 0.001, "loss": 1.3312, "step": 280900 }, { "epoch": 90.82094376212024, "grad_norm": 1.5602014064788818, "learning_rate": 0.001, "loss": 1.3473, "step": 281000 }, { "epoch": 90.85326438267614, "grad_norm": 1.5325652360916138, "learning_rate": 0.001, "loss": 1.3635, "step": 281100 }, { "epoch": 90.88558500323207, "grad_norm": 1.6036229133605957, "learning_rate": 0.001, "loss": 1.3579, "step": 281200 }, { "epoch": 90.91790562378797, "grad_norm": 1.6385998725891113, "learning_rate": 0.001, "loss": 1.3632, "step": 281300 }, { "epoch": 90.9502262443439, "grad_norm": 1.8205591440200806, "learning_rate": 0.001, "loss": 1.3462, "step": 281400 }, { "epoch": 90.9825468648998, "grad_norm": 1.7536708116531372, "learning_rate": 0.001, "loss": 1.3528, "step": 281500 }, { "epoch": 91.01486748545572, "grad_norm": 1.2094005346298218, "learning_rate": 0.001, "loss": 1.2857, "step": 281600 }, { "epoch": 91.04718810601163, "grad_norm": 2.219756841659546, "learning_rate": 0.001, "loss": 1.2417, "step": 281700 }, { "epoch": 91.07950872656755, "grad_norm": 2.2155497074127197, "learning_rate": 0.001, "loss": 1.2447, "step": 281800 }, { "epoch": 91.11182934712346, "grad_norm": 2.1346004009246826, "learning_rate": 0.001, "loss": 1.2614, "step": 281900 }, { "epoch": 91.14414996767938, "grad_norm": 1.6468212604522705, "learning_rate": 0.001, "loss": 1.2746, "step": 282000 }, { "epoch": 91.17647058823529, "grad_norm": 1.9895401000976562, "learning_rate": 0.001, "loss": 1.2707, "step": 282100 }, { "epoch": 91.20879120879121, "grad_norm": 2.380171298980713, "learning_rate": 0.001, "loss": 1.2691, "step": 282200 }, { "epoch": 91.24111182934712, "grad_norm": 1.7068277597427368, "learning_rate": 0.001, "loss": 1.271, "step": 282300 }, { "epoch": 91.27343244990304, "grad_norm": 1.8234671354293823, "learning_rate": 0.001, "loss": 1.2688, "step": 282400 }, { "epoch": 91.30575307045895, "grad_norm": 1.459957480430603, "learning_rate": 0.001, "loss": 1.2912, "step": 282500 }, { "epoch": 91.33807369101487, "grad_norm": 1.7018338441848755, "learning_rate": 0.001, "loss": 1.2881, "step": 282600 }, { "epoch": 91.37039431157078, "grad_norm": 1.862154483795166, "learning_rate": 0.001, "loss": 1.2976, "step": 282700 }, { "epoch": 91.4027149321267, "grad_norm": 1.876776933670044, "learning_rate": 0.001, "loss": 1.2975, "step": 282800 }, { "epoch": 91.4350355526826, "grad_norm": 2.4025259017944336, "learning_rate": 0.001, "loss": 1.2911, "step": 282900 }, { "epoch": 91.46735617323853, "grad_norm": 1.6050946712493896, "learning_rate": 0.001, "loss": 1.297, "step": 283000 }, { "epoch": 91.49967679379444, "grad_norm": 1.7577511072158813, "learning_rate": 0.001, "loss": 1.311, "step": 283100 }, { "epoch": 91.53199741435036, "grad_norm": 1.6006762981414795, "learning_rate": 0.001, "loss": 1.3057, "step": 283200 }, { "epoch": 91.56431803490626, "grad_norm": 1.7033071517944336, "learning_rate": 0.001, "loss": 1.3044, "step": 283300 }, { "epoch": 91.59663865546219, "grad_norm": 1.6848423480987549, "learning_rate": 0.001, "loss": 1.3219, "step": 283400 }, { "epoch": 91.6289592760181, "grad_norm": 1.886040449142456, "learning_rate": 0.001, "loss": 1.314, "step": 283500 }, { "epoch": 91.66127989657402, "grad_norm": 1.5038381814956665, "learning_rate": 0.001, "loss": 1.3306, "step": 283600 }, { "epoch": 91.69360051712992, "grad_norm": 1.553776741027832, "learning_rate": 0.001, "loss": 1.3221, "step": 283700 }, { "epoch": 91.72592113768584, "grad_norm": 1.759239912033081, "learning_rate": 0.001, "loss": 1.3269, "step": 283800 }, { "epoch": 91.75824175824175, "grad_norm": 2.047476291656494, "learning_rate": 0.001, "loss": 1.3342, "step": 283900 }, { "epoch": 91.79056237879767, "grad_norm": 1.8062100410461426, "learning_rate": 0.001, "loss": 1.3431, "step": 284000 }, { "epoch": 91.82288299935358, "grad_norm": 1.7818137407302856, "learning_rate": 0.001, "loss": 1.3331, "step": 284100 }, { "epoch": 91.8552036199095, "grad_norm": 1.8032774925231934, "learning_rate": 0.001, "loss": 1.3454, "step": 284200 }, { "epoch": 91.88752424046541, "grad_norm": 1.5718131065368652, "learning_rate": 0.001, "loss": 1.336, "step": 284300 }, { "epoch": 91.91984486102133, "grad_norm": 1.789857029914856, "learning_rate": 0.001, "loss": 1.3349, "step": 284400 }, { "epoch": 91.95216548157724, "grad_norm": 1.8243910074234009, "learning_rate": 0.001, "loss": 1.347, "step": 284500 }, { "epoch": 91.98448610213316, "grad_norm": 2.6804580688476562, "learning_rate": 0.001, "loss": 1.3379, "step": 284600 }, { "epoch": 92.01680672268908, "grad_norm": 1.7941631078720093, "learning_rate": 0.001, "loss": 1.2708, "step": 284700 }, { "epoch": 92.04912734324499, "grad_norm": 1.7584716081619263, "learning_rate": 0.001, "loss": 1.2498, "step": 284800 }, { "epoch": 92.08144796380091, "grad_norm": 1.6040362119674683, "learning_rate": 0.001, "loss": 1.2461, "step": 284900 }, { "epoch": 92.11376858435682, "grad_norm": 1.6586319208145142, "learning_rate": 0.001, "loss": 1.2484, "step": 285000 }, { "epoch": 92.14608920491274, "grad_norm": 1.869040846824646, "learning_rate": 0.001, "loss": 1.2543, "step": 285100 }, { "epoch": 92.17840982546865, "grad_norm": 1.9890912771224976, "learning_rate": 0.001, "loss": 1.2621, "step": 285200 }, { "epoch": 92.21073044602457, "grad_norm": 1.7279739379882812, "learning_rate": 0.001, "loss": 1.2587, "step": 285300 }, { "epoch": 92.24305106658048, "grad_norm": 1.527393102645874, "learning_rate": 0.001, "loss": 1.2656, "step": 285400 }, { "epoch": 92.2753716871364, "grad_norm": 2.0269243717193604, "learning_rate": 0.001, "loss": 1.2876, "step": 285500 }, { "epoch": 92.3076923076923, "grad_norm": 2.9750266075134277, "learning_rate": 0.001, "loss": 1.2761, "step": 285600 }, { "epoch": 92.34001292824823, "grad_norm": 1.699476718902588, "learning_rate": 0.001, "loss": 1.279, "step": 285700 }, { "epoch": 92.37233354880414, "grad_norm": 1.8639898300170898, "learning_rate": 0.001, "loss": 1.2823, "step": 285800 }, { "epoch": 92.40465416936006, "grad_norm": 1.3638863563537598, "learning_rate": 0.001, "loss": 1.284, "step": 285900 }, { "epoch": 92.43697478991596, "grad_norm": 1.5757185220718384, "learning_rate": 0.001, "loss": 1.2923, "step": 286000 }, { "epoch": 92.46929541047189, "grad_norm": 1.8936587572097778, "learning_rate": 0.001, "loss": 1.3045, "step": 286100 }, { "epoch": 92.5016160310278, "grad_norm": 2.0886335372924805, "learning_rate": 0.001, "loss": 1.2932, "step": 286200 }, { "epoch": 92.53393665158372, "grad_norm": 2.6820244789123535, "learning_rate": 0.001, "loss": 1.3142, "step": 286300 }, { "epoch": 92.56625727213962, "grad_norm": 1.8949545621871948, "learning_rate": 0.001, "loss": 1.3036, "step": 286400 }, { "epoch": 92.59857789269554, "grad_norm": 1.7866016626358032, "learning_rate": 0.001, "loss": 1.3071, "step": 286500 }, { "epoch": 92.63089851325145, "grad_norm": 2.0061726570129395, "learning_rate": 0.001, "loss": 1.3092, "step": 286600 }, { "epoch": 92.66321913380737, "grad_norm": 1.950553059577942, "learning_rate": 0.001, "loss": 1.3201, "step": 286700 }, { "epoch": 92.69553975436328, "grad_norm": 2.6318914890289307, "learning_rate": 0.001, "loss": 1.3117, "step": 286800 }, { "epoch": 92.7278603749192, "grad_norm": 1.6457170248031616, "learning_rate": 0.001, "loss": 1.3083, "step": 286900 }, { "epoch": 92.76018099547511, "grad_norm": 2.6424992084503174, "learning_rate": 0.001, "loss": 1.3334, "step": 287000 }, { "epoch": 92.79250161603103, "grad_norm": 1.5322871208190918, "learning_rate": 0.001, "loss": 1.3171, "step": 287100 }, { "epoch": 92.82482223658694, "grad_norm": 1.9201703071594238, "learning_rate": 0.001, "loss": 1.3377, "step": 287200 }, { "epoch": 92.85714285714286, "grad_norm": 2.1163434982299805, "learning_rate": 0.001, "loss": 1.333, "step": 287300 }, { "epoch": 92.88946347769877, "grad_norm": 1.438014268875122, "learning_rate": 0.001, "loss": 1.3275, "step": 287400 }, { "epoch": 92.92178409825469, "grad_norm": 1.786887764930725, "learning_rate": 0.001, "loss": 1.3299, "step": 287500 }, { "epoch": 92.9541047188106, "grad_norm": 1.8135703802108765, "learning_rate": 0.001, "loss": 1.3392, "step": 287600 }, { "epoch": 92.98642533936652, "grad_norm": 1.705079436302185, "learning_rate": 0.001, "loss": 1.3478, "step": 287700 }, { "epoch": 93.01874595992243, "grad_norm": 2.3613109588623047, "learning_rate": 0.001, "loss": 1.2827, "step": 287800 }, { "epoch": 93.05106658047835, "grad_norm": 1.4064383506774902, "learning_rate": 0.001, "loss": 1.2314, "step": 287900 }, { "epoch": 93.08338720103426, "grad_norm": 1.7855405807495117, "learning_rate": 0.001, "loss": 1.2275, "step": 288000 }, { "epoch": 93.11570782159018, "grad_norm": 1.2022162675857544, "learning_rate": 0.001, "loss": 1.2441, "step": 288100 }, { "epoch": 93.14802844214609, "grad_norm": 1.501390814781189, "learning_rate": 0.001, "loss": 1.2495, "step": 288200 }, { "epoch": 93.180349062702, "grad_norm": 1.2558820247650146, "learning_rate": 0.001, "loss": 1.2427, "step": 288300 }, { "epoch": 93.21266968325791, "grad_norm": 1.2189136743545532, "learning_rate": 0.001, "loss": 1.2495, "step": 288400 }, { "epoch": 93.24499030381384, "grad_norm": 1.3827097415924072, "learning_rate": 0.001, "loss": 1.249, "step": 288500 }, { "epoch": 93.27731092436974, "grad_norm": 1.5776780843734741, "learning_rate": 0.001, "loss": 1.2677, "step": 288600 }, { "epoch": 93.30963154492567, "grad_norm": 1.501815915107727, "learning_rate": 0.001, "loss": 1.2673, "step": 288700 }, { "epoch": 93.34195216548157, "grad_norm": 1.7654236555099487, "learning_rate": 0.001, "loss": 1.2822, "step": 288800 }, { "epoch": 93.3742727860375, "grad_norm": 1.8030627965927124, "learning_rate": 0.001, "loss": 1.2816, "step": 288900 }, { "epoch": 93.4065934065934, "grad_norm": 1.3551745414733887, "learning_rate": 0.001, "loss": 1.2697, "step": 289000 }, { "epoch": 93.43891402714932, "grad_norm": 1.3812977075576782, "learning_rate": 0.001, "loss": 1.2881, "step": 289100 }, { "epoch": 93.47123464770523, "grad_norm": 1.2390937805175781, "learning_rate": 0.001, "loss": 1.2882, "step": 289200 }, { "epoch": 93.50355526826115, "grad_norm": 1.3505865335464478, "learning_rate": 0.001, "loss": 1.2881, "step": 289300 }, { "epoch": 93.53587588881706, "grad_norm": 1.8913111686706543, "learning_rate": 0.001, "loss": 1.3037, "step": 289400 }, { "epoch": 93.56819650937298, "grad_norm": 2.1353416442871094, "learning_rate": 0.001, "loss": 1.3182, "step": 289500 }, { "epoch": 93.60051712992889, "grad_norm": 1.4333372116088867, "learning_rate": 0.001, "loss": 1.3018, "step": 289600 }, { "epoch": 93.63283775048481, "grad_norm": 1.4343737363815308, "learning_rate": 0.001, "loss": 1.2958, "step": 289700 }, { "epoch": 93.66515837104072, "grad_norm": 1.8866076469421387, "learning_rate": 0.001, "loss": 1.3095, "step": 289800 }, { "epoch": 93.69747899159664, "grad_norm": 1.802651047706604, "learning_rate": 0.001, "loss": 1.3296, "step": 289900 }, { "epoch": 93.72979961215255, "grad_norm": 1.3045099973678589, "learning_rate": 0.001, "loss": 1.3177, "step": 290000 }, { "epoch": 93.76212023270847, "grad_norm": 1.24888277053833, "learning_rate": 0.001, "loss": 1.3263, "step": 290100 }, { "epoch": 93.79444085326438, "grad_norm": 1.5360679626464844, "learning_rate": 0.001, "loss": 1.3371, "step": 290200 }, { "epoch": 93.8267614738203, "grad_norm": 1.491134524345398, "learning_rate": 0.001, "loss": 1.3337, "step": 290300 }, { "epoch": 93.8590820943762, "grad_norm": 1.2645684480667114, "learning_rate": 0.001, "loss": 1.3272, "step": 290400 }, { "epoch": 93.89140271493213, "grad_norm": 1.7189093828201294, "learning_rate": 0.001, "loss": 1.331, "step": 290500 }, { "epoch": 93.92372333548803, "grad_norm": 1.4772014617919922, "learning_rate": 0.001, "loss": 1.3186, "step": 290600 }, { "epoch": 93.95604395604396, "grad_norm": 1.5750775337219238, "learning_rate": 0.001, "loss": 1.3494, "step": 290700 }, { "epoch": 93.98836457659988, "grad_norm": 1.560361623764038, "learning_rate": 0.001, "loss": 1.3308, "step": 290800 }, { "epoch": 94.02068519715579, "grad_norm": 1.4874534606933594, "learning_rate": 0.001, "loss": 1.2771, "step": 290900 }, { "epoch": 94.0530058177117, "grad_norm": 1.5172135829925537, "learning_rate": 0.001, "loss": 1.2225, "step": 291000 }, { "epoch": 94.08532643826761, "grad_norm": 1.3056913614273071, "learning_rate": 0.001, "loss": 1.2375, "step": 291100 }, { "epoch": 94.11764705882354, "grad_norm": 1.163444995880127, "learning_rate": 0.001, "loss": 1.2395, "step": 291200 }, { "epoch": 94.14996767937944, "grad_norm": 1.3650455474853516, "learning_rate": 0.001, "loss": 1.2509, "step": 291300 }, { "epoch": 94.18228829993537, "grad_norm": 1.5594156980514526, "learning_rate": 0.001, "loss": 1.2378, "step": 291400 }, { "epoch": 94.21460892049127, "grad_norm": 1.3483871221542358, "learning_rate": 0.001, "loss": 1.2524, "step": 291500 }, { "epoch": 94.2469295410472, "grad_norm": 1.5659109354019165, "learning_rate": 0.001, "loss": 1.2509, "step": 291600 }, { "epoch": 94.2792501616031, "grad_norm": 2.272768497467041, "learning_rate": 0.001, "loss": 1.2641, "step": 291700 }, { "epoch": 94.31157078215902, "grad_norm": 1.498880386352539, "learning_rate": 0.001, "loss": 1.2594, "step": 291800 }, { "epoch": 94.34389140271493, "grad_norm": 3.030276298522949, "learning_rate": 0.001, "loss": 1.2609, "step": 291900 }, { "epoch": 94.37621202327085, "grad_norm": 1.6324968338012695, "learning_rate": 0.001, "loss": 1.2594, "step": 292000 }, { "epoch": 94.40853264382676, "grad_norm": 1.801637053489685, "learning_rate": 0.001, "loss": 1.276, "step": 292100 }, { "epoch": 94.44085326438268, "grad_norm": 2.261122941970825, "learning_rate": 0.001, "loss": 1.2843, "step": 292200 }, { "epoch": 94.47317388493859, "grad_norm": 2.3047244548797607, "learning_rate": 0.001, "loss": 1.2973, "step": 292300 }, { "epoch": 94.50549450549451, "grad_norm": 2.708813428878784, "learning_rate": 0.001, "loss": 1.2903, "step": 292400 }, { "epoch": 94.53781512605042, "grad_norm": 2.372790575027466, "learning_rate": 0.001, "loss": 1.2972, "step": 292500 }, { "epoch": 94.57013574660634, "grad_norm": 1.5244978666305542, "learning_rate": 0.001, "loss": 1.2814, "step": 292600 }, { "epoch": 94.60245636716225, "grad_norm": 2.138533592224121, "learning_rate": 0.001, "loss": 1.2886, "step": 292700 }, { "epoch": 94.63477698771817, "grad_norm": 1.3166296482086182, "learning_rate": 0.001, "loss": 1.2948, "step": 292800 }, { "epoch": 94.66709760827408, "grad_norm": 1.5935583114624023, "learning_rate": 0.001, "loss": 1.3114, "step": 292900 }, { "epoch": 94.69941822883, "grad_norm": 1.9687855243682861, "learning_rate": 0.001, "loss": 1.3064, "step": 293000 }, { "epoch": 94.7317388493859, "grad_norm": 1.6364773511886597, "learning_rate": 0.001, "loss": 1.3186, "step": 293100 }, { "epoch": 94.76405946994183, "grad_norm": 1.3022596836090088, "learning_rate": 0.001, "loss": 1.3239, "step": 293200 }, { "epoch": 94.79638009049773, "grad_norm": 1.4766348600387573, "learning_rate": 0.001, "loss": 1.3321, "step": 293300 }, { "epoch": 94.82870071105366, "grad_norm": 2.4459609985351562, "learning_rate": 0.001, "loss": 1.3089, "step": 293400 }, { "epoch": 94.86102133160956, "grad_norm": 1.246825098991394, "learning_rate": 0.001, "loss": 1.3026, "step": 293500 }, { "epoch": 94.89334195216549, "grad_norm": 2.1387414932250977, "learning_rate": 0.001, "loss": 1.3201, "step": 293600 }, { "epoch": 94.9256625727214, "grad_norm": 1.2364811897277832, "learning_rate": 0.001, "loss": 1.3298, "step": 293700 }, { "epoch": 94.95798319327731, "grad_norm": 1.3467936515808105, "learning_rate": 0.001, "loss": 1.3386, "step": 293800 }, { "epoch": 94.99030381383322, "grad_norm": 1.5889116525650024, "learning_rate": 0.001, "loss": 1.3304, "step": 293900 }, { "epoch": 95.02262443438914, "grad_norm": 1.1252005100250244, "learning_rate": 0.001, "loss": 1.2648, "step": 294000 }, { "epoch": 95.05494505494505, "grad_norm": 1.8917295932769775, "learning_rate": 0.001, "loss": 1.2281, "step": 294100 }, { "epoch": 95.08726567550097, "grad_norm": 1.2165217399597168, "learning_rate": 0.001, "loss": 1.2275, "step": 294200 }, { "epoch": 95.11958629605688, "grad_norm": 1.483284831047058, "learning_rate": 0.001, "loss": 1.2294, "step": 294300 }, { "epoch": 95.1519069166128, "grad_norm": 1.5502418279647827, "learning_rate": 0.001, "loss": 1.2497, "step": 294400 }, { "epoch": 95.18422753716871, "grad_norm": 1.3309822082519531, "learning_rate": 0.001, "loss": 1.2413, "step": 294500 }, { "epoch": 95.21654815772463, "grad_norm": 1.3194745779037476, "learning_rate": 0.001, "loss": 1.2473, "step": 294600 }, { "epoch": 95.24886877828054, "grad_norm": 1.3569340705871582, "learning_rate": 0.001, "loss": 1.2371, "step": 294700 }, { "epoch": 95.28118939883646, "grad_norm": 1.3349523544311523, "learning_rate": 0.001, "loss": 1.2471, "step": 294800 }, { "epoch": 95.31351001939237, "grad_norm": 2.0123939514160156, "learning_rate": 0.001, "loss": 1.2701, "step": 294900 }, { "epoch": 95.34583063994829, "grad_norm": 1.622585654258728, "learning_rate": 0.001, "loss": 1.2563, "step": 295000 }, { "epoch": 95.3781512605042, "grad_norm": 1.2212409973144531, "learning_rate": 0.001, "loss": 1.2853, "step": 295100 }, { "epoch": 95.41047188106012, "grad_norm": 1.120845079421997, "learning_rate": 0.001, "loss": 1.2622, "step": 295200 }, { "epoch": 95.44279250161603, "grad_norm": 1.106087565422058, "learning_rate": 0.001, "loss": 1.2759, "step": 295300 }, { "epoch": 95.47511312217195, "grad_norm": 1.985410213470459, "learning_rate": 0.001, "loss": 1.2846, "step": 295400 }, { "epoch": 95.50743374272786, "grad_norm": 1.1984760761260986, "learning_rate": 0.001, "loss": 1.2756, "step": 295500 }, { "epoch": 95.53975436328378, "grad_norm": 1.565770149230957, "learning_rate": 0.001, "loss": 1.2914, "step": 295600 }, { "epoch": 95.57207498383968, "grad_norm": 1.3624826669692993, "learning_rate": 0.001, "loss": 1.2835, "step": 295700 }, { "epoch": 95.6043956043956, "grad_norm": 1.0454158782958984, "learning_rate": 0.001, "loss": 1.287, "step": 295800 }, { "epoch": 95.63671622495151, "grad_norm": 1.0237979888916016, "learning_rate": 0.001, "loss": 1.2997, "step": 295900 }, { "epoch": 95.66903684550743, "grad_norm": 1.2886712551116943, "learning_rate": 0.001, "loss": 1.2895, "step": 296000 }, { "epoch": 95.70135746606334, "grad_norm": 1.6183465719223022, "learning_rate": 0.001, "loss": 1.297, "step": 296100 }, { "epoch": 95.73367808661926, "grad_norm": 1.7799890041351318, "learning_rate": 0.001, "loss": 1.3121, "step": 296200 }, { "epoch": 95.76599870717517, "grad_norm": 1.240616798400879, "learning_rate": 0.001, "loss": 1.2888, "step": 296300 }, { "epoch": 95.7983193277311, "grad_norm": 1.814924955368042, "learning_rate": 0.001, "loss": 1.3176, "step": 296400 }, { "epoch": 95.830639948287, "grad_norm": 1.1921446323394775, "learning_rate": 0.001, "loss": 1.3089, "step": 296500 }, { "epoch": 95.86296056884292, "grad_norm": 1.1958500146865845, "learning_rate": 0.001, "loss": 1.3175, "step": 296600 }, { "epoch": 95.89528118939883, "grad_norm": 1.388808250427246, "learning_rate": 0.001, "loss": 1.3137, "step": 296700 }, { "epoch": 95.92760180995475, "grad_norm": 1.6556081771850586, "learning_rate": 0.001, "loss": 1.324, "step": 296800 }, { "epoch": 95.95992243051066, "grad_norm": 1.1369446516036987, "learning_rate": 0.001, "loss": 1.3176, "step": 296900 }, { "epoch": 95.99224305106658, "grad_norm": 1.4303462505340576, "learning_rate": 0.001, "loss": 1.3213, "step": 297000 }, { "epoch": 96.0245636716225, "grad_norm": 1.7964837551116943, "learning_rate": 0.001, "loss": 1.2639, "step": 297100 }, { "epoch": 96.05688429217841, "grad_norm": 2.2737128734588623, "learning_rate": 0.001, "loss": 1.2169, "step": 297200 }, { "epoch": 96.08920491273433, "grad_norm": 1.316617727279663, "learning_rate": 0.001, "loss": 1.2285, "step": 297300 }, { "epoch": 96.12152553329024, "grad_norm": 1.4206103086471558, "learning_rate": 0.001, "loss": 1.2253, "step": 297400 }, { "epoch": 96.15384615384616, "grad_norm": 1.3422598838806152, "learning_rate": 0.001, "loss": 1.2307, "step": 297500 }, { "epoch": 96.18616677440207, "grad_norm": 1.8675673007965088, "learning_rate": 0.001, "loss": 1.2436, "step": 297600 }, { "epoch": 96.21848739495799, "grad_norm": 1.1970405578613281, "learning_rate": 0.001, "loss": 1.2485, "step": 297700 }, { "epoch": 96.2508080155139, "grad_norm": 1.6834607124328613, "learning_rate": 0.001, "loss": 1.2387, "step": 297800 }, { "epoch": 96.28312863606982, "grad_norm": 1.3664674758911133, "learning_rate": 0.001, "loss": 1.2287, "step": 297900 }, { "epoch": 96.31544925662573, "grad_norm": 1.3463797569274902, "learning_rate": 0.001, "loss": 1.2583, "step": 298000 }, { "epoch": 96.34776987718165, "grad_norm": 1.318743109703064, "learning_rate": 0.001, "loss": 1.2707, "step": 298100 }, { "epoch": 96.38009049773756, "grad_norm": 1.703060269355774, "learning_rate": 0.001, "loss": 1.2761, "step": 298200 }, { "epoch": 96.41241111829348, "grad_norm": 1.5728152990341187, "learning_rate": 0.001, "loss": 1.2626, "step": 298300 }, { "epoch": 96.44473173884938, "grad_norm": 1.3078418970108032, "learning_rate": 0.001, "loss": 1.2598, "step": 298400 }, { "epoch": 96.4770523594053, "grad_norm": 1.6507360935211182, "learning_rate": 0.001, "loss": 1.2679, "step": 298500 }, { "epoch": 96.50937297996121, "grad_norm": 1.6421030759811401, "learning_rate": 0.001, "loss": 1.277, "step": 298600 }, { "epoch": 96.54169360051714, "grad_norm": 1.5586189031600952, "learning_rate": 0.001, "loss": 1.2862, "step": 298700 }, { "epoch": 96.57401422107304, "grad_norm": 1.9439473152160645, "learning_rate": 0.001, "loss": 1.283, "step": 298800 }, { "epoch": 96.60633484162896, "grad_norm": 1.1835355758666992, "learning_rate": 0.001, "loss": 1.2904, "step": 298900 }, { "epoch": 96.63865546218487, "grad_norm": 2.370730400085449, "learning_rate": 0.001, "loss": 1.2695, "step": 299000 }, { "epoch": 96.6709760827408, "grad_norm": 1.409196376800537, "learning_rate": 0.001, "loss": 1.2858, "step": 299100 }, { "epoch": 96.7032967032967, "grad_norm": 1.6037887334823608, "learning_rate": 0.001, "loss": 1.2972, "step": 299200 }, { "epoch": 96.73561732385262, "grad_norm": 1.8080615997314453, "learning_rate": 0.001, "loss": 1.2963, "step": 299300 }, { "epoch": 96.76793794440853, "grad_norm": 1.3899056911468506, "learning_rate": 0.001, "loss": 1.3146, "step": 299400 }, { "epoch": 96.80025856496445, "grad_norm": 1.2540109157562256, "learning_rate": 0.001, "loss": 1.3093, "step": 299500 }, { "epoch": 96.83257918552036, "grad_norm": 1.2840240001678467, "learning_rate": 0.001, "loss": 1.3132, "step": 299600 }, { "epoch": 96.86489980607628, "grad_norm": 2.769483804702759, "learning_rate": 0.001, "loss": 1.3173, "step": 299700 }, { "epoch": 96.89722042663219, "grad_norm": 1.460633397102356, "learning_rate": 0.001, "loss": 1.3161, "step": 299800 }, { "epoch": 96.92954104718811, "grad_norm": 1.941650390625, "learning_rate": 0.001, "loss": 1.3029, "step": 299900 }, { "epoch": 96.96186166774402, "grad_norm": 1.4547431468963623, "learning_rate": 0.001, "loss": 1.3191, "step": 300000 }, { "epoch": 96.99418228829994, "grad_norm": 1.1949020624160767, "learning_rate": 0.001, "loss": 1.3033, "step": 300100 }, { "epoch": 97.02650290885585, "grad_norm": 1.4821178913116455, "learning_rate": 0.001, "loss": 1.2344, "step": 300200 }, { "epoch": 97.05882352941177, "grad_norm": 2.7771239280700684, "learning_rate": 0.001, "loss": 1.2183, "step": 300300 }, { "epoch": 97.09114414996768, "grad_norm": 1.1940053701400757, "learning_rate": 0.001, "loss": 1.218, "step": 300400 }, { "epoch": 97.1234647705236, "grad_norm": 1.2287544012069702, "learning_rate": 0.001, "loss": 1.2185, "step": 300500 }, { "epoch": 97.1557853910795, "grad_norm": 1.2844361066818237, "learning_rate": 0.001, "loss": 1.2334, "step": 300600 }, { "epoch": 97.18810601163543, "grad_norm": 1.7986161708831787, "learning_rate": 0.001, "loss": 1.2308, "step": 300700 }, { "epoch": 97.22042663219133, "grad_norm": 2.0331642627716064, "learning_rate": 0.001, "loss": 1.2359, "step": 300800 }, { "epoch": 97.25274725274726, "grad_norm": 1.260055422782898, "learning_rate": 0.001, "loss": 1.2452, "step": 300900 }, { "epoch": 97.28506787330316, "grad_norm": 1.5814372301101685, "learning_rate": 0.001, "loss": 1.2561, "step": 301000 }, { "epoch": 97.31738849385908, "grad_norm": 1.2864145040512085, "learning_rate": 0.001, "loss": 1.2641, "step": 301100 }, { "epoch": 97.34970911441499, "grad_norm": 1.138455867767334, "learning_rate": 0.001, "loss": 1.2574, "step": 301200 }, { "epoch": 97.38202973497091, "grad_norm": 1.9856462478637695, "learning_rate": 0.001, "loss": 1.2631, "step": 301300 }, { "epoch": 97.41435035552682, "grad_norm": 1.3746229410171509, "learning_rate": 0.001, "loss": 1.2537, "step": 301400 }, { "epoch": 97.44667097608274, "grad_norm": 1.5862191915512085, "learning_rate": 0.001, "loss": 1.2684, "step": 301500 }, { "epoch": 97.47899159663865, "grad_norm": 2.5958855152130127, "learning_rate": 0.001, "loss": 1.2607, "step": 301600 }, { "epoch": 97.51131221719457, "grad_norm": 1.5952675342559814, "learning_rate": 0.001, "loss": 1.2868, "step": 301700 }, { "epoch": 97.54363283775048, "grad_norm": 1.5219841003417969, "learning_rate": 0.001, "loss": 1.2769, "step": 301800 }, { "epoch": 97.5759534583064, "grad_norm": 1.7715810537338257, "learning_rate": 0.001, "loss": 1.2765, "step": 301900 }, { "epoch": 97.60827407886231, "grad_norm": 1.4715001583099365, "learning_rate": 0.001, "loss": 1.2896, "step": 302000 }, { "epoch": 97.64059469941823, "grad_norm": 1.7011549472808838, "learning_rate": 0.001, "loss": 1.2806, "step": 302100 }, { "epoch": 97.67291531997414, "grad_norm": 1.7269728183746338, "learning_rate": 0.001, "loss": 1.2731, "step": 302200 }, { "epoch": 97.70523594053006, "grad_norm": 1.3083314895629883, "learning_rate": 0.001, "loss": 1.2806, "step": 302300 }, { "epoch": 97.73755656108597, "grad_norm": 1.3248363733291626, "learning_rate": 0.001, "loss": 1.2924, "step": 302400 }, { "epoch": 97.76987718164189, "grad_norm": 1.6495850086212158, "learning_rate": 0.001, "loss": 1.3018, "step": 302500 }, { "epoch": 97.8021978021978, "grad_norm": 1.550723910331726, "learning_rate": 0.001, "loss": 1.3067, "step": 302600 }, { "epoch": 97.83451842275372, "grad_norm": 1.4989489316940308, "learning_rate": 0.001, "loss": 1.3016, "step": 302700 }, { "epoch": 97.86683904330962, "grad_norm": 1.403679370880127, "learning_rate": 0.001, "loss": 1.3039, "step": 302800 }, { "epoch": 97.89915966386555, "grad_norm": 1.3110108375549316, "learning_rate": 0.001, "loss": 1.2979, "step": 302900 }, { "epoch": 97.93148028442145, "grad_norm": 1.416511058807373, "learning_rate": 0.001, "loss": 1.3202, "step": 303000 }, { "epoch": 97.96380090497738, "grad_norm": 2.252964735031128, "learning_rate": 0.001, "loss": 1.3163, "step": 303100 }, { "epoch": 97.99612152553328, "grad_norm": 1.4983501434326172, "learning_rate": 0.001, "loss": 1.3016, "step": 303200 }, { "epoch": 98.0284421460892, "grad_norm": 1.783571720123291, "learning_rate": 0.001, "loss": 1.2019, "step": 303300 }, { "epoch": 98.06076276664513, "grad_norm": 1.5965900421142578, "learning_rate": 0.001, "loss": 1.2057, "step": 303400 }, { "epoch": 98.09308338720103, "grad_norm": 1.517217755317688, "learning_rate": 0.001, "loss": 1.2201, "step": 303500 }, { "epoch": 98.12540400775696, "grad_norm": 1.5559382438659668, "learning_rate": 0.001, "loss": 1.2208, "step": 303600 }, { "epoch": 98.15772462831286, "grad_norm": 1.6932978630065918, "learning_rate": 0.001, "loss": 1.2203, "step": 303700 }, { "epoch": 98.19004524886878, "grad_norm": 1.4250338077545166, "learning_rate": 0.001, "loss": 1.2283, "step": 303800 }, { "epoch": 98.22236586942469, "grad_norm": 1.4945716857910156, "learning_rate": 0.001, "loss": 1.2295, "step": 303900 }, { "epoch": 98.25468648998061, "grad_norm": 1.9087316989898682, "learning_rate": 0.001, "loss": 1.2385, "step": 304000 }, { "epoch": 98.28700711053652, "grad_norm": 1.467848777770996, "learning_rate": 0.001, "loss": 1.2366, "step": 304100 }, { "epoch": 98.31932773109244, "grad_norm": 1.7165251970291138, "learning_rate": 0.001, "loss": 1.2442, "step": 304200 }, { "epoch": 98.35164835164835, "grad_norm": 1.2803969383239746, "learning_rate": 0.001, "loss": 1.2407, "step": 304300 }, { "epoch": 98.38396897220427, "grad_norm": 1.6253660917282104, "learning_rate": 0.001, "loss": 1.2553, "step": 304400 }, { "epoch": 98.41628959276018, "grad_norm": 1.965498447418213, "learning_rate": 0.001, "loss": 1.2583, "step": 304500 }, { "epoch": 98.4486102133161, "grad_norm": 1.4171141386032104, "learning_rate": 0.001, "loss": 1.2593, "step": 304600 }, { "epoch": 98.48093083387201, "grad_norm": 1.5294244289398193, "learning_rate": 0.001, "loss": 1.2507, "step": 304700 }, { "epoch": 98.51325145442793, "grad_norm": 1.8220654726028442, "learning_rate": 0.001, "loss": 1.2554, "step": 304800 }, { "epoch": 98.54557207498384, "grad_norm": 2.1719462871551514, "learning_rate": 0.001, "loss": 1.2863, "step": 304900 }, { "epoch": 98.57789269553976, "grad_norm": 1.4300793409347534, "learning_rate": 0.001, "loss": 1.2797, "step": 305000 }, { "epoch": 98.61021331609567, "grad_norm": 2.0281763076782227, "learning_rate": 0.001, "loss": 1.2714, "step": 305100 }, { "epoch": 98.64253393665159, "grad_norm": 2.3171002864837646, "learning_rate": 0.001, "loss": 1.2826, "step": 305200 }, { "epoch": 98.6748545572075, "grad_norm": 1.9922728538513184, "learning_rate": 0.001, "loss": 1.2843, "step": 305300 }, { "epoch": 98.70717517776342, "grad_norm": 1.9118890762329102, "learning_rate": 0.001, "loss": 1.2728, "step": 305400 }, { "epoch": 98.73949579831933, "grad_norm": 2.1546452045440674, "learning_rate": 0.001, "loss": 1.2845, "step": 305500 }, { "epoch": 98.77181641887525, "grad_norm": 1.4353299140930176, "learning_rate": 0.001, "loss": 1.2918, "step": 305600 }, { "epoch": 98.80413703943115, "grad_norm": 1.4719852209091187, "learning_rate": 0.001, "loss": 1.2865, "step": 305700 }, { "epoch": 98.83645765998708, "grad_norm": 1.617453694343567, "learning_rate": 0.001, "loss": 1.2939, "step": 305800 }, { "epoch": 98.86877828054298, "grad_norm": 1.263366460800171, "learning_rate": 0.001, "loss": 1.2932, "step": 305900 }, { "epoch": 98.9010989010989, "grad_norm": 1.5086780786514282, "learning_rate": 0.001, "loss": 1.2963, "step": 306000 }, { "epoch": 98.93341952165481, "grad_norm": 1.6143393516540527, "learning_rate": 0.001, "loss": 1.3091, "step": 306100 }, { "epoch": 98.96574014221073, "grad_norm": 1.3707364797592163, "learning_rate": 0.001, "loss": 1.3095, "step": 306200 }, { "epoch": 98.99806076276664, "grad_norm": 2.250619649887085, "learning_rate": 0.001, "loss": 1.2998, "step": 306300 }, { "epoch": 99.03038138332256, "grad_norm": 1.6753365993499756, "learning_rate": 0.001, "loss": 1.1985, "step": 306400 }, { "epoch": 99.06270200387847, "grad_norm": 1.8750298023223877, "learning_rate": 0.001, "loss": 1.1983, "step": 306500 }, { "epoch": 99.09502262443439, "grad_norm": 1.4335172176361084, "learning_rate": 0.001, "loss": 1.214, "step": 306600 }, { "epoch": 99.1273432449903, "grad_norm": 1.7616883516311646, "learning_rate": 0.001, "loss": 1.2191, "step": 306700 }, { "epoch": 99.15966386554622, "grad_norm": 1.9447722434997559, "learning_rate": 0.001, "loss": 1.2059, "step": 306800 }, { "epoch": 99.19198448610213, "grad_norm": 1.7094531059265137, "learning_rate": 0.001, "loss": 1.2185, "step": 306900 }, { "epoch": 99.22430510665805, "grad_norm": 2.3314735889434814, "learning_rate": 0.001, "loss": 1.2213, "step": 307000 }, { "epoch": 99.25662572721396, "grad_norm": 1.6030874252319336, "learning_rate": 0.001, "loss": 1.2341, "step": 307100 }, { "epoch": 99.28894634776988, "grad_norm": 1.6456331014633179, "learning_rate": 0.001, "loss": 1.2499, "step": 307200 }, { "epoch": 99.32126696832579, "grad_norm": 1.9911348819732666, "learning_rate": 0.001, "loss": 1.2435, "step": 307300 }, { "epoch": 99.35358758888171, "grad_norm": 1.4561117887496948, "learning_rate": 0.001, "loss": 1.2372, "step": 307400 }, { "epoch": 99.38590820943762, "grad_norm": 2.072667121887207, "learning_rate": 0.001, "loss": 1.2546, "step": 307500 }, { "epoch": 99.41822882999354, "grad_norm": 1.9429185390472412, "learning_rate": 0.001, "loss": 1.2557, "step": 307600 }, { "epoch": 99.45054945054945, "grad_norm": 1.8015973567962646, "learning_rate": 0.001, "loss": 1.2533, "step": 307700 }, { "epoch": 99.48287007110537, "grad_norm": 1.6045411825180054, "learning_rate": 0.001, "loss": 1.2552, "step": 307800 }, { "epoch": 99.51519069166127, "grad_norm": 1.370804786682129, "learning_rate": 0.001, "loss": 1.2708, "step": 307900 }, { "epoch": 99.5475113122172, "grad_norm": 1.4999316930770874, "learning_rate": 0.001, "loss": 1.2585, "step": 308000 }, { "epoch": 99.5798319327731, "grad_norm": 1.7964613437652588, "learning_rate": 0.001, "loss": 1.2804, "step": 308100 }, { "epoch": 99.61215255332903, "grad_norm": 1.7374740839004517, "learning_rate": 0.001, "loss": 1.2789, "step": 308200 }, { "epoch": 99.64447317388493, "grad_norm": 1.268214225769043, "learning_rate": 0.001, "loss": 1.2704, "step": 308300 }, { "epoch": 99.67679379444085, "grad_norm": 2.9452226161956787, "learning_rate": 0.001, "loss": 1.2776, "step": 308400 }, { "epoch": 99.70911441499676, "grad_norm": 2.1403918266296387, "learning_rate": 0.001, "loss": 1.2827, "step": 308500 }, { "epoch": 99.74143503555268, "grad_norm": 1.5536611080169678, "learning_rate": 0.001, "loss": 1.2794, "step": 308600 }, { "epoch": 99.77375565610859, "grad_norm": 2.0996694564819336, "learning_rate": 0.001, "loss": 1.2843, "step": 308700 }, { "epoch": 99.80607627666451, "grad_norm": 1.7627614736557007, "learning_rate": 0.001, "loss": 1.2726, "step": 308800 }, { "epoch": 99.83839689722042, "grad_norm": 1.360751748085022, "learning_rate": 0.001, "loss": 1.2856, "step": 308900 }, { "epoch": 99.87071751777634, "grad_norm": 1.9599181413650513, "learning_rate": 0.001, "loss": 1.2973, "step": 309000 }, { "epoch": 99.90303813833225, "grad_norm": 1.8387882709503174, "learning_rate": 0.001, "loss": 1.306, "step": 309100 }, { "epoch": 99.93535875888817, "grad_norm": 1.9950312376022339, "learning_rate": 0.001, "loss": 1.3021, "step": 309200 }, { "epoch": 99.96767937944408, "grad_norm": 1.6858352422714233, "learning_rate": 0.001, "loss": 1.2996, "step": 309300 }, { "epoch": 100.0, "grad_norm": 2.136234998703003, "learning_rate": 0.001, "loss": 1.2644, "step": 309400 }, { "epoch": 100.0, "step": 309400, "total_flos": 3.2122963901952e+18, "train_loss": 1.7981142194978483, "train_runtime": 30930.6445, "train_samples_per_second": 320.071, "train_steps_per_second": 10.003 } ], "logging_steps": 100, "max_steps": 309400, "num_input_tokens_seen": 0, "num_train_epochs": 100, "save_steps": 2000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 3.2122963901952e+18, "train_batch_size": 32, "trial_name": null, "trial_params": null }