| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9994186181716704, | |
| "eval_steps": 500, | |
| "global_step": 1343, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0007441687402618543, | |
| "grad_norm": 40.28213183146898, | |
| "learning_rate": 0.0, | |
| "loss": 5.7576, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0014883374805237087, | |
| "grad_norm": 37.954846879312846, | |
| "learning_rate": 2.6164804129562553e-06, | |
| "loss": 5.742, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.002232506220785563, | |
| "grad_norm": 40.293921615599, | |
| "learning_rate": 4.1470233384070705e-06, | |
| "loss": 5.769, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0029766749610474173, | |
| "grad_norm": 32.025476247408996, | |
| "learning_rate": 5.2329608259125105e-06, | |
| "loss": 5.6641, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.003720843701309272, | |
| "grad_norm": 17.42845549889482, | |
| "learning_rate": 6.0752793805656175e-06, | |
| "loss": 5.5356, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004465012441571126, | |
| "grad_norm": 16.319584815304317, | |
| "learning_rate": 6.763503751363326e-06, | |
| "loss": 5.4658, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.00520918118183298, | |
| "grad_norm": 12.406340568802632, | |
| "learning_rate": 7.345389165780057e-06, | |
| "loss": 5.0944, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005953349922094835, | |
| "grad_norm": 8.9792930282741, | |
| "learning_rate": 7.849441238868767e-06, | |
| "loss": 4.8825, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006697518662356689, | |
| "grad_norm": 4.473612590732251, | |
| "learning_rate": 8.294046676814141e-06, | |
| "loss": 4.7899, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.007441687402618544, | |
| "grad_norm": 3.0570747744666575, | |
| "learning_rate": 8.691759793521874e-06, | |
| "loss": 4.6911, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.008185856142880399, | |
| "grad_norm": 2.622215094814623, | |
| "learning_rate": 9.051535070126044e-06, | |
| "loss": 4.6286, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.008930024883142252, | |
| "grad_norm": 2.1643357269799623, | |
| "learning_rate": 9.379984164319582e-06, | |
| "loss": 4.5094, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.009674193623404107, | |
| "grad_norm": 2.072554773904808, | |
| "learning_rate": 9.682128041841535e-06, | |
| "loss": 4.4681, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.01041836236366596, | |
| "grad_norm": 2.147660885872088, | |
| "learning_rate": 9.961869578736312e-06, | |
| "loss": 4.4617, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.011162531103927816, | |
| "grad_norm": 1.9963899271010772, | |
| "learning_rate": 1.022230271897269e-05, | |
| "loss": 4.4622, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.01190669984418967, | |
| "grad_norm": 1.8600670612864825, | |
| "learning_rate": 1.0465921651825021e-05, | |
| "loss": 4.3966, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.012650868584451525, | |
| "grad_norm": 1.636497444567462, | |
| "learning_rate": 1.0694766462818038e-05, | |
| "loss": 4.4334, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.013395037324713378, | |
| "grad_norm": 1.5041799977146408, | |
| "learning_rate": 1.0910527089770397e-05, | |
| "loss": 4.3613, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.014139206064975233, | |
| "grad_norm": 1.4087110131660556, | |
| "learning_rate": 1.1114619134583114e-05, | |
| "loss": 4.357, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.014883374805237088, | |
| "grad_norm": 1.3312378823006819, | |
| "learning_rate": 1.1308240206478129e-05, | |
| "loss": 4.3417, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.015627543545498942, | |
| "grad_norm": 1.3916103687434003, | |
| "learning_rate": 1.1492412504187127e-05, | |
| "loss": 4.3748, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.016371712285760797, | |
| "grad_norm": 1.2115008828872131, | |
| "learning_rate": 1.16680154830823e-05, | |
| "loss": 4.2836, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.017115881026022652, | |
| "grad_norm": 1.2767121737908453, | |
| "learning_rate": 1.1835811254817263e-05, | |
| "loss": 4.2817, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.017860049766284504, | |
| "grad_norm": 1.1856461823529274, | |
| "learning_rate": 1.199646457727584e-05, | |
| "loss": 4.255, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01860421850654636, | |
| "grad_norm": 1.2189618649742828, | |
| "learning_rate": 1.2150558761131235e-05, | |
| "loss": 4.2333, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.019348387246808214, | |
| "grad_norm": 1.1814426942666032, | |
| "learning_rate": 1.2298608454797791e-05, | |
| "loss": 4.2717, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.02009255598707007, | |
| "grad_norm": 1.2135253960285517, | |
| "learning_rate": 1.2441070015221212e-05, | |
| "loss": 4.2211, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.02083672472733192, | |
| "grad_norm": 1.1319015604750045, | |
| "learning_rate": 1.2578349991692568e-05, | |
| "loss": 4.2182, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.021580893467593776, | |
| "grad_norm": 1.051774275255818, | |
| "learning_rate": 1.2710812120265033e-05, | |
| "loss": 4.2365, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.02232506220785563, | |
| "grad_norm": 1.100570126333654, | |
| "learning_rate": 1.2838783131928944e-05, | |
| "loss": 4.2004, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.023069230948117487, | |
| "grad_norm": 1.1518958788113443, | |
| "learning_rate": 1.296255760806741e-05, | |
| "loss": 4.2277, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.02381339968837934, | |
| "grad_norm": 1.04528479799854, | |
| "learning_rate": 1.3082402064781279e-05, | |
| "loss": 4.1865, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.024557568428641194, | |
| "grad_norm": 1.0243455443660119, | |
| "learning_rate": 1.3198558408533116e-05, | |
| "loss": 4.1941, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.02530173716890305, | |
| "grad_norm": 1.0565306632295002, | |
| "learning_rate": 1.3311246875774294e-05, | |
| "loss": 4.1809, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.026045905909164904, | |
| "grad_norm": 1.0539485341060912, | |
| "learning_rate": 1.3420668546345674e-05, | |
| "loss": 4.1751, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.026790074649426756, | |
| "grad_norm": 1.056986642115358, | |
| "learning_rate": 1.3527007502726652e-05, | |
| "loss": 4.1259, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.02753424338968861, | |
| "grad_norm": 1.0593678198418355, | |
| "learning_rate": 1.3630432693377191e-05, | |
| "loss": 4.1554, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.028278412129950466, | |
| "grad_norm": 1.0523046068647912, | |
| "learning_rate": 1.3731099547539368e-05, | |
| "loss": 4.1521, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.02902258087021232, | |
| "grad_norm": 1.1044755308381733, | |
| "learning_rate": 1.3829151380248607e-05, | |
| "loss": 4.1524, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.029766749610474177, | |
| "grad_norm": 1.108136345717576, | |
| "learning_rate": 1.3924720619434385e-05, | |
| "loss": 4.1606, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03051091835073603, | |
| "grad_norm": 1.1207185669171094, | |
| "learning_rate": 1.4017929881477741e-05, | |
| "loss": 4.1489, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.031255087090997884, | |
| "grad_norm": 1.062031659188879, | |
| "learning_rate": 1.4108892917143383e-05, | |
| "loss": 4.0714, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.03199925583125974, | |
| "grad_norm": 1.0837395060391366, | |
| "learning_rate": 1.4197715446192922e-05, | |
| "loss": 4.0949, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.032743424571521594, | |
| "grad_norm": 1.124135212493442, | |
| "learning_rate": 1.4284495896038555e-05, | |
| "loss": 4.1546, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.03348759331178345, | |
| "grad_norm": 1.0604800276669095, | |
| "learning_rate": 1.4369326057379758e-05, | |
| "loss": 4.103, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.034231762052045304, | |
| "grad_norm": 1.0839103604910125, | |
| "learning_rate": 1.4452291667773517e-05, | |
| "loss": 4.0937, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.03497593079230715, | |
| "grad_norm": 1.0430276195739923, | |
| "learning_rate": 1.4533472932439719e-05, | |
| "loss": 4.1243, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03572009953256901, | |
| "grad_norm": 1.0142332098960116, | |
| "learning_rate": 1.4612944990232094e-05, | |
| "loss": 4.0517, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.03646426827283086, | |
| "grad_norm": 1.0773686541540213, | |
| "learning_rate": 1.4690778331560114e-05, | |
| "loss": 4.1042, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03720843701309272, | |
| "grad_norm": 1.0866711243976093, | |
| "learning_rate": 1.4767039174087491e-05, | |
| "loss": 4.065, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.037952605753354574, | |
| "grad_norm": 1.0449386312700308, | |
| "learning_rate": 1.4841789801225108e-05, | |
| "loss": 4.078, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03869677449361643, | |
| "grad_norm": 1.0043721671548014, | |
| "learning_rate": 1.491508886775405e-05, | |
| "loss": 4.0739, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.039440943233878284, | |
| "grad_norm": 1.0593671885336344, | |
| "learning_rate": 1.4986991676336104e-05, | |
| "loss": 4.0572, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.04018511197414014, | |
| "grad_norm": 1.1336381730686247, | |
| "learning_rate": 1.5057550428177467e-05, | |
| "loss": 4.0897, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.04092928071440199, | |
| "grad_norm": 1.0651935068897007, | |
| "learning_rate": 1.5126814450691663e-05, | |
| "loss": 4.0533, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.04167344945466384, | |
| "grad_norm": 1.0470533883830977, | |
| "learning_rate": 1.5194830404648826e-05, | |
| "loss": 4.0569, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0424176181949257, | |
| "grad_norm": 1.02131208058522, | |
| "learning_rate": 1.5261642472990182e-05, | |
| "loss": 4.075, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.04316178693518755, | |
| "grad_norm": 0.9978789039090863, | |
| "learning_rate": 1.5327292533221287e-05, | |
| "loss": 4.08, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.04390595567544941, | |
| "grad_norm": 1.005388655066182, | |
| "learning_rate": 1.539182031506852e-05, | |
| "loss": 4.0402, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.04465012441571126, | |
| "grad_norm": 1.1167047788609468, | |
| "learning_rate": 1.54552635448852e-05, | |
| "loss": 4.0594, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04539429315597312, | |
| "grad_norm": 1.041248072313915, | |
| "learning_rate": 1.5517658078121626e-05, | |
| "loss": 4.0769, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.046138461896234974, | |
| "grad_norm": 1.0241035033662262, | |
| "learning_rate": 1.5579038021023668e-05, | |
| "loss": 4.0444, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.04688263063649683, | |
| "grad_norm": 1.0734353487861803, | |
| "learning_rate": 1.5639435842594197e-05, | |
| "loss": 4.0047, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.04762679937675868, | |
| "grad_norm": 1.040455084752415, | |
| "learning_rate": 1.5698882477737533e-05, | |
| "loss": 4.065, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.04837096811702053, | |
| "grad_norm": 1.1039376623928785, | |
| "learning_rate": 1.575740742240715e-05, | |
| "loss": 4.05, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.04911513685728239, | |
| "grad_norm": 1.0412052698427907, | |
| "learning_rate": 1.581503882148937e-05, | |
| "loss": 4.0374, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.04985930559754424, | |
| "grad_norm": 1.1296810898624878, | |
| "learning_rate": 1.587180355007843e-05, | |
| "loss": 4.0766, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0506034743378061, | |
| "grad_norm": 1.0504232245348173, | |
| "learning_rate": 1.5927727288730554e-05, | |
| "loss": 4.0142, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.05134764307806795, | |
| "grad_norm": 1.11921482395444, | |
| "learning_rate": 1.598283459322433e-05, | |
| "loss": 4.1023, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.05209181181832981, | |
| "grad_norm": 1.1058159022728031, | |
| "learning_rate": 1.6037148959301934e-05, | |
| "loss": 4.0784, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.052835980558591664, | |
| "grad_norm": 1.1767577075976265, | |
| "learning_rate": 1.6090692882818154e-05, | |
| "loss": 4.0295, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.05358014929885351, | |
| "grad_norm": 1.0463748666754686, | |
| "learning_rate": 1.614348791568291e-05, | |
| "loss": 4.0473, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.05432431803911537, | |
| "grad_norm": 1.038483489022193, | |
| "learning_rate": 1.6195554717945162e-05, | |
| "loss": 4.0035, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.05506848677937722, | |
| "grad_norm": 1.0837638408441645, | |
| "learning_rate": 1.6246913106333447e-05, | |
| "loss": 4.0431, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.05581265551963908, | |
| "grad_norm": 1.0310698958053535, | |
| "learning_rate": 1.6297582099538306e-05, | |
| "loss": 4.0091, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.05655682425990093, | |
| "grad_norm": 1.0602478710208674, | |
| "learning_rate": 1.6347579960495625e-05, | |
| "loss": 4.0102, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.05730099300016279, | |
| "grad_norm": 1.0802699317216853, | |
| "learning_rate": 1.63969242359061e-05, | |
| "loss": 3.9805, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.05804516174042464, | |
| "grad_norm": 1.0408226478834655, | |
| "learning_rate": 1.6445631793204863e-05, | |
| "loss": 4.0151, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0587893304806865, | |
| "grad_norm": 1.0310241855703013, | |
| "learning_rate": 1.6493718855176122e-05, | |
| "loss": 3.9576, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.059533499220948354, | |
| "grad_norm": 1.0914051950397774, | |
| "learning_rate": 1.6541201032390638e-05, | |
| "loss": 4.0054, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0602776679612102, | |
| "grad_norm": 1.0365938954589975, | |
| "learning_rate": 1.6588093353628282e-05, | |
| "loss": 4.0384, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.06102183670147206, | |
| "grad_norm": 1.0793361858298907, | |
| "learning_rate": 1.6634410294433997e-05, | |
| "loss": 4.0102, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.06176600544173391, | |
| "grad_norm": 1.1256865378091168, | |
| "learning_rate": 1.6680165803943017e-05, | |
| "loss": 4.0045, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.06251017418199577, | |
| "grad_norm": 1.02834469225436, | |
| "learning_rate": 1.672537333009964e-05, | |
| "loss": 4.0228, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.06325434292225762, | |
| "grad_norm": 1.0553223698349907, | |
| "learning_rate": 1.677004584338366e-05, | |
| "loss": 3.9772, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.06399851166251948, | |
| "grad_norm": 1.081083606766696, | |
| "learning_rate": 1.6814195859149176e-05, | |
| "loss": 4.0488, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.06474268040278133, | |
| "grad_norm": 1.087880426024787, | |
| "learning_rate": 1.6857835458672104e-05, | |
| "loss": 4.0206, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.06548684914304319, | |
| "grad_norm": 1.0436873463050869, | |
| "learning_rate": 1.690097630899481e-05, | |
| "loss": 3.9702, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.06623101788330504, | |
| "grad_norm": 1.0267988724124961, | |
| "learning_rate": 1.6943629681649592e-05, | |
| "loss": 3.9782, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.0669751866235669, | |
| "grad_norm": 1.1385464168004937, | |
| "learning_rate": 1.6985806470336015e-05, | |
| "loss": 3.9942, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06771935536382875, | |
| "grad_norm": 1.0432403937747066, | |
| "learning_rate": 1.702751720762159e-05, | |
| "loss": 4.0206, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.06846352410409061, | |
| "grad_norm": 1.1798870993524078, | |
| "learning_rate": 1.7068772080729773e-05, | |
| "loss": 3.9927, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.06920769284435245, | |
| "grad_norm": 1.0251165578409693, | |
| "learning_rate": 1.710958094647448e-05, | |
| "loss": 3.9898, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0699518615846143, | |
| "grad_norm": 1.050368231727308, | |
| "learning_rate": 1.7149953345395975e-05, | |
| "loss": 3.9685, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.07069603032487616, | |
| "grad_norm": 1.1551778718206707, | |
| "learning_rate": 1.7189898515148732e-05, | |
| "loss": 4.0315, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.07144019906513802, | |
| "grad_norm": 1.0218383615477624, | |
| "learning_rate": 1.722942540318835e-05, | |
| "loss": 3.9895, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.07218436780539987, | |
| "grad_norm": 1.2730542158096376, | |
| "learning_rate": 1.726854267880107e-05, | |
| "loss": 3.9863, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.07292853654566173, | |
| "grad_norm": 1.059834593309155, | |
| "learning_rate": 1.730725874451637e-05, | |
| "loss": 3.9932, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.07367270528592358, | |
| "grad_norm": 1.0414768916549912, | |
| "learning_rate": 1.7345581746940185e-05, | |
| "loss": 3.97, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.07441687402618544, | |
| "grad_norm": 1.0459663928761072, | |
| "learning_rate": 1.738351958704375e-05, | |
| "loss": 4.0594, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07516104276644729, | |
| "grad_norm": 1.0079698932917285, | |
| "learning_rate": 1.7421079929940502e-05, | |
| "loss": 3.9647, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.07590521150670915, | |
| "grad_norm": 1.104403325606981, | |
| "learning_rate": 1.7458270214181364e-05, | |
| "loss": 3.9891, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.076649380246971, | |
| "grad_norm": 1.0899824311589137, | |
| "learning_rate": 1.749509766059657e-05, | |
| "loss": 3.9572, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.07739354898723286, | |
| "grad_norm": 1.0256662708529325, | |
| "learning_rate": 1.7531569280710302e-05, | |
| "loss": 4.0154, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07813771772749471, | |
| "grad_norm": 1.003879061204672, | |
| "learning_rate": 1.7567691884752744e-05, | |
| "loss": 3.9524, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07888188646775657, | |
| "grad_norm": 1.0294158850936943, | |
| "learning_rate": 1.760347208929236e-05, | |
| "loss": 3.9769, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07962605520801842, | |
| "grad_norm": 1.034265397813248, | |
| "learning_rate": 1.7638916324509837e-05, | |
| "loss": 3.9981, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.08037022394828028, | |
| "grad_norm": 1.0635676727786938, | |
| "learning_rate": 1.7674030841133725e-05, | |
| "loss": 3.9391, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.08111439268854213, | |
| "grad_norm": 1.0382548490363581, | |
| "learning_rate": 1.770882171705639e-05, | |
| "loss": 3.994, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.08185856142880397, | |
| "grad_norm": 1.24849674389146, | |
| "learning_rate": 1.774329486364792e-05, | |
| "loss": 4.0045, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.08260273016906583, | |
| "grad_norm": 1.0543355977515012, | |
| "learning_rate": 1.777745603178426e-05, | |
| "loss": 3.9475, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.08334689890932769, | |
| "grad_norm": 1.0429344446319329, | |
| "learning_rate": 1.781131081760508e-05, | |
| "loss": 3.9676, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.08409106764958954, | |
| "grad_norm": 1.13971103799619, | |
| "learning_rate": 1.784486466801566e-05, | |
| "loss": 3.9757, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.0848352363898514, | |
| "grad_norm": 1.0843231601372652, | |
| "learning_rate": 1.787812288594644e-05, | |
| "loss": 3.9669, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.08557940513011325, | |
| "grad_norm": 1.13253776607225, | |
| "learning_rate": 1.7911090635382878e-05, | |
| "loss": 3.9361, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.0863235738703751, | |
| "grad_norm": 1.123191082800284, | |
| "learning_rate": 1.7943772946177543e-05, | |
| "loss": 3.9607, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.08706774261063696, | |
| "grad_norm": 1.0399771364371393, | |
| "learning_rate": 1.7976174718655676e-05, | |
| "loss": 3.9544, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.08781191135089882, | |
| "grad_norm": 1.0927560633321864, | |
| "learning_rate": 1.8008300728024776e-05, | |
| "loss": 3.9131, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.08855608009116067, | |
| "grad_norm": 1.1176242088766228, | |
| "learning_rate": 1.8040155628598096e-05, | |
| "loss": 3.9274, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.08930024883142253, | |
| "grad_norm": 1.1688430079344734, | |
| "learning_rate": 1.8071743957841455e-05, | |
| "loss": 3.9952, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.09004441757168438, | |
| "grad_norm": 1.0186311612749528, | |
| "learning_rate": 1.810307014025209e-05, | |
| "loss": 3.9239, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.09078858631194624, | |
| "grad_norm": 1.1046314658484728, | |
| "learning_rate": 1.8134138491077882e-05, | |
| "loss": 3.925, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.09153275505220809, | |
| "grad_norm": 1.0187952497474388, | |
| "learning_rate": 1.8164953219884808e-05, | |
| "loss": 3.9322, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.09227692379246995, | |
| "grad_norm": 1.043346272780104, | |
| "learning_rate": 1.819551843397992e-05, | |
| "loss": 3.928, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0930210925327318, | |
| "grad_norm": 1.0382874379396565, | |
| "learning_rate": 1.8225838141696857e-05, | |
| "loss": 3.9733, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.09376526127299366, | |
| "grad_norm": 1.045358915304584, | |
| "learning_rate": 1.8255916255550453e-05, | |
| "loss": 3.9359, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0945094300132555, | |
| "grad_norm": 1.0229110283842722, | |
| "learning_rate": 1.8285756595266698e-05, | |
| "loss": 3.9206, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.09525359875351735, | |
| "grad_norm": 1.0252642835887658, | |
| "learning_rate": 1.831536289069379e-05, | |
| "loss": 3.9905, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.09599776749377921, | |
| "grad_norm": 1.0416563264436998, | |
| "learning_rate": 1.834473878459999e-05, | |
| "loss": 3.9558, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.09674193623404106, | |
| "grad_norm": 0.9906081180533991, | |
| "learning_rate": 1.837388783536341e-05, | |
| "loss": 3.8951, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.09748610497430292, | |
| "grad_norm": 1.0247527219623138, | |
| "learning_rate": 1.8402813519558736e-05, | |
| "loss": 3.9471, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.09823027371456478, | |
| "grad_norm": 1.019253253192708, | |
| "learning_rate": 1.8431519234445625e-05, | |
| "loss": 3.9687, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.09897444245482663, | |
| "grad_norm": 1.0466127238185976, | |
| "learning_rate": 1.846000830036317e-05, | |
| "loss": 3.9819, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.09971861119508849, | |
| "grad_norm": 1.0144237877513214, | |
| "learning_rate": 1.848828396303469e-05, | |
| "loss": 3.9304, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.10046277993535034, | |
| "grad_norm": 1.012969708052044, | |
| "learning_rate": 1.851634939578683e-05, | |
| "loss": 3.9286, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.1012069486756122, | |
| "grad_norm": 1.0382905383836176, | |
| "learning_rate": 1.8544207701686807e-05, | |
| "loss": 3.9226, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.10195111741587405, | |
| "grad_norm": 1.0089812321886908, | |
| "learning_rate": 1.8571861915601312e-05, | |
| "loss": 3.9373, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1026952861561359, | |
| "grad_norm": 1.0018437345247462, | |
| "learning_rate": 1.859931500618059e-05, | |
| "loss": 3.9369, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.10343945489639776, | |
| "grad_norm": 1.0002270998081322, | |
| "learning_rate": 1.8626569877770852e-05, | |
| "loss": 3.9793, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.10418362363665962, | |
| "grad_norm": 0.9979219477879793, | |
| "learning_rate": 1.8653629372258186e-05, | |
| "loss": 3.9185, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10492779237692147, | |
| "grad_norm": 1.0293079269842134, | |
| "learning_rate": 1.8680496270846792e-05, | |
| "loss": 3.9116, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.10567196111718333, | |
| "grad_norm": 1.0051994987775417, | |
| "learning_rate": 1.8707173295774414e-05, | |
| "loss": 3.9653, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.10641612985744518, | |
| "grad_norm": 0.9917678837635974, | |
| "learning_rate": 1.873366311196758e-05, | |
| "loss": 3.9347, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.10716029859770702, | |
| "grad_norm": 1.0134566401575213, | |
| "learning_rate": 1.8759968328639164e-05, | |
| "loss": 3.972, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.10790446733796888, | |
| "grad_norm": 1.0079991313690682, | |
| "learning_rate": 1.878609150083065e-05, | |
| "loss": 3.979, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.10864863607823073, | |
| "grad_norm": 0.9724491790561419, | |
| "learning_rate": 1.8812035130901415e-05, | |
| "loss": 3.9103, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.10939280481849259, | |
| "grad_norm": 1.0266607847449294, | |
| "learning_rate": 1.8837801669967185e-05, | |
| "loss": 3.9415, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.11013697355875444, | |
| "grad_norm": 1.014059449927133, | |
| "learning_rate": 1.88633935192897e-05, | |
| "loss": 3.9271, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.1108811422990163, | |
| "grad_norm": 1.008729408663488, | |
| "learning_rate": 1.888881303161964e-05, | |
| "loss": 3.9352, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.11162531103927816, | |
| "grad_norm": 0.974217214758975, | |
| "learning_rate": 1.8914062512494562e-05, | |
| "loss": 3.9607, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.11236947977954001, | |
| "grad_norm": 1.0034271324861523, | |
| "learning_rate": 1.8939144221493803e-05, | |
| "loss": 3.9491, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.11311364851980187, | |
| "grad_norm": 1.0185215737707887, | |
| "learning_rate": 1.896406037345188e-05, | |
| "loss": 3.9278, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.11385781726006372, | |
| "grad_norm": 0.9878339771587401, | |
| "learning_rate": 1.8988813139632178e-05, | |
| "loss": 3.9317, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.11460198600032558, | |
| "grad_norm": 1.016349316845976, | |
| "learning_rate": 1.901340464886236e-05, | |
| "loss": 3.909, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.11534615474058743, | |
| "grad_norm": 1.027101832954034, | |
| "learning_rate": 1.903783698863303e-05, | |
| "loss": 3.9199, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.11609032348084929, | |
| "grad_norm": 0.9818722079648826, | |
| "learning_rate": 1.906211220616112e-05, | |
| "loss": 3.9179, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.11683449222111114, | |
| "grad_norm": 1.0162140165365403, | |
| "learning_rate": 1.9086232309419236e-05, | |
| "loss": 3.9235, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.117578660961373, | |
| "grad_norm": 0.974360824840726, | |
| "learning_rate": 1.9110199268132375e-05, | |
| "loss": 3.9338, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.11832282970163485, | |
| "grad_norm": 1.034581455039987, | |
| "learning_rate": 1.9134015014743175e-05, | |
| "loss": 3.9137, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.11906699844189671, | |
| "grad_norm": 1.040262330326242, | |
| "learning_rate": 1.9157681445346897e-05, | |
| "loss": 3.9329, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11981116718215855, | |
| "grad_norm": 1.0392129101325005, | |
| "learning_rate": 1.918120042059732e-05, | |
| "loss": 3.8952, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.1205553359224204, | |
| "grad_norm": 1.0088871617648623, | |
| "learning_rate": 1.9204573766584535e-05, | |
| "loss": 3.9125, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.12129950466268226, | |
| "grad_norm": 0.9805897104680126, | |
| "learning_rate": 1.9227803275685793e-05, | |
| "loss": 3.9114, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.12204367340294411, | |
| "grad_norm": 1.028458200077777, | |
| "learning_rate": 1.9250890707390254e-05, | |
| "loss": 3.9254, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.12278784214320597, | |
| "grad_norm": 1.0256714252479524, | |
| "learning_rate": 1.927383778909873e-05, | |
| "loss": 3.9434, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.12353201088346782, | |
| "grad_norm": 0.9893765209388768, | |
| "learning_rate": 1.9296646216899273e-05, | |
| "loss": 3.9391, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.12427617962372968, | |
| "grad_norm": 1.0295325339316823, | |
| "learning_rate": 1.9319317656319386e-05, | |
| "loss": 3.9206, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.12502034836399153, | |
| "grad_norm": 1.023371515755248, | |
| "learning_rate": 1.9341853743055896e-05, | |
| "loss": 3.8887, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.12576451710425338, | |
| "grad_norm": 1.0384189813636024, | |
| "learning_rate": 1.936425608368307e-05, | |
| "loss": 3.9451, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.12650868584451525, | |
| "grad_norm": 1.0651166771961635, | |
| "learning_rate": 1.9386526256339914e-05, | |
| "loss": 3.947, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1272528545847771, | |
| "grad_norm": 1.0554377556818044, | |
| "learning_rate": 1.9408665811397255e-05, | |
| "loss": 3.9436, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.12799702332503896, | |
| "grad_norm": 1.009366062125802, | |
| "learning_rate": 1.9430676272105432e-05, | |
| "loss": 3.9107, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.1287411920653008, | |
| "grad_norm": 1.0098888559915948, | |
| "learning_rate": 1.9452559135223174e-05, | |
| "loss": 3.9054, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.12948536080556267, | |
| "grad_norm": 1.0250201494226787, | |
| "learning_rate": 1.947431587162836e-05, | |
| "loss": 3.9077, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.1302295295458245, | |
| "grad_norm": 1.0394330956543154, | |
| "learning_rate": 1.9495947926911298e-05, | |
| "loss": 3.9272, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.13097369828608638, | |
| "grad_norm": 1.0680327402257843, | |
| "learning_rate": 1.9517456721951064e-05, | |
| "loss": 3.9178, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.13171786702634822, | |
| "grad_norm": 0.9817097875184574, | |
| "learning_rate": 1.9538843653475586e-05, | |
| "loss": 3.9195, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.1324620357666101, | |
| "grad_norm": 1.0338017849525705, | |
| "learning_rate": 1.9560110094605845e-05, | |
| "loss": 3.9298, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.13320620450687193, | |
| "grad_norm": 1.1392590141700782, | |
| "learning_rate": 1.958125739538492e-05, | |
| "loss": 3.9225, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.1339503732471338, | |
| "grad_norm": 1.0033539794562174, | |
| "learning_rate": 1.960228688329227e-05, | |
| "loss": 3.8781, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.13469454198739564, | |
| "grad_norm": 1.0845274277424417, | |
| "learning_rate": 1.962319986374374e-05, | |
| "loss": 3.8564, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.1354387107276575, | |
| "grad_norm": 1.0050811751546636, | |
| "learning_rate": 1.9643997620577847e-05, | |
| "loss": 3.9363, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.13618287946791935, | |
| "grad_norm": 1.018459581160744, | |
| "learning_rate": 1.9664681416528695e-05, | |
| "loss": 3.9206, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.13692704820818122, | |
| "grad_norm": 1.07961071902412, | |
| "learning_rate": 1.968525249368603e-05, | |
| "loss": 3.8778, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.13767121694844306, | |
| "grad_norm": 0.9977770305620137, | |
| "learning_rate": 1.9705712073942808e-05, | |
| "loss": 3.9081, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.1384153856887049, | |
| "grad_norm": 1.051869571399524, | |
| "learning_rate": 1.9726061359430737e-05, | |
| "loss": 3.8971, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.13915955442896677, | |
| "grad_norm": 0.9899020097355119, | |
| "learning_rate": 1.9746301532944084e-05, | |
| "loss": 3.8765, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1399037231692286, | |
| "grad_norm": 1.0129926093719666, | |
| "learning_rate": 1.976643375835223e-05, | |
| "loss": 3.9175, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.14064789190949048, | |
| "grad_norm": 0.9775394421834881, | |
| "learning_rate": 1.978645918100127e-05, | |
| "loss": 3.9126, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.14139206064975232, | |
| "grad_norm": 1.0273712313694896, | |
| "learning_rate": 1.980637892810499e-05, | |
| "loss": 3.8938, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1421362293900142, | |
| "grad_norm": 1.0551565867845918, | |
| "learning_rate": 1.9826194109125613e-05, | |
| "loss": 3.8725, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.14288039813027603, | |
| "grad_norm": 0.9578898896374625, | |
| "learning_rate": 1.9845905816144606e-05, | |
| "loss": 3.8855, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.1436245668705379, | |
| "grad_norm": 1.0228519978608672, | |
| "learning_rate": 1.986551512422382e-05, | |
| "loss": 3.8465, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.14436873561079974, | |
| "grad_norm": 1.0541556826203027, | |
| "learning_rate": 1.9885023091757327e-05, | |
| "loss": 3.8942, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1451129043510616, | |
| "grad_norm": 1.070516953608261, | |
| "learning_rate": 1.9904430760814223e-05, | |
| "loss": 3.9217, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.14585707309132345, | |
| "grad_norm": 1.101856520977726, | |
| "learning_rate": 1.9923739157472624e-05, | |
| "loss": 3.8681, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.14660124183158532, | |
| "grad_norm": 0.9911877882058072, | |
| "learning_rate": 1.99429492921452e-05, | |
| "loss": 3.8727, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.14734541057184716, | |
| "grad_norm": 1.0837838674233822, | |
| "learning_rate": 1.996206215989644e-05, | |
| "loss": 3.9013, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.14808957931210903, | |
| "grad_norm": 1.107213213883806, | |
| "learning_rate": 1.9981078740751956e-05, | |
| "loss": 3.9185, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.14883374805237087, | |
| "grad_norm": 1.0147691167060533, | |
| "learning_rate": 2e-05, | |
| "loss": 3.8887, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14957791679263274, | |
| "grad_norm": 1.108760980265153, | |
| "learning_rate": 2e-05, | |
| "loss": 3.8852, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.15032208553289458, | |
| "grad_norm": 1.1606797193437703, | |
| "learning_rate": 1.9982502187226597e-05, | |
| "loss": 3.9103, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.15106625427315643, | |
| "grad_norm": 0.9780382066447286, | |
| "learning_rate": 1.9965004374453195e-05, | |
| "loss": 3.8907, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.1518104230134183, | |
| "grad_norm": 1.0923767759334193, | |
| "learning_rate": 1.9947506561679793e-05, | |
| "loss": 3.8657, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.15255459175368014, | |
| "grad_norm": 1.058031328390541, | |
| "learning_rate": 1.993000874890639e-05, | |
| "loss": 3.9221, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.153298760493942, | |
| "grad_norm": 1.1144653724813092, | |
| "learning_rate": 1.9912510936132984e-05, | |
| "loss": 3.9018, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.15404292923420385, | |
| "grad_norm": 1.086555993942653, | |
| "learning_rate": 1.9895013123359582e-05, | |
| "loss": 3.8904, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.15478709797446571, | |
| "grad_norm": 1.0928089337465563, | |
| "learning_rate": 1.9877515310586177e-05, | |
| "loss": 3.845, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.15553126671472756, | |
| "grad_norm": 1.0117451997906663, | |
| "learning_rate": 1.9860017497812775e-05, | |
| "loss": 3.8596, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.15627543545498943, | |
| "grad_norm": 0.9860809835720783, | |
| "learning_rate": 1.984251968503937e-05, | |
| "loss": 3.8698, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.15701960419525127, | |
| "grad_norm": 1.0469938436783353, | |
| "learning_rate": 1.982502187226597e-05, | |
| "loss": 3.8949, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.15776377293551314, | |
| "grad_norm": 1.0778915521743153, | |
| "learning_rate": 1.9807524059492564e-05, | |
| "loss": 3.902, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.15850794167577498, | |
| "grad_norm": 1.0081826770062852, | |
| "learning_rate": 1.9790026246719162e-05, | |
| "loss": 3.8825, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.15925211041603685, | |
| "grad_norm": 1.0811363993261065, | |
| "learning_rate": 1.977252843394576e-05, | |
| "loss": 3.8684, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.1599962791562987, | |
| "grad_norm": 1.0518907982175134, | |
| "learning_rate": 1.9755030621172356e-05, | |
| "loss": 3.8782, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.16074044789656056, | |
| "grad_norm": 1.017099834407515, | |
| "learning_rate": 1.973753280839895e-05, | |
| "loss": 3.8891, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.1614846166368224, | |
| "grad_norm": 1.0662397354594775, | |
| "learning_rate": 1.972003499562555e-05, | |
| "loss": 3.9223, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.16222878537708427, | |
| "grad_norm": 1.0142656780583652, | |
| "learning_rate": 1.9702537182852148e-05, | |
| "loss": 3.8803, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.1629729541173461, | |
| "grad_norm": 0.9969241687057226, | |
| "learning_rate": 1.9685039370078743e-05, | |
| "loss": 3.9017, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.16371712285760795, | |
| "grad_norm": 1.1145463057895744, | |
| "learning_rate": 1.9667541557305338e-05, | |
| "loss": 3.9077, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.16446129159786982, | |
| "grad_norm": 1.0422175718039535, | |
| "learning_rate": 1.9650043744531936e-05, | |
| "loss": 3.8816, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.16520546033813166, | |
| "grad_norm": 0.9775337504899944, | |
| "learning_rate": 1.963254593175853e-05, | |
| "loss": 3.8886, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.16594962907839353, | |
| "grad_norm": 1.1305819397718035, | |
| "learning_rate": 1.961504811898513e-05, | |
| "loss": 3.9078, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.16669379781865537, | |
| "grad_norm": 1.0523575031319667, | |
| "learning_rate": 1.9597550306211725e-05, | |
| "loss": 3.8921, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.16743796655891724, | |
| "grad_norm": 1.0920228946290917, | |
| "learning_rate": 1.9580052493438323e-05, | |
| "loss": 3.9123, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.16818213529917908, | |
| "grad_norm": 0.9856472634349347, | |
| "learning_rate": 1.9562554680664918e-05, | |
| "loss": 3.8984, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.16892630403944095, | |
| "grad_norm": 0.9874379178084494, | |
| "learning_rate": 1.9545056867891513e-05, | |
| "loss": 3.8544, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.1696704727797028, | |
| "grad_norm": 0.9878390504781994, | |
| "learning_rate": 1.952755905511811e-05, | |
| "loss": 3.8749, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.17041464151996466, | |
| "grad_norm": 0.9665431347439873, | |
| "learning_rate": 1.951006124234471e-05, | |
| "loss": 3.8414, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.1711588102602265, | |
| "grad_norm": 1.0396612817965534, | |
| "learning_rate": 1.9492563429571305e-05, | |
| "loss": 3.8815, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.17190297900048837, | |
| "grad_norm": 0.987556682544731, | |
| "learning_rate": 1.94750656167979e-05, | |
| "loss": 3.9001, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.1726471477407502, | |
| "grad_norm": 1.0723173968695912, | |
| "learning_rate": 1.9457567804024498e-05, | |
| "loss": 3.8743, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.17339131648101208, | |
| "grad_norm": 0.9561409858596005, | |
| "learning_rate": 1.9440069991251097e-05, | |
| "loss": 3.8725, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.17413548522127392, | |
| "grad_norm": 0.9999998650740891, | |
| "learning_rate": 1.9422572178477692e-05, | |
| "loss": 3.8607, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.1748796539615358, | |
| "grad_norm": 0.9775634013017566, | |
| "learning_rate": 1.9405074365704287e-05, | |
| "loss": 3.9033, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.17562382270179763, | |
| "grad_norm": 0.9799845717044794, | |
| "learning_rate": 1.9387576552930885e-05, | |
| "loss": 3.8833, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.17636799144205947, | |
| "grad_norm": 0.9901840199690045, | |
| "learning_rate": 1.937007874015748e-05, | |
| "loss": 3.8233, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.17711216018232134, | |
| "grad_norm": 1.0243609832848712, | |
| "learning_rate": 1.935258092738408e-05, | |
| "loss": 3.8821, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.17785632892258318, | |
| "grad_norm": 1.0121061826434752, | |
| "learning_rate": 1.9335083114610677e-05, | |
| "loss": 3.8677, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.17860049766284505, | |
| "grad_norm": 1.0219606453728582, | |
| "learning_rate": 1.9317585301837272e-05, | |
| "loss": 3.8683, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.1793446664031069, | |
| "grad_norm": 1.0187095816659832, | |
| "learning_rate": 1.9300087489063867e-05, | |
| "loss": 3.8461, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.18008883514336876, | |
| "grad_norm": 0.979083170079006, | |
| "learning_rate": 1.9282589676290465e-05, | |
| "loss": 3.912, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.1808330038836306, | |
| "grad_norm": 1.016894358077027, | |
| "learning_rate": 1.9265091863517064e-05, | |
| "loss": 3.8574, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.18157717262389247, | |
| "grad_norm": 1.0508869962411294, | |
| "learning_rate": 1.924759405074366e-05, | |
| "loss": 3.8491, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.18232134136415432, | |
| "grad_norm": 1.0242007832907005, | |
| "learning_rate": 1.9230096237970254e-05, | |
| "loss": 3.8711, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.18306551010441618, | |
| "grad_norm": 1.0357575567045123, | |
| "learning_rate": 1.9212598425196852e-05, | |
| "loss": 3.8688, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.18380967884467803, | |
| "grad_norm": 1.0059880577184972, | |
| "learning_rate": 1.919510061242345e-05, | |
| "loss": 3.8836, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.1845538475849399, | |
| "grad_norm": 0.9404617991229203, | |
| "learning_rate": 1.9177602799650046e-05, | |
| "loss": 3.8558, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.18529801632520174, | |
| "grad_norm": 0.9979847270034242, | |
| "learning_rate": 1.916010498687664e-05, | |
| "loss": 3.8518, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.1860421850654636, | |
| "grad_norm": 1.1095814655217553, | |
| "learning_rate": 1.914260717410324e-05, | |
| "loss": 3.8487, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.18678635380572545, | |
| "grad_norm": 1.000582783386416, | |
| "learning_rate": 1.9125109361329834e-05, | |
| "loss": 3.8805, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.18753052254598732, | |
| "grad_norm": 1.0532889783814745, | |
| "learning_rate": 1.9107611548556433e-05, | |
| "loss": 3.8464, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.18827469128624916, | |
| "grad_norm": 1.0705950262609198, | |
| "learning_rate": 1.9090113735783028e-05, | |
| "loss": 3.8643, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.189018860026511, | |
| "grad_norm": 1.0634602575628862, | |
| "learning_rate": 1.9072615923009626e-05, | |
| "loss": 3.8727, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.18976302876677287, | |
| "grad_norm": 1.0199100667707268, | |
| "learning_rate": 1.905511811023622e-05, | |
| "loss": 3.8534, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.1905071975070347, | |
| "grad_norm": 1.0896591938046292, | |
| "learning_rate": 1.9037620297462816e-05, | |
| "loss": 3.8771, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.19125136624729658, | |
| "grad_norm": 1.002967767986152, | |
| "learning_rate": 1.9020122484689415e-05, | |
| "loss": 3.8521, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.19199553498755842, | |
| "grad_norm": 0.9962346001278949, | |
| "learning_rate": 1.9002624671916013e-05, | |
| "loss": 3.9142, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.1927397037278203, | |
| "grad_norm": 1.0012651710583431, | |
| "learning_rate": 1.8985126859142608e-05, | |
| "loss": 3.8916, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.19348387246808213, | |
| "grad_norm": 0.996035226541302, | |
| "learning_rate": 1.8967629046369206e-05, | |
| "loss": 3.8357, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.194228041208344, | |
| "grad_norm": 1.066900772364287, | |
| "learning_rate": 1.89501312335958e-05, | |
| "loss": 3.8662, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.19497220994860584, | |
| "grad_norm": 0.9617199906790946, | |
| "learning_rate": 1.89326334208224e-05, | |
| "loss": 3.9121, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.1957163786888677, | |
| "grad_norm": 0.9906912403247032, | |
| "learning_rate": 1.8915135608048995e-05, | |
| "loss": 3.8664, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.19646054742912955, | |
| "grad_norm": 1.0191639183560413, | |
| "learning_rate": 1.8897637795275593e-05, | |
| "loss": 3.8374, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.19720471616939142, | |
| "grad_norm": 0.9586133729145878, | |
| "learning_rate": 1.888013998250219e-05, | |
| "loss": 3.847, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.19794888490965326, | |
| "grad_norm": 1.0575916580617752, | |
| "learning_rate": 1.8862642169728787e-05, | |
| "loss": 3.8834, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.19869305364991513, | |
| "grad_norm": 1.0294047706672742, | |
| "learning_rate": 1.8845144356955382e-05, | |
| "loss": 3.8477, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.19943722239017697, | |
| "grad_norm": 1.010678236243312, | |
| "learning_rate": 1.882764654418198e-05, | |
| "loss": 3.8553, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.20018139113043884, | |
| "grad_norm": 1.0577025095744086, | |
| "learning_rate": 1.8810148731408575e-05, | |
| "loss": 3.8823, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.20092555987070068, | |
| "grad_norm": 1.0128353534644974, | |
| "learning_rate": 1.879265091863517e-05, | |
| "loss": 3.8689, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.20166972861096252, | |
| "grad_norm": 0.9490557753959974, | |
| "learning_rate": 1.877515310586177e-05, | |
| "loss": 3.8495, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.2024138973512244, | |
| "grad_norm": 1.021426802980482, | |
| "learning_rate": 1.8757655293088367e-05, | |
| "loss": 3.8716, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.20315806609148623, | |
| "grad_norm": 1.0508105936689038, | |
| "learning_rate": 1.8740157480314962e-05, | |
| "loss": 3.8883, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.2039022348317481, | |
| "grad_norm": 0.9908159724843973, | |
| "learning_rate": 1.8722659667541557e-05, | |
| "loss": 3.8433, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.20464640357200994, | |
| "grad_norm": 0.9822654130662549, | |
| "learning_rate": 1.8705161854768156e-05, | |
| "loss": 3.845, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.2053905723122718, | |
| "grad_norm": 1.0235735438121543, | |
| "learning_rate": 1.8687664041994754e-05, | |
| "loss": 3.8143, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.20613474105253365, | |
| "grad_norm": 1.0777376830856633, | |
| "learning_rate": 1.867016622922135e-05, | |
| "loss": 3.8522, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.20687890979279552, | |
| "grad_norm": 0.9509750335824687, | |
| "learning_rate": 1.8652668416447944e-05, | |
| "loss": 3.8426, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.20762307853305736, | |
| "grad_norm": 1.0352386200369608, | |
| "learning_rate": 1.8635170603674542e-05, | |
| "loss": 3.8617, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.20836724727331923, | |
| "grad_norm": 1.0137217012340711, | |
| "learning_rate": 1.8617672790901137e-05, | |
| "loss": 3.8773, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.20911141601358108, | |
| "grad_norm": 0.9781047524239378, | |
| "learning_rate": 1.8600174978127736e-05, | |
| "loss": 3.8504, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.20985558475384294, | |
| "grad_norm": 0.912774361475487, | |
| "learning_rate": 1.858267716535433e-05, | |
| "loss": 3.811, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.21059975349410479, | |
| "grad_norm": 0.9514914941163664, | |
| "learning_rate": 1.856517935258093e-05, | |
| "loss": 3.8215, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.21134392223436665, | |
| "grad_norm": 1.0064067729864992, | |
| "learning_rate": 1.8547681539807524e-05, | |
| "loss": 3.8769, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.2120880909746285, | |
| "grad_norm": 0.9414676630740266, | |
| "learning_rate": 1.8530183727034123e-05, | |
| "loss": 3.8823, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.21283225971489036, | |
| "grad_norm": 1.0259064161379845, | |
| "learning_rate": 1.851268591426072e-05, | |
| "loss": 3.8583, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.2135764284551522, | |
| "grad_norm": 0.9869337022008011, | |
| "learning_rate": 1.8495188101487316e-05, | |
| "loss": 3.8383, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.21432059719541405, | |
| "grad_norm": 1.008334446459325, | |
| "learning_rate": 1.847769028871391e-05, | |
| "loss": 3.8475, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.21506476593567592, | |
| "grad_norm": 0.9597002003760106, | |
| "learning_rate": 1.846019247594051e-05, | |
| "loss": 3.8507, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.21580893467593776, | |
| "grad_norm": 0.9763029386561717, | |
| "learning_rate": 1.8442694663167108e-05, | |
| "loss": 3.8381, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.21655310341619963, | |
| "grad_norm": 0.9722245409891449, | |
| "learning_rate": 1.8425196850393703e-05, | |
| "loss": 3.818, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.21729727215646147, | |
| "grad_norm": 1.0084103663323618, | |
| "learning_rate": 1.8407699037620298e-05, | |
| "loss": 3.8432, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.21804144089672334, | |
| "grad_norm": 0.9803183375223226, | |
| "learning_rate": 1.8390201224846896e-05, | |
| "loss": 3.8625, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.21878560963698518, | |
| "grad_norm": 0.9821195446082526, | |
| "learning_rate": 1.837270341207349e-05, | |
| "loss": 3.8274, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.21952977837724705, | |
| "grad_norm": 0.9561554154411983, | |
| "learning_rate": 1.835520559930009e-05, | |
| "loss": 3.8391, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.2202739471175089, | |
| "grad_norm": 1.0080169273302677, | |
| "learning_rate": 1.8337707786526685e-05, | |
| "loss": 3.8387, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.22101811585777076, | |
| "grad_norm": 0.9365439054047329, | |
| "learning_rate": 1.8320209973753283e-05, | |
| "loss": 3.8395, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.2217622845980326, | |
| "grad_norm": 0.9566068516646122, | |
| "learning_rate": 1.830271216097988e-05, | |
| "loss": 3.8311, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.22250645333829447, | |
| "grad_norm": 0.9809445508275404, | |
| "learning_rate": 1.8285214348206473e-05, | |
| "loss": 3.8423, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.2232506220785563, | |
| "grad_norm": 0.996778230346756, | |
| "learning_rate": 1.8267716535433072e-05, | |
| "loss": 3.8599, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.22399479081881818, | |
| "grad_norm": 1.0219966391264999, | |
| "learning_rate": 1.825021872265967e-05, | |
| "loss": 3.8603, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.22473895955908002, | |
| "grad_norm": 0.9500010861925763, | |
| "learning_rate": 1.8232720909886265e-05, | |
| "loss": 3.824, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.2254831282993419, | |
| "grad_norm": 0.9245280909799222, | |
| "learning_rate": 1.821522309711286e-05, | |
| "loss": 3.7937, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.22622729703960373, | |
| "grad_norm": 0.9353631270587811, | |
| "learning_rate": 1.819772528433946e-05, | |
| "loss": 3.8239, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.22697146577986557, | |
| "grad_norm": 1.1676600172924565, | |
| "learning_rate": 1.8180227471566057e-05, | |
| "loss": 3.8399, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.22771563452012744, | |
| "grad_norm": 0.9417402551191273, | |
| "learning_rate": 1.8162729658792652e-05, | |
| "loss": 3.8348, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.22845980326038928, | |
| "grad_norm": 0.9741963709755517, | |
| "learning_rate": 1.814523184601925e-05, | |
| "loss": 3.8279, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.22920397200065115, | |
| "grad_norm": 0.9618250929918145, | |
| "learning_rate": 1.8127734033245846e-05, | |
| "loss": 3.8222, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.229948140740913, | |
| "grad_norm": 0.9925656091589423, | |
| "learning_rate": 1.811023622047244e-05, | |
| "loss": 3.8729, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.23069230948117486, | |
| "grad_norm": 0.9799458103917179, | |
| "learning_rate": 1.809273840769904e-05, | |
| "loss": 3.8393, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.2314364782214367, | |
| "grad_norm": 0.9895334302302228, | |
| "learning_rate": 1.8075240594925637e-05, | |
| "loss": 3.8563, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.23218064696169857, | |
| "grad_norm": 0.9415421869224918, | |
| "learning_rate": 1.8057742782152232e-05, | |
| "loss": 3.8297, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.2329248157019604, | |
| "grad_norm": 0.9328098871714081, | |
| "learning_rate": 1.8040244969378827e-05, | |
| "loss": 3.8063, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.23366898444222228, | |
| "grad_norm": 0.9511674392032746, | |
| "learning_rate": 1.8022747156605426e-05, | |
| "loss": 3.8153, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.23441315318248412, | |
| "grad_norm": 0.957234329473841, | |
| "learning_rate": 1.8005249343832024e-05, | |
| "loss": 3.8295, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.235157321922746, | |
| "grad_norm": 0.9715760658208371, | |
| "learning_rate": 1.798775153105862e-05, | |
| "loss": 3.8279, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.23590149066300783, | |
| "grad_norm": 0.97769159593366, | |
| "learning_rate": 1.7970253718285214e-05, | |
| "loss": 3.8354, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.2366456594032697, | |
| "grad_norm": 0.9346484566202329, | |
| "learning_rate": 1.7952755905511813e-05, | |
| "loss": 3.7971, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.23738982814353154, | |
| "grad_norm": 0.9705099334979708, | |
| "learning_rate": 1.793525809273841e-05, | |
| "loss": 3.8564, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.23813399688379341, | |
| "grad_norm": 1.1099301539791957, | |
| "learning_rate": 1.7917760279965006e-05, | |
| "loss": 3.8433, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.23887816562405526, | |
| "grad_norm": 0.9551539639565874, | |
| "learning_rate": 1.79002624671916e-05, | |
| "loss": 3.8432, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.2396223343643171, | |
| "grad_norm": 0.9701221306559589, | |
| "learning_rate": 1.78827646544182e-05, | |
| "loss": 3.8389, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.24036650310457897, | |
| "grad_norm": 0.9554314456364148, | |
| "learning_rate": 1.7865266841644795e-05, | |
| "loss": 3.8319, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.2411106718448408, | |
| "grad_norm": 0.9383545116848294, | |
| "learning_rate": 1.7847769028871393e-05, | |
| "loss": 3.8066, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.24185484058510268, | |
| "grad_norm": 0.9701025751910185, | |
| "learning_rate": 1.7830271216097988e-05, | |
| "loss": 3.8136, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.24259900932536452, | |
| "grad_norm": 0.9844490795639008, | |
| "learning_rate": 1.7812773403324587e-05, | |
| "loss": 3.8305, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.2433431780656264, | |
| "grad_norm": 1.005549071084577, | |
| "learning_rate": 1.779527559055118e-05, | |
| "loss": 3.8188, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.24408734680588823, | |
| "grad_norm": 0.9708372972977228, | |
| "learning_rate": 1.7777777777777777e-05, | |
| "loss": 3.817, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.2448315155461501, | |
| "grad_norm": 0.9683854193265783, | |
| "learning_rate": 1.7760279965004375e-05, | |
| "loss": 3.866, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.24557568428641194, | |
| "grad_norm": 0.9492825740287544, | |
| "learning_rate": 1.7742782152230973e-05, | |
| "loss": 3.8014, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.2463198530266738, | |
| "grad_norm": 0.9608240879918067, | |
| "learning_rate": 1.772528433945757e-05, | |
| "loss": 3.8203, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.24706402176693565, | |
| "grad_norm": 0.9792629860236627, | |
| "learning_rate": 1.7707786526684167e-05, | |
| "loss": 3.8158, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.24780819050719752, | |
| "grad_norm": 0.9367234899477028, | |
| "learning_rate": 1.7690288713910762e-05, | |
| "loss": 3.8162, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.24855235924745936, | |
| "grad_norm": 0.9507748684637178, | |
| "learning_rate": 1.767279090113736e-05, | |
| "loss": 3.8289, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.24929652798772123, | |
| "grad_norm": 0.9716744099237822, | |
| "learning_rate": 1.7655293088363955e-05, | |
| "loss": 3.8119, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.25004069672798307, | |
| "grad_norm": 0.93112642802389, | |
| "learning_rate": 1.7637795275590554e-05, | |
| "loss": 3.8328, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.2507848654682449, | |
| "grad_norm": 0.9858118569043646, | |
| "learning_rate": 1.762029746281715e-05, | |
| "loss": 3.8218, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.25152903420850675, | |
| "grad_norm": 0.9805460730995765, | |
| "learning_rate": 1.7602799650043747e-05, | |
| "loss": 3.8313, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.25227320294876865, | |
| "grad_norm": 0.9442803485035726, | |
| "learning_rate": 1.7585301837270342e-05, | |
| "loss": 3.8267, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.2530173716890305, | |
| "grad_norm": 0.9536842069799742, | |
| "learning_rate": 1.756780402449694e-05, | |
| "loss": 3.817, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.25376154042929233, | |
| "grad_norm": 0.9720643665777429, | |
| "learning_rate": 1.7550306211723536e-05, | |
| "loss": 3.8408, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.2545057091695542, | |
| "grad_norm": 0.9611314016677492, | |
| "learning_rate": 1.753280839895013e-05, | |
| "loss": 3.7951, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.25524987790981607, | |
| "grad_norm": 0.9475559333772892, | |
| "learning_rate": 1.751531058617673e-05, | |
| "loss": 3.7945, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.2559940466500779, | |
| "grad_norm": 0.9605355040964934, | |
| "learning_rate": 1.7497812773403328e-05, | |
| "loss": 3.8345, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.25673821539033975, | |
| "grad_norm": 1.003535227647403, | |
| "learning_rate": 1.7480314960629923e-05, | |
| "loss": 3.8322, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.2574823841306016, | |
| "grad_norm": 1.009164516519548, | |
| "learning_rate": 1.7462817147856518e-05, | |
| "loss": 3.8355, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.2582265528708635, | |
| "grad_norm": 0.9807594504747283, | |
| "learning_rate": 1.7445319335083116e-05, | |
| "loss": 3.7864, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.25897072161112533, | |
| "grad_norm": 0.9409257276808156, | |
| "learning_rate": 1.7427821522309714e-05, | |
| "loss": 3.7952, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.2597148903513872, | |
| "grad_norm": 1.0244125797583925, | |
| "learning_rate": 1.741032370953631e-05, | |
| "loss": 3.83, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.260459059091649, | |
| "grad_norm": 0.9951698994611993, | |
| "learning_rate": 1.7392825896762904e-05, | |
| "loss": 3.858, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2612032278319109, | |
| "grad_norm": 0.9806922973506748, | |
| "learning_rate": 1.7375328083989503e-05, | |
| "loss": 3.7881, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.26194739657217275, | |
| "grad_norm": 0.984847523036345, | |
| "learning_rate": 1.7357830271216098e-05, | |
| "loss": 3.8272, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.2626915653124346, | |
| "grad_norm": 0.9792691855962944, | |
| "learning_rate": 1.7340332458442696e-05, | |
| "loss": 3.7709, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.26343573405269644, | |
| "grad_norm": 0.967021079137601, | |
| "learning_rate": 1.7322834645669295e-05, | |
| "loss": 3.8213, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.2641799027929583, | |
| "grad_norm": 0.9748972959061579, | |
| "learning_rate": 1.730533683289589e-05, | |
| "loss": 3.8653, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.2649240715332202, | |
| "grad_norm": 0.9476925673020904, | |
| "learning_rate": 1.7287839020122485e-05, | |
| "loss": 3.8135, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.265668240273482, | |
| "grad_norm": 0.951502832746964, | |
| "learning_rate": 1.7270341207349083e-05, | |
| "loss": 3.7835, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.26641240901374386, | |
| "grad_norm": 0.9630370910506539, | |
| "learning_rate": 1.725284339457568e-05, | |
| "loss": 3.8145, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.2671565777540057, | |
| "grad_norm": 0.9330174001044523, | |
| "learning_rate": 1.7235345581802277e-05, | |
| "loss": 3.822, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.2679007464942676, | |
| "grad_norm": 0.9761850719824813, | |
| "learning_rate": 1.721784776902887e-05, | |
| "loss": 3.8138, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.26864491523452944, | |
| "grad_norm": 0.9833932122929374, | |
| "learning_rate": 1.720034995625547e-05, | |
| "loss": 3.8238, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.2693890839747913, | |
| "grad_norm": 0.9369826731076517, | |
| "learning_rate": 1.718285214348207e-05, | |
| "loss": 3.7953, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.2701332527150531, | |
| "grad_norm": 0.9335849679421255, | |
| "learning_rate": 1.7165354330708663e-05, | |
| "loss": 3.7838, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.270877421455315, | |
| "grad_norm": 0.9762929217169171, | |
| "learning_rate": 1.714785651793526e-05, | |
| "loss": 3.8413, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.27162159019557686, | |
| "grad_norm": 0.9827434941126122, | |
| "learning_rate": 1.7130358705161857e-05, | |
| "loss": 3.8285, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.2723657589358387, | |
| "grad_norm": 0.9170865514853238, | |
| "learning_rate": 1.7112860892388452e-05, | |
| "loss": 3.8116, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.27310992767610054, | |
| "grad_norm": 0.9532632558724176, | |
| "learning_rate": 1.709536307961505e-05, | |
| "loss": 3.8196, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.27385409641636244, | |
| "grad_norm": 0.9803488922020432, | |
| "learning_rate": 1.7077865266841645e-05, | |
| "loss": 3.7641, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.2745982651566243, | |
| "grad_norm": 1.002427031417992, | |
| "learning_rate": 1.7060367454068244e-05, | |
| "loss": 3.8342, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.2753424338968861, | |
| "grad_norm": 0.958447228313314, | |
| "learning_rate": 1.704286964129484e-05, | |
| "loss": 3.7685, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.27608660263714796, | |
| "grad_norm": 0.9404991317392366, | |
| "learning_rate": 1.7025371828521434e-05, | |
| "loss": 3.808, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.2768307713774098, | |
| "grad_norm": 0.9875389612468736, | |
| "learning_rate": 1.7007874015748032e-05, | |
| "loss": 3.7922, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.2775749401176717, | |
| "grad_norm": 0.9515246471349947, | |
| "learning_rate": 1.699037620297463e-05, | |
| "loss": 3.7999, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.27831910885793354, | |
| "grad_norm": 0.9319490110821051, | |
| "learning_rate": 1.6972878390201226e-05, | |
| "loss": 3.7861, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.2790632775981954, | |
| "grad_norm": 1.0434293926890705, | |
| "learning_rate": 1.695538057742782e-05, | |
| "loss": 3.8314, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.2798074463384572, | |
| "grad_norm": 0.9520259607128255, | |
| "learning_rate": 1.693788276465442e-05, | |
| "loss": 3.8132, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.2805516150787191, | |
| "grad_norm": 1.068501040936156, | |
| "learning_rate": 1.6920384951881018e-05, | |
| "loss": 3.8213, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.28129578381898096, | |
| "grad_norm": 0.9531491825635828, | |
| "learning_rate": 1.6902887139107613e-05, | |
| "loss": 3.7791, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.2820399525592428, | |
| "grad_norm": 1.0312513466488231, | |
| "learning_rate": 1.688538932633421e-05, | |
| "loss": 3.7959, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.28278412129950464, | |
| "grad_norm": 0.9902039686579069, | |
| "learning_rate": 1.6867891513560806e-05, | |
| "loss": 3.8191, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.28352829003976654, | |
| "grad_norm": 0.9963337480602784, | |
| "learning_rate": 1.68503937007874e-05, | |
| "loss": 3.7965, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.2842724587800284, | |
| "grad_norm": 0.9759590219193287, | |
| "learning_rate": 1.6832895888014e-05, | |
| "loss": 3.7743, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.2850166275202902, | |
| "grad_norm": 0.9605408266457366, | |
| "learning_rate": 1.6815398075240598e-05, | |
| "loss": 3.8125, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.28576079626055206, | |
| "grad_norm": 0.9570391628220699, | |
| "learning_rate": 1.6797900262467193e-05, | |
| "loss": 3.8301, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.28650496500081396, | |
| "grad_norm": 1.0068456268067085, | |
| "learning_rate": 1.6780402449693788e-05, | |
| "loss": 3.8519, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.2872491337410758, | |
| "grad_norm": 0.9403733609852776, | |
| "learning_rate": 1.6762904636920386e-05, | |
| "loss": 3.8299, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.28799330248133764, | |
| "grad_norm": 0.9730361354303355, | |
| "learning_rate": 1.6745406824146985e-05, | |
| "loss": 3.8336, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.2887374712215995, | |
| "grad_norm": 0.9952669975221121, | |
| "learning_rate": 1.672790901137358e-05, | |
| "loss": 3.8152, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.2894816399618613, | |
| "grad_norm": 0.9804272160629643, | |
| "learning_rate": 1.6710411198600175e-05, | |
| "loss": 3.806, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.2902258087021232, | |
| "grad_norm": 1.0098979335848115, | |
| "learning_rate": 1.6692913385826773e-05, | |
| "loss": 3.809, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.29096997744238506, | |
| "grad_norm": 1.0025553767384172, | |
| "learning_rate": 1.667541557305337e-05, | |
| "loss": 3.8192, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.2917141461826469, | |
| "grad_norm": 1.0303619821013217, | |
| "learning_rate": 1.6657917760279967e-05, | |
| "loss": 3.7828, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.29245831492290875, | |
| "grad_norm": 0.9217411357903023, | |
| "learning_rate": 1.6640419947506562e-05, | |
| "loss": 3.7729, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.29320248366317064, | |
| "grad_norm": 0.9448735713336124, | |
| "learning_rate": 1.662292213473316e-05, | |
| "loss": 3.7839, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.2939466524034325, | |
| "grad_norm": 1.021720584912747, | |
| "learning_rate": 1.6605424321959755e-05, | |
| "loss": 3.808, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.2946908211436943, | |
| "grad_norm": 0.9559575771046955, | |
| "learning_rate": 1.6587926509186354e-05, | |
| "loss": 3.8294, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.29543498988395617, | |
| "grad_norm": 0.9398919831924532, | |
| "learning_rate": 1.657042869641295e-05, | |
| "loss": 3.8027, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.29617915862421806, | |
| "grad_norm": 0.951425114817361, | |
| "learning_rate": 1.6552930883639547e-05, | |
| "loss": 3.8497, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.2969233273644799, | |
| "grad_norm": 0.9252600876035637, | |
| "learning_rate": 1.6535433070866142e-05, | |
| "loss": 3.7777, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.29766749610474175, | |
| "grad_norm": 1.0192622035625145, | |
| "learning_rate": 1.651793525809274e-05, | |
| "loss": 3.8578, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.2984116648450036, | |
| "grad_norm": 0.9576726481255868, | |
| "learning_rate": 1.650043744531934e-05, | |
| "loss": 3.8026, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.2991558335852655, | |
| "grad_norm": 0.9511629980234442, | |
| "learning_rate": 1.6482939632545934e-05, | |
| "loss": 3.7691, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.2999000023255273, | |
| "grad_norm": 0.9608208478510485, | |
| "learning_rate": 1.646544181977253e-05, | |
| "loss": 3.7922, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.30064417106578917, | |
| "grad_norm": 0.9967867012275484, | |
| "learning_rate": 1.6447944006999127e-05, | |
| "loss": 3.7713, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.301388339806051, | |
| "grad_norm": 0.9498476123824345, | |
| "learning_rate": 1.6430446194225722e-05, | |
| "loss": 3.8003, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.30213250854631285, | |
| "grad_norm": 1.024134460096829, | |
| "learning_rate": 1.641294838145232e-05, | |
| "loss": 3.8153, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.30287667728657475, | |
| "grad_norm": 0.9722587118015115, | |
| "learning_rate": 1.6395450568678916e-05, | |
| "loss": 3.7734, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.3036208460268366, | |
| "grad_norm": 0.962795292422573, | |
| "learning_rate": 1.6377952755905514e-05, | |
| "loss": 3.8208, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.30436501476709843, | |
| "grad_norm": 0.9843260599547673, | |
| "learning_rate": 1.636045494313211e-05, | |
| "loss": 3.7989, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.30510918350736027, | |
| "grad_norm": 0.9567270986792265, | |
| "learning_rate": 1.6342957130358708e-05, | |
| "loss": 3.7955, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.30585335224762217, | |
| "grad_norm": 0.9459031543867924, | |
| "learning_rate": 1.6325459317585303e-05, | |
| "loss": 3.7832, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.306597520987884, | |
| "grad_norm": 0.9520781659689155, | |
| "learning_rate": 1.63079615048119e-05, | |
| "loss": 3.7955, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.30734168972814585, | |
| "grad_norm": 0.9668587125905648, | |
| "learning_rate": 1.6290463692038496e-05, | |
| "loss": 3.7695, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.3080858584684077, | |
| "grad_norm": 0.9523697048350952, | |
| "learning_rate": 1.627296587926509e-05, | |
| "loss": 3.839, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.3088300272086696, | |
| "grad_norm": 0.9342685197245886, | |
| "learning_rate": 1.625546806649169e-05, | |
| "loss": 3.8372, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.30957419594893143, | |
| "grad_norm": 0.966741891470495, | |
| "learning_rate": 1.6237970253718288e-05, | |
| "loss": 3.8398, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.31031836468919327, | |
| "grad_norm": 0.92938190601994, | |
| "learning_rate": 1.6220472440944883e-05, | |
| "loss": 3.8017, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.3110625334294551, | |
| "grad_norm": 0.9287876995591162, | |
| "learning_rate": 1.6202974628171478e-05, | |
| "loss": 3.812, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.311806702169717, | |
| "grad_norm": 0.9216202563270554, | |
| "learning_rate": 1.6185476815398076e-05, | |
| "loss": 3.7826, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.31255087090997885, | |
| "grad_norm": 0.9165097806683913, | |
| "learning_rate": 1.6167979002624675e-05, | |
| "loss": 3.8153, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.3132950396502407, | |
| "grad_norm": 0.9402474753615747, | |
| "learning_rate": 1.615048118985127e-05, | |
| "loss": 3.8631, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.31403920839050253, | |
| "grad_norm": 0.9781273181958364, | |
| "learning_rate": 1.6132983377077865e-05, | |
| "loss": 3.7967, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.3147833771307644, | |
| "grad_norm": 0.9460589528067613, | |
| "learning_rate": 1.6115485564304463e-05, | |
| "loss": 3.7985, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.31552754587102627, | |
| "grad_norm": 0.9714250213043516, | |
| "learning_rate": 1.609798775153106e-05, | |
| "loss": 3.8166, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.3162717146112881, | |
| "grad_norm": 0.9829307370971013, | |
| "learning_rate": 1.6080489938757657e-05, | |
| "loss": 3.8165, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.31701588335154995, | |
| "grad_norm": 0.9251032591358892, | |
| "learning_rate": 1.6062992125984255e-05, | |
| "loss": 3.7634, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.3177600520918118, | |
| "grad_norm": 1.0860715473556946, | |
| "learning_rate": 1.604549431321085e-05, | |
| "loss": 3.7986, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.3185042208320737, | |
| "grad_norm": 1.0129028451443784, | |
| "learning_rate": 1.6027996500437445e-05, | |
| "loss": 3.8347, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.31924838957233553, | |
| "grad_norm": 0.9649138844021552, | |
| "learning_rate": 1.6010498687664044e-05, | |
| "loss": 3.7858, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.3199925583125974, | |
| "grad_norm": 0.9977728439833579, | |
| "learning_rate": 1.5993000874890642e-05, | |
| "loss": 3.7886, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.3207367270528592, | |
| "grad_norm": 0.9718918964826525, | |
| "learning_rate": 1.5975503062117237e-05, | |
| "loss": 3.8128, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.3214808957931211, | |
| "grad_norm": 0.9323266257911378, | |
| "learning_rate": 1.5958005249343832e-05, | |
| "loss": 3.8095, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.32222506453338295, | |
| "grad_norm": 0.9664946571852979, | |
| "learning_rate": 1.594050743657043e-05, | |
| "loss": 3.81, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.3229692332736448, | |
| "grad_norm": 0.9773888950754641, | |
| "learning_rate": 1.592300962379703e-05, | |
| "loss": 3.7752, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.32371340201390664, | |
| "grad_norm": 0.9933759208621031, | |
| "learning_rate": 1.5905511811023624e-05, | |
| "loss": 3.8152, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.32445757075416853, | |
| "grad_norm": 0.9679243482081298, | |
| "learning_rate": 1.588801399825022e-05, | |
| "loss": 3.8008, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.3252017394944304, | |
| "grad_norm": 1.0016374759708353, | |
| "learning_rate": 1.5870516185476817e-05, | |
| "loss": 3.8252, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.3259459082346922, | |
| "grad_norm": 1.0395890601012985, | |
| "learning_rate": 1.5853018372703412e-05, | |
| "loss": 3.7948, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.32669007697495406, | |
| "grad_norm": 1.0157254292561269, | |
| "learning_rate": 1.583552055993001e-05, | |
| "loss": 3.8016, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.3274342457152159, | |
| "grad_norm": 0.9230482453498372, | |
| "learning_rate": 1.5818022747156606e-05, | |
| "loss": 3.7616, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3281784144554778, | |
| "grad_norm": 0.9954450314658444, | |
| "learning_rate": 1.5800524934383204e-05, | |
| "loss": 3.8016, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.32892258319573964, | |
| "grad_norm": 0.9337022396007288, | |
| "learning_rate": 1.57830271216098e-05, | |
| "loss": 3.7886, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.3296667519360015, | |
| "grad_norm": 0.9401905186033193, | |
| "learning_rate": 1.5765529308836394e-05, | |
| "loss": 3.7729, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.3304109206762633, | |
| "grad_norm": 0.9667790119135654, | |
| "learning_rate": 1.5748031496062993e-05, | |
| "loss": 3.7889, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.3311550894165252, | |
| "grad_norm": 0.9572104283519749, | |
| "learning_rate": 1.573053368328959e-05, | |
| "loss": 3.7678, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.33189925815678706, | |
| "grad_norm": 0.985911712968553, | |
| "learning_rate": 1.5713035870516186e-05, | |
| "loss": 3.7842, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.3326434268970489, | |
| "grad_norm": 0.9318539202302681, | |
| "learning_rate": 1.5695538057742785e-05, | |
| "loss": 3.7943, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.33338759563731074, | |
| "grad_norm": 0.9258857375642625, | |
| "learning_rate": 1.567804024496938e-05, | |
| "loss": 3.7878, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.33413176437757264, | |
| "grad_norm": 1.010698434340366, | |
| "learning_rate": 1.5660542432195978e-05, | |
| "loss": 3.8491, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.3348759331178345, | |
| "grad_norm": 0.9494257580942008, | |
| "learning_rate": 1.5643044619422573e-05, | |
| "loss": 3.7783, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3356201018580963, | |
| "grad_norm": 0.9598513580980321, | |
| "learning_rate": 1.562554680664917e-05, | |
| "loss": 3.8046, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.33636427059835816, | |
| "grad_norm": 0.9493292344115629, | |
| "learning_rate": 1.5608048993875766e-05, | |
| "loss": 3.796, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.33710843933862006, | |
| "grad_norm": 0.9619517089228916, | |
| "learning_rate": 1.559055118110236e-05, | |
| "loss": 3.805, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.3378526080788819, | |
| "grad_norm": 0.9208013559551588, | |
| "learning_rate": 1.557305336832896e-05, | |
| "loss": 3.7853, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.33859677681914374, | |
| "grad_norm": 0.9439981485196373, | |
| "learning_rate": 1.555555555555556e-05, | |
| "loss": 3.8097, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.3393409455594056, | |
| "grad_norm": 0.96027785466224, | |
| "learning_rate": 1.5538057742782153e-05, | |
| "loss": 3.7868, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.3400851142996674, | |
| "grad_norm": 0.965956168946772, | |
| "learning_rate": 1.552055993000875e-05, | |
| "loss": 3.8006, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.3408292830399293, | |
| "grad_norm": 0.9460486085700003, | |
| "learning_rate": 1.5503062117235347e-05, | |
| "loss": 3.8047, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.34157345178019116, | |
| "grad_norm": 0.9929210152749046, | |
| "learning_rate": 1.5485564304461945e-05, | |
| "loss": 3.8073, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.342317620520453, | |
| "grad_norm": 1.0313656786339396, | |
| "learning_rate": 1.546806649168854e-05, | |
| "loss": 3.7894, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.34306178926071484, | |
| "grad_norm": 0.9490812944842836, | |
| "learning_rate": 1.5450568678915135e-05, | |
| "loss": 3.8136, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.34380595800097674, | |
| "grad_norm": 0.9888856525762676, | |
| "learning_rate": 1.5433070866141734e-05, | |
| "loss": 3.7679, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.3445501267412386, | |
| "grad_norm": 1.0035794822656299, | |
| "learning_rate": 1.5415573053368332e-05, | |
| "loss": 3.782, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.3452942954815004, | |
| "grad_norm": 1.0082546014186038, | |
| "learning_rate": 1.5398075240594927e-05, | |
| "loss": 3.7748, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.34603846422176227, | |
| "grad_norm": 0.9958456422128024, | |
| "learning_rate": 1.5380577427821522e-05, | |
| "loss": 3.8027, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.34678263296202416, | |
| "grad_norm": 0.9475959488740303, | |
| "learning_rate": 1.536307961504812e-05, | |
| "loss": 3.7688, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.347526801702286, | |
| "grad_norm": 1.0813850817079405, | |
| "learning_rate": 1.5345581802274716e-05, | |
| "loss": 3.8006, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.34827097044254784, | |
| "grad_norm": 0.9739338163117937, | |
| "learning_rate": 1.5328083989501314e-05, | |
| "loss": 3.7831, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.3490151391828097, | |
| "grad_norm": 1.0133726260753495, | |
| "learning_rate": 1.531058617672791e-05, | |
| "loss": 3.8092, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.3497593079230716, | |
| "grad_norm": 1.0312722572893473, | |
| "learning_rate": 1.5293088363954507e-05, | |
| "loss": 3.7829, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3505034766633334, | |
| "grad_norm": 0.97516265212767, | |
| "learning_rate": 1.5275590551181102e-05, | |
| "loss": 3.7868, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.35124764540359527, | |
| "grad_norm": 0.9423433483864794, | |
| "learning_rate": 1.52580927384077e-05, | |
| "loss": 3.8105, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.3519918141438571, | |
| "grad_norm": 0.9707512410482256, | |
| "learning_rate": 1.5240594925634298e-05, | |
| "loss": 3.7755, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.35273598288411895, | |
| "grad_norm": 1.0138155372961248, | |
| "learning_rate": 1.5223097112860894e-05, | |
| "loss": 3.798, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.35348015162438085, | |
| "grad_norm": 0.9941688770397961, | |
| "learning_rate": 1.520559930008749e-05, | |
| "loss": 3.8154, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.3542243203646427, | |
| "grad_norm": 1.0434098062595074, | |
| "learning_rate": 1.5188101487314086e-05, | |
| "loss": 3.7941, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.3549684891049045, | |
| "grad_norm": 0.9356294811696706, | |
| "learning_rate": 1.5170603674540683e-05, | |
| "loss": 3.7729, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.35571265784516637, | |
| "grad_norm": 0.9417573773786343, | |
| "learning_rate": 1.5153105861767281e-05, | |
| "loss": 3.8063, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.35645682658542827, | |
| "grad_norm": 1.0073140791947333, | |
| "learning_rate": 1.5135608048993878e-05, | |
| "loss": 3.7701, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.3572009953256901, | |
| "grad_norm": 0.9625455703121449, | |
| "learning_rate": 1.5118110236220473e-05, | |
| "loss": 3.7915, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.35794516406595195, | |
| "grad_norm": 0.9876809782187296, | |
| "learning_rate": 1.510061242344707e-05, | |
| "loss": 3.7704, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.3586893328062138, | |
| "grad_norm": 0.9516656440173605, | |
| "learning_rate": 1.5083114610673668e-05, | |
| "loss": 3.7841, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.3594335015464757, | |
| "grad_norm": 0.9834694781125857, | |
| "learning_rate": 1.5065616797900265e-05, | |
| "loss": 3.8076, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.36017767028673753, | |
| "grad_norm": 0.9365474949803332, | |
| "learning_rate": 1.504811898512686e-05, | |
| "loss": 3.7853, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.36092183902699937, | |
| "grad_norm": 0.9569500502686391, | |
| "learning_rate": 1.5030621172353457e-05, | |
| "loss": 3.7832, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.3616660077672612, | |
| "grad_norm": 0.9448678531324651, | |
| "learning_rate": 1.5013123359580053e-05, | |
| "loss": 3.7669, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.3624101765075231, | |
| "grad_norm": 0.9651182833965192, | |
| "learning_rate": 1.4995625546806652e-05, | |
| "loss": 3.7908, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.36315434524778495, | |
| "grad_norm": 0.9253793479231828, | |
| "learning_rate": 1.4978127734033248e-05, | |
| "loss": 3.7622, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.3638985139880468, | |
| "grad_norm": 0.9362082814638715, | |
| "learning_rate": 1.4960629921259843e-05, | |
| "loss": 3.7602, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.36464268272830863, | |
| "grad_norm": 0.933892614825818, | |
| "learning_rate": 1.494313210848644e-05, | |
| "loss": 3.7841, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3653868514685705, | |
| "grad_norm": 0.9222228105776019, | |
| "learning_rate": 1.4925634295713037e-05, | |
| "loss": 3.7563, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.36613102020883237, | |
| "grad_norm": 0.9271089821224759, | |
| "learning_rate": 1.4908136482939635e-05, | |
| "loss": 3.7678, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.3668751889490942, | |
| "grad_norm": 0.9478434554825043, | |
| "learning_rate": 1.489063867016623e-05, | |
| "loss": 3.7753, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.36761935768935605, | |
| "grad_norm": 0.9759073300463119, | |
| "learning_rate": 1.4873140857392827e-05, | |
| "loss": 3.8118, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.3683635264296179, | |
| "grad_norm": 0.9280736352573646, | |
| "learning_rate": 1.4855643044619424e-05, | |
| "loss": 3.7732, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.3691076951698798, | |
| "grad_norm": 0.892166594793543, | |
| "learning_rate": 1.4838145231846019e-05, | |
| "loss": 3.7444, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.36985186391014163, | |
| "grad_norm": 0.9598271946693212, | |
| "learning_rate": 1.4820647419072617e-05, | |
| "loss": 3.7899, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.3705960326504035, | |
| "grad_norm": 0.9259731162718675, | |
| "learning_rate": 1.4803149606299214e-05, | |
| "loss": 3.7805, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.3713402013906653, | |
| "grad_norm": 0.9580328563783095, | |
| "learning_rate": 1.478565179352581e-05, | |
| "loss": 3.8399, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.3720843701309272, | |
| "grad_norm": 0.9354325315456309, | |
| "learning_rate": 1.4768153980752407e-05, | |
| "loss": 3.8173, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.37282853887118905, | |
| "grad_norm": 0.9214794690237775, | |
| "learning_rate": 1.4750656167979002e-05, | |
| "loss": 3.7557, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.3735727076114509, | |
| "grad_norm": 0.9185618712789796, | |
| "learning_rate": 1.47331583552056e-05, | |
| "loss": 3.7596, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.37431687635171274, | |
| "grad_norm": 0.9506157478611675, | |
| "learning_rate": 1.4715660542432198e-05, | |
| "loss": 3.7836, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.37506104509197463, | |
| "grad_norm": 0.8915296706416388, | |
| "learning_rate": 1.4698162729658794e-05, | |
| "loss": 3.783, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.3758052138322365, | |
| "grad_norm": 0.9433799044674582, | |
| "learning_rate": 1.468066491688539e-05, | |
| "loss": 3.7715, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.3765493825724983, | |
| "grad_norm": 1.0140275318280914, | |
| "learning_rate": 1.4663167104111988e-05, | |
| "loss": 3.8057, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.37729355131276016, | |
| "grad_norm": 0.9749884633717492, | |
| "learning_rate": 1.4645669291338584e-05, | |
| "loss": 3.778, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.378037720053022, | |
| "grad_norm": 0.9646416485042524, | |
| "learning_rate": 1.4628171478565181e-05, | |
| "loss": 3.8243, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.3787818887932839, | |
| "grad_norm": 0.925391826009976, | |
| "learning_rate": 1.4610673665791776e-05, | |
| "loss": 3.7428, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.37952605753354574, | |
| "grad_norm": 0.9305303702533825, | |
| "learning_rate": 1.4593175853018373e-05, | |
| "loss": 3.7835, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.3802702262738076, | |
| "grad_norm": 0.9586858488453346, | |
| "learning_rate": 1.4575678040244971e-05, | |
| "loss": 3.7562, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.3810143950140694, | |
| "grad_norm": 0.9470302040277444, | |
| "learning_rate": 1.4558180227471568e-05, | |
| "loss": 3.7721, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.3817585637543313, | |
| "grad_norm": 0.9933248991511777, | |
| "learning_rate": 1.4540682414698165e-05, | |
| "loss": 3.7901, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.38250273249459316, | |
| "grad_norm": 0.9687696405664761, | |
| "learning_rate": 1.452318460192476e-05, | |
| "loss": 3.8056, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.383246901234855, | |
| "grad_norm": 0.9640795694499766, | |
| "learning_rate": 1.4505686789151356e-05, | |
| "loss": 3.78, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.38399106997511684, | |
| "grad_norm": 0.9485837288832386, | |
| "learning_rate": 1.4488188976377955e-05, | |
| "loss": 3.7826, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.38473523871537874, | |
| "grad_norm": 0.9397094999448056, | |
| "learning_rate": 1.4470691163604552e-05, | |
| "loss": 3.7561, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.3854794074556406, | |
| "grad_norm": 0.9428037237200803, | |
| "learning_rate": 1.4453193350831147e-05, | |
| "loss": 3.7889, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.3862235761959024, | |
| "grad_norm": 0.9257624554662638, | |
| "learning_rate": 1.4435695538057743e-05, | |
| "loss": 3.7701, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.38696774493616426, | |
| "grad_norm": 0.9463229158883213, | |
| "learning_rate": 1.441819772528434e-05, | |
| "loss": 3.769, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.38771191367642616, | |
| "grad_norm": 0.9335440776009483, | |
| "learning_rate": 1.4400699912510938e-05, | |
| "loss": 3.765, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.388456082416688, | |
| "grad_norm": 0.9139627100829676, | |
| "learning_rate": 1.4383202099737535e-05, | |
| "loss": 3.8021, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.38920025115694984, | |
| "grad_norm": 0.9115865431810327, | |
| "learning_rate": 1.436570428696413e-05, | |
| "loss": 3.713, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.3899444198972117, | |
| "grad_norm": 0.9630094807256917, | |
| "learning_rate": 1.4348206474190727e-05, | |
| "loss": 3.7946, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.3906885886374735, | |
| "grad_norm": 0.9458356064792427, | |
| "learning_rate": 1.4330708661417324e-05, | |
| "loss": 3.74, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.3914327573777354, | |
| "grad_norm": 0.9112746630556541, | |
| "learning_rate": 1.4313210848643922e-05, | |
| "loss": 3.8015, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.39217692611799726, | |
| "grad_norm": 0.9215709415337336, | |
| "learning_rate": 1.4295713035870517e-05, | |
| "loss": 3.7859, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.3929210948582591, | |
| "grad_norm": 0.9278107665256279, | |
| "learning_rate": 1.4278215223097114e-05, | |
| "loss": 3.7888, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.39366526359852094, | |
| "grad_norm": 0.9370825701249599, | |
| "learning_rate": 1.426071741032371e-05, | |
| "loss": 3.771, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.39440943233878284, | |
| "grad_norm": 0.9687733498431801, | |
| "learning_rate": 1.4243219597550309e-05, | |
| "loss": 3.7751, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3951536010790447, | |
| "grad_norm": 0.9621308515369155, | |
| "learning_rate": 1.4225721784776904e-05, | |
| "loss": 3.7773, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.3958977698193065, | |
| "grad_norm": 0.9415390670131851, | |
| "learning_rate": 1.42082239720035e-05, | |
| "loss": 3.7995, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.39664193855956836, | |
| "grad_norm": 0.9347171689102004, | |
| "learning_rate": 1.4190726159230097e-05, | |
| "loss": 3.7585, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.39738610729983026, | |
| "grad_norm": 0.9313402926291189, | |
| "learning_rate": 1.4173228346456694e-05, | |
| "loss": 3.7496, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.3981302760400921, | |
| "grad_norm": 0.9692686120292682, | |
| "learning_rate": 1.4155730533683293e-05, | |
| "loss": 3.7512, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.39887444478035394, | |
| "grad_norm": 0.9451037152530462, | |
| "learning_rate": 1.4138232720909888e-05, | |
| "loss": 3.8102, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.3996186135206158, | |
| "grad_norm": 0.9172650100774835, | |
| "learning_rate": 1.4120734908136484e-05, | |
| "loss": 3.7457, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.4003627822608777, | |
| "grad_norm": 0.9784240205762622, | |
| "learning_rate": 1.4103237095363081e-05, | |
| "loss": 3.7899, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.4011069510011395, | |
| "grad_norm": 0.9417258737128843, | |
| "learning_rate": 1.4085739282589676e-05, | |
| "loss": 3.8133, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.40185111974140136, | |
| "grad_norm": 0.9448294128582327, | |
| "learning_rate": 1.4068241469816274e-05, | |
| "loss": 3.7319, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.4025952884816632, | |
| "grad_norm": 0.9642189400694554, | |
| "learning_rate": 1.4050743657042871e-05, | |
| "loss": 3.7769, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.40333945722192505, | |
| "grad_norm": 0.967540786812905, | |
| "learning_rate": 1.4033245844269468e-05, | |
| "loss": 3.7546, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.40408362596218694, | |
| "grad_norm": 0.9358118679309165, | |
| "learning_rate": 1.4015748031496063e-05, | |
| "loss": 3.7713, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.4048277947024488, | |
| "grad_norm": 0.9770669617105951, | |
| "learning_rate": 1.399825021872266e-05, | |
| "loss": 3.7593, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.4055719634427106, | |
| "grad_norm": 0.9772696106724039, | |
| "learning_rate": 1.3980752405949258e-05, | |
| "loss": 3.7284, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.40631613218297247, | |
| "grad_norm": 1.0189494718662118, | |
| "learning_rate": 1.3963254593175855e-05, | |
| "loss": 3.7814, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.40706030092323436, | |
| "grad_norm": 0.950842147856222, | |
| "learning_rate": 1.3945756780402451e-05, | |
| "loss": 3.7626, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.4078044696634962, | |
| "grad_norm": 0.9328714293277736, | |
| "learning_rate": 1.3928258967629047e-05, | |
| "loss": 3.7669, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.40854863840375805, | |
| "grad_norm": 0.9934502163863145, | |
| "learning_rate": 1.3910761154855643e-05, | |
| "loss": 3.7612, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.4092928071440199, | |
| "grad_norm": 1.0388812690511289, | |
| "learning_rate": 1.3893263342082242e-05, | |
| "loss": 3.8132, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.4100369758842818, | |
| "grad_norm": 0.9152661126350029, | |
| "learning_rate": 1.3875765529308838e-05, | |
| "loss": 3.7961, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.4107811446245436, | |
| "grad_norm": 0.9174977829855058, | |
| "learning_rate": 1.3858267716535433e-05, | |
| "loss": 3.7541, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.41152531336480547, | |
| "grad_norm": 1.0800210967477823, | |
| "learning_rate": 1.384076990376203e-05, | |
| "loss": 3.8074, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.4122694821050673, | |
| "grad_norm": 0.9325570154755897, | |
| "learning_rate": 1.3823272090988629e-05, | |
| "loss": 3.7669, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.4130136508453292, | |
| "grad_norm": 1.0347319563708866, | |
| "learning_rate": 1.3805774278215225e-05, | |
| "loss": 3.7824, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.41375781958559105, | |
| "grad_norm": 0.9588558780946068, | |
| "learning_rate": 1.378827646544182e-05, | |
| "loss": 3.7217, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.4145019883258529, | |
| "grad_norm": 0.9707926089761055, | |
| "learning_rate": 1.3770778652668417e-05, | |
| "loss": 3.7904, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.41524615706611473, | |
| "grad_norm": 0.9495221955767716, | |
| "learning_rate": 1.3753280839895014e-05, | |
| "loss": 3.7833, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.41599032580637657, | |
| "grad_norm": 0.9413223501380156, | |
| "learning_rate": 1.3735783027121612e-05, | |
| "loss": 3.7486, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.41673449454663847, | |
| "grad_norm": 0.934981281172684, | |
| "learning_rate": 1.3718285214348209e-05, | |
| "loss": 3.7703, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.4174786632869003, | |
| "grad_norm": 0.9265601977150958, | |
| "learning_rate": 1.3700787401574804e-05, | |
| "loss": 3.7445, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.41822283202716215, | |
| "grad_norm": 0.9172634636607314, | |
| "learning_rate": 1.36832895888014e-05, | |
| "loss": 3.7575, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.418967000767424, | |
| "grad_norm": 0.915934911289319, | |
| "learning_rate": 1.3665791776027997e-05, | |
| "loss": 3.7581, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.4197111695076859, | |
| "grad_norm": 0.9041905901291603, | |
| "learning_rate": 1.3648293963254596e-05, | |
| "loss": 3.7299, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.42045533824794773, | |
| "grad_norm": 0.9322497549181782, | |
| "learning_rate": 1.363079615048119e-05, | |
| "loss": 3.7489, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.42119950698820957, | |
| "grad_norm": 0.9723893468768703, | |
| "learning_rate": 1.3613298337707787e-05, | |
| "loss": 3.7886, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.4219436757284714, | |
| "grad_norm": 0.9127801934094371, | |
| "learning_rate": 1.3595800524934384e-05, | |
| "loss": 3.7241, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.4226878444687333, | |
| "grad_norm": 0.9494757307909666, | |
| "learning_rate": 1.357830271216098e-05, | |
| "loss": 3.7456, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.42343201320899515, | |
| "grad_norm": 0.9217902761545226, | |
| "learning_rate": 1.356080489938758e-05, | |
| "loss": 3.7559, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.424176181949257, | |
| "grad_norm": 0.9372868932724301, | |
| "learning_rate": 1.3543307086614174e-05, | |
| "loss": 3.7461, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.42492035068951883, | |
| "grad_norm": 0.903487799144028, | |
| "learning_rate": 1.3525809273840771e-05, | |
| "loss": 3.7373, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.42566451942978073, | |
| "grad_norm": 1.0021462405588577, | |
| "learning_rate": 1.3508311461067368e-05, | |
| "loss": 3.7539, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.42640868817004257, | |
| "grad_norm": 0.9170697635627474, | |
| "learning_rate": 1.3490813648293963e-05, | |
| "loss": 3.7593, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.4271528569103044, | |
| "grad_norm": 0.9780678447111848, | |
| "learning_rate": 1.3473315835520561e-05, | |
| "loss": 3.7466, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.42789702565056625, | |
| "grad_norm": 0.939306813513707, | |
| "learning_rate": 1.3455818022747158e-05, | |
| "loss": 3.7612, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.4286411943908281, | |
| "grad_norm": 0.9121457078551923, | |
| "learning_rate": 1.3438320209973755e-05, | |
| "loss": 3.7744, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.42938536313109, | |
| "grad_norm": 0.984904616833892, | |
| "learning_rate": 1.342082239720035e-05, | |
| "loss": 3.7846, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.43012953187135183, | |
| "grad_norm": 0.8921835163850447, | |
| "learning_rate": 1.3403324584426948e-05, | |
| "loss": 3.7853, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.4308737006116137, | |
| "grad_norm": 0.9944594284213352, | |
| "learning_rate": 1.3385826771653545e-05, | |
| "loss": 3.7975, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.4316178693518755, | |
| "grad_norm": 0.9214531243911438, | |
| "learning_rate": 1.3368328958880142e-05, | |
| "loss": 3.75, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4323620380921374, | |
| "grad_norm": 0.9357426741367952, | |
| "learning_rate": 1.3350831146106738e-05, | |
| "loss": 3.7505, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.43310620683239925, | |
| "grad_norm": 0.8885664966826696, | |
| "learning_rate": 1.3333333333333333e-05, | |
| "loss": 3.771, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.4338503755726611, | |
| "grad_norm": 0.9631367636447117, | |
| "learning_rate": 1.3315835520559932e-05, | |
| "loss": 3.7641, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.43459454431292294, | |
| "grad_norm": 0.9729414336714773, | |
| "learning_rate": 1.3298337707786528e-05, | |
| "loss": 3.7452, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.43533871305318483, | |
| "grad_norm": 0.9656622437569762, | |
| "learning_rate": 1.3280839895013125e-05, | |
| "loss": 3.7818, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.4360828817934467, | |
| "grad_norm": 0.9212738650340485, | |
| "learning_rate": 1.326334208223972e-05, | |
| "loss": 3.7752, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.4368270505337085, | |
| "grad_norm": 1.0004027463636957, | |
| "learning_rate": 1.3245844269466317e-05, | |
| "loss": 3.7376, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.43757121927397036, | |
| "grad_norm": 0.9624166951011305, | |
| "learning_rate": 1.3228346456692915e-05, | |
| "loss": 3.7563, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.43831538801423225, | |
| "grad_norm": 0.920221506643196, | |
| "learning_rate": 1.3210848643919512e-05, | |
| "loss": 3.7269, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.4390595567544941, | |
| "grad_norm": 0.9407118174286582, | |
| "learning_rate": 1.3193350831146107e-05, | |
| "loss": 3.7374, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.43980372549475594, | |
| "grad_norm": 0.988088762408026, | |
| "learning_rate": 1.3175853018372704e-05, | |
| "loss": 3.7444, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.4405478942350178, | |
| "grad_norm": 0.9425880469768513, | |
| "learning_rate": 1.31583552055993e-05, | |
| "loss": 3.7467, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.4412920629752796, | |
| "grad_norm": 0.9561722897127888, | |
| "learning_rate": 1.3140857392825899e-05, | |
| "loss": 3.7767, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.4420362317155415, | |
| "grad_norm": 0.9759470851832611, | |
| "learning_rate": 1.3123359580052496e-05, | |
| "loss": 3.79, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.44278040045580336, | |
| "grad_norm": 0.8985177219017222, | |
| "learning_rate": 1.310586176727909e-05, | |
| "loss": 3.7762, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.4435245691960652, | |
| "grad_norm": 0.9243856222044661, | |
| "learning_rate": 1.3088363954505687e-05, | |
| "loss": 3.7793, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.44426873793632704, | |
| "grad_norm": 0.9633753620949967, | |
| "learning_rate": 1.3070866141732284e-05, | |
| "loss": 3.776, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.44501290667658894, | |
| "grad_norm": 0.9529338097712216, | |
| "learning_rate": 1.3053368328958883e-05, | |
| "loss": 3.7433, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.4457570754168508, | |
| "grad_norm": 0.9697423353697971, | |
| "learning_rate": 1.3035870516185478e-05, | |
| "loss": 3.7931, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.4465012441571126, | |
| "grad_norm": 0.9409054129900818, | |
| "learning_rate": 1.3018372703412074e-05, | |
| "loss": 3.7311, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.44724541289737446, | |
| "grad_norm": 0.96521759555247, | |
| "learning_rate": 1.3000874890638671e-05, | |
| "loss": 3.7619, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.44798958163763636, | |
| "grad_norm": 0.9346366786393363, | |
| "learning_rate": 1.2983377077865266e-05, | |
| "loss": 3.7698, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.4487337503778982, | |
| "grad_norm": 0.938739635878293, | |
| "learning_rate": 1.2965879265091864e-05, | |
| "loss": 3.731, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.44947791911816004, | |
| "grad_norm": 0.9241036716938422, | |
| "learning_rate": 1.2948381452318461e-05, | |
| "loss": 3.737, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.4502220878584219, | |
| "grad_norm": 0.9660482036828824, | |
| "learning_rate": 1.2930883639545058e-05, | |
| "loss": 3.7873, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.4509662565986838, | |
| "grad_norm": 0.9233519670555015, | |
| "learning_rate": 1.2913385826771655e-05, | |
| "loss": 3.7365, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.4517104253389456, | |
| "grad_norm": 0.9344628517479248, | |
| "learning_rate": 1.2895888013998253e-05, | |
| "loss": 3.7717, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.45245459407920746, | |
| "grad_norm": 0.9737586377450018, | |
| "learning_rate": 1.2878390201224848e-05, | |
| "loss": 3.7326, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.4531987628194693, | |
| "grad_norm": 0.9241078674212275, | |
| "learning_rate": 1.2860892388451445e-05, | |
| "loss": 3.7916, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.45394293155973114, | |
| "grad_norm": 0.9151759561152131, | |
| "learning_rate": 1.2843394575678041e-05, | |
| "loss": 3.7169, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.45468710029999304, | |
| "grad_norm": 0.9315595024686493, | |
| "learning_rate": 1.2825896762904636e-05, | |
| "loss": 3.7702, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.4554312690402549, | |
| "grad_norm": 0.927863415922818, | |
| "learning_rate": 1.2808398950131235e-05, | |
| "loss": 3.7497, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.4561754377805167, | |
| "grad_norm": 0.9498495195332954, | |
| "learning_rate": 1.2790901137357832e-05, | |
| "loss": 3.7436, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.45691960652077857, | |
| "grad_norm": 0.9228543633814827, | |
| "learning_rate": 1.2773403324584428e-05, | |
| "loss": 3.763, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.45766377526104046, | |
| "grad_norm": 0.9441116525408387, | |
| "learning_rate": 1.2755905511811025e-05, | |
| "loss": 3.7699, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.4584079440013023, | |
| "grad_norm": 0.9124720958078346, | |
| "learning_rate": 1.273840769903762e-05, | |
| "loss": 3.731, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.45915211274156414, | |
| "grad_norm": 0.969270174124955, | |
| "learning_rate": 1.2720909886264218e-05, | |
| "loss": 3.7369, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.459896281481826, | |
| "grad_norm": 0.9417570689889766, | |
| "learning_rate": 1.2703412073490815e-05, | |
| "loss": 3.7715, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.4606404502220879, | |
| "grad_norm": 0.9694376890653595, | |
| "learning_rate": 1.2685914260717412e-05, | |
| "loss": 3.7164, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.4613846189623497, | |
| "grad_norm": 0.9357401686835035, | |
| "learning_rate": 1.2668416447944007e-05, | |
| "loss": 3.7231, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.46212878770261157, | |
| "grad_norm": 0.9319914795955363, | |
| "learning_rate": 1.2650918635170604e-05, | |
| "loss": 3.7598, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.4628729564428734, | |
| "grad_norm": 0.9647643937665509, | |
| "learning_rate": 1.2633420822397202e-05, | |
| "loss": 3.7607, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.4636171251831353, | |
| "grad_norm": 0.9586340367436607, | |
| "learning_rate": 1.2615923009623799e-05, | |
| "loss": 3.7668, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.46436129392339714, | |
| "grad_norm": 0.9384923146548246, | |
| "learning_rate": 1.2598425196850394e-05, | |
| "loss": 3.7261, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.465105462663659, | |
| "grad_norm": 0.9418499103789868, | |
| "learning_rate": 1.258092738407699e-05, | |
| "loss": 3.716, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.4658496314039208, | |
| "grad_norm": 0.9221070761689966, | |
| "learning_rate": 1.2563429571303587e-05, | |
| "loss": 3.7539, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.46659380014418267, | |
| "grad_norm": 0.9313710862333304, | |
| "learning_rate": 1.2545931758530186e-05, | |
| "loss": 3.7459, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.46733796888444457, | |
| "grad_norm": 0.913666541102956, | |
| "learning_rate": 1.2528433945756782e-05, | |
| "loss": 3.7388, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.4680821376247064, | |
| "grad_norm": 1.0117601404423218, | |
| "learning_rate": 1.2510936132983377e-05, | |
| "loss": 3.761, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.46882630636496825, | |
| "grad_norm": 0.931462165605379, | |
| "learning_rate": 1.2493438320209974e-05, | |
| "loss": 3.7779, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4695704751052301, | |
| "grad_norm": 0.8921798000100868, | |
| "learning_rate": 1.2475940507436573e-05, | |
| "loss": 3.7274, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.470314643845492, | |
| "grad_norm": 0.9833649357167662, | |
| "learning_rate": 1.245844269466317e-05, | |
| "loss": 3.7735, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.4710588125857538, | |
| "grad_norm": 0.9817263283671807, | |
| "learning_rate": 1.2440944881889764e-05, | |
| "loss": 3.7059, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.47180298132601567, | |
| "grad_norm": 0.9468548025361827, | |
| "learning_rate": 1.2423447069116361e-05, | |
| "loss": 3.7305, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.4725471500662775, | |
| "grad_norm": 0.9614019179863994, | |
| "learning_rate": 1.2405949256342958e-05, | |
| "loss": 3.7726, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.4732913188065394, | |
| "grad_norm": 0.9826247724524748, | |
| "learning_rate": 1.2388451443569556e-05, | |
| "loss": 3.7848, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.47403548754680125, | |
| "grad_norm": 0.9394678730738416, | |
| "learning_rate": 1.2370953630796151e-05, | |
| "loss": 3.7803, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.4747796562870631, | |
| "grad_norm": 0.9672498329008435, | |
| "learning_rate": 1.2353455818022748e-05, | |
| "loss": 3.7735, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.47552382502732493, | |
| "grad_norm": 0.9613007807263887, | |
| "learning_rate": 1.2335958005249345e-05, | |
| "loss": 3.7412, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.47626799376758683, | |
| "grad_norm": 0.9284161362855874, | |
| "learning_rate": 1.2318460192475941e-05, | |
| "loss": 3.7538, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.47701216250784867, | |
| "grad_norm": 0.9388027767530401, | |
| "learning_rate": 1.230096237970254e-05, | |
| "loss": 3.7246, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.4777563312481105, | |
| "grad_norm": 0.9209459605261991, | |
| "learning_rate": 1.2283464566929135e-05, | |
| "loss": 3.7795, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.47850049998837235, | |
| "grad_norm": 0.9159627348354152, | |
| "learning_rate": 1.2265966754155732e-05, | |
| "loss": 3.7693, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.4792446687286342, | |
| "grad_norm": 1.06422740213832, | |
| "learning_rate": 1.2248468941382328e-05, | |
| "loss": 3.7476, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.4799888374688961, | |
| "grad_norm": 0.9302132866953204, | |
| "learning_rate": 1.2230971128608923e-05, | |
| "loss": 3.7497, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.48073300620915793, | |
| "grad_norm": 0.9137505158132918, | |
| "learning_rate": 1.2213473315835522e-05, | |
| "loss": 3.7603, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.4814771749494198, | |
| "grad_norm": 0.911111902951911, | |
| "learning_rate": 1.2195975503062118e-05, | |
| "loss": 3.7487, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.4822213436896816, | |
| "grad_norm": 0.9286806220441729, | |
| "learning_rate": 1.2178477690288715e-05, | |
| "loss": 3.7296, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.4829655124299435, | |
| "grad_norm": 0.9309723209596665, | |
| "learning_rate": 1.216097987751531e-05, | |
| "loss": 3.7546, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.48370968117020535, | |
| "grad_norm": 0.9168890468262008, | |
| "learning_rate": 1.2143482064741907e-05, | |
| "loss": 3.7386, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.4844538499104672, | |
| "grad_norm": 0.9576447211774173, | |
| "learning_rate": 1.2125984251968505e-05, | |
| "loss": 3.7617, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.48519801865072903, | |
| "grad_norm": 0.9005824146791948, | |
| "learning_rate": 1.2108486439195102e-05, | |
| "loss": 3.7512, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.48594218739099093, | |
| "grad_norm": 0.9605003380888615, | |
| "learning_rate": 1.2090988626421699e-05, | |
| "loss": 3.766, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.4866863561312528, | |
| "grad_norm": 0.9158571511770911, | |
| "learning_rate": 1.2073490813648294e-05, | |
| "loss": 3.7371, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.4874305248715146, | |
| "grad_norm": 0.9404423107775526, | |
| "learning_rate": 1.2055993000874892e-05, | |
| "loss": 3.6947, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.48817469361177646, | |
| "grad_norm": 0.9630396935995491, | |
| "learning_rate": 1.2038495188101489e-05, | |
| "loss": 3.7194, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.4889188623520383, | |
| "grad_norm": 0.9115954906073751, | |
| "learning_rate": 1.2020997375328086e-05, | |
| "loss": 3.7475, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.4896630310923002, | |
| "grad_norm": 0.9153324795680369, | |
| "learning_rate": 1.200349956255468e-05, | |
| "loss": 3.7315, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.49040719983256204, | |
| "grad_norm": 0.9310455410660707, | |
| "learning_rate": 1.1986001749781277e-05, | |
| "loss": 3.7598, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.4911513685728239, | |
| "grad_norm": 0.9193330735114456, | |
| "learning_rate": 1.1968503937007876e-05, | |
| "loss": 3.7479, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.4918955373130857, | |
| "grad_norm": 0.9639203499584349, | |
| "learning_rate": 1.1951006124234472e-05, | |
| "loss": 3.7287, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.4926397060533476, | |
| "grad_norm": 0.9504104811432326, | |
| "learning_rate": 1.193350831146107e-05, | |
| "loss": 3.7334, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.49338387479360946, | |
| "grad_norm": 0.9842616180005365, | |
| "learning_rate": 1.1916010498687664e-05, | |
| "loss": 3.7497, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.4941280435338713, | |
| "grad_norm": 0.9480670476907875, | |
| "learning_rate": 1.1898512685914261e-05, | |
| "loss": 3.7742, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.49487221227413314, | |
| "grad_norm": 1.0182107506666231, | |
| "learning_rate": 1.188101487314086e-05, | |
| "loss": 3.761, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.49561638101439504, | |
| "grad_norm": 0.9873374240360752, | |
| "learning_rate": 1.1863517060367456e-05, | |
| "loss": 3.7764, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.4963605497546569, | |
| "grad_norm": 0.9505943348582996, | |
| "learning_rate": 1.1846019247594051e-05, | |
| "loss": 3.7927, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.4971047184949187, | |
| "grad_norm": 0.9170703978205829, | |
| "learning_rate": 1.1828521434820648e-05, | |
| "loss": 3.784, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.49784888723518056, | |
| "grad_norm": 0.9080293952409417, | |
| "learning_rate": 1.1811023622047245e-05, | |
| "loss": 3.7128, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.49859305597544246, | |
| "grad_norm": 0.952640559952456, | |
| "learning_rate": 1.1793525809273843e-05, | |
| "loss": 3.7419, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.4993372247157043, | |
| "grad_norm": 0.9588939244992026, | |
| "learning_rate": 1.1776027996500438e-05, | |
| "loss": 3.7713, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.5000813934559661, | |
| "grad_norm": 0.9207359215318365, | |
| "learning_rate": 1.1758530183727035e-05, | |
| "loss": 3.7376, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.500825562196228, | |
| "grad_norm": 0.9134681029350289, | |
| "learning_rate": 1.1741032370953631e-05, | |
| "loss": 3.7317, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.5015697309364898, | |
| "grad_norm": 0.9568293496711773, | |
| "learning_rate": 1.1723534558180228e-05, | |
| "loss": 3.7457, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.5023138996767517, | |
| "grad_norm": 0.9015502539329926, | |
| "learning_rate": 1.1706036745406827e-05, | |
| "loss": 3.756, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.5030580684170135, | |
| "grad_norm": 0.9279677891267879, | |
| "learning_rate": 1.1688538932633422e-05, | |
| "loss": 3.7373, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.5038022371572755, | |
| "grad_norm": 0.9472715150201526, | |
| "learning_rate": 1.1671041119860018e-05, | |
| "loss": 3.7581, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.5045464058975373, | |
| "grad_norm": 0.9416538339047547, | |
| "learning_rate": 1.1653543307086615e-05, | |
| "loss": 3.7965, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.5052905746377991, | |
| "grad_norm": 0.9075181125934488, | |
| "learning_rate": 1.1636045494313213e-05, | |
| "loss": 3.7481, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.506034743378061, | |
| "grad_norm": 0.9304375855060932, | |
| "learning_rate": 1.1618547681539808e-05, | |
| "loss": 3.7622, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.5067789121183228, | |
| "grad_norm": 0.9472195087267425, | |
| "learning_rate": 1.1601049868766405e-05, | |
| "loss": 3.763, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.5075230808585847, | |
| "grad_norm": 0.925724234282432, | |
| "learning_rate": 1.1583552055993002e-05, | |
| "loss": 3.7613, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.5082672495988465, | |
| "grad_norm": 0.9276913831412554, | |
| "learning_rate": 1.1566054243219597e-05, | |
| "loss": 3.7386, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.5090114183391083, | |
| "grad_norm": 0.9885568366975916, | |
| "learning_rate": 1.1548556430446195e-05, | |
| "loss": 3.7436, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.5097555870793703, | |
| "grad_norm": 0.9229622222674994, | |
| "learning_rate": 1.1531058617672792e-05, | |
| "loss": 3.7649, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.5104997558196321, | |
| "grad_norm": 0.945768605368974, | |
| "learning_rate": 1.1513560804899389e-05, | |
| "loss": 3.7418, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.511243924559894, | |
| "grad_norm": 0.9074701723359645, | |
| "learning_rate": 1.1496062992125985e-05, | |
| "loss": 3.7191, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.5119880933001558, | |
| "grad_norm": 0.9271772135729172, | |
| "learning_rate": 1.147856517935258e-05, | |
| "loss": 3.7497, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.5127322620404177, | |
| "grad_norm": 0.9344396498455457, | |
| "learning_rate": 1.1461067366579179e-05, | |
| "loss": 3.7364, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.5134764307806795, | |
| "grad_norm": 0.9142577806432824, | |
| "learning_rate": 1.1443569553805776e-05, | |
| "loss": 3.7593, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.5142205995209413, | |
| "grad_norm": 0.9528470025067902, | |
| "learning_rate": 1.1426071741032372e-05, | |
| "loss": 3.7372, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.5149647682612032, | |
| "grad_norm": 0.963219635752827, | |
| "learning_rate": 1.1408573928258967e-05, | |
| "loss": 3.7455, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.515708937001465, | |
| "grad_norm": 0.9176569009714276, | |
| "learning_rate": 1.1391076115485564e-05, | |
| "loss": 3.7636, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.516453105741727, | |
| "grad_norm": 0.9159648354843738, | |
| "learning_rate": 1.1373578302712163e-05, | |
| "loss": 3.7407, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.5171972744819888, | |
| "grad_norm": 0.9606165501943866, | |
| "learning_rate": 1.135608048993876e-05, | |
| "loss": 3.7179, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.5179414432222507, | |
| "grad_norm": 0.9039458639318599, | |
| "learning_rate": 1.1338582677165354e-05, | |
| "loss": 3.7091, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.5186856119625125, | |
| "grad_norm": 0.9120152018457843, | |
| "learning_rate": 1.1321084864391951e-05, | |
| "loss": 3.7265, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.5194297807027743, | |
| "grad_norm": 0.9280086126476629, | |
| "learning_rate": 1.1303587051618548e-05, | |
| "loss": 3.7774, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.5201739494430362, | |
| "grad_norm": 0.9101955901865356, | |
| "learning_rate": 1.1286089238845146e-05, | |
| "loss": 3.7347, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.520918118183298, | |
| "grad_norm": 0.9364025811887693, | |
| "learning_rate": 1.1268591426071743e-05, | |
| "loss": 3.7343, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.5216622869235599, | |
| "grad_norm": 0.9431452849198947, | |
| "learning_rate": 1.1251093613298338e-05, | |
| "loss": 3.7321, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.5224064556638218, | |
| "grad_norm": 0.9063871496948601, | |
| "learning_rate": 1.1233595800524935e-05, | |
| "loss": 3.7262, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.5231506244040837, | |
| "grad_norm": 0.9357837963387813, | |
| "learning_rate": 1.1216097987751533e-05, | |
| "loss": 3.704, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.5238947931443455, | |
| "grad_norm": 0.9056255141892088, | |
| "learning_rate": 1.119860017497813e-05, | |
| "loss": 3.7469, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.5246389618846073, | |
| "grad_norm": 0.9000693227345116, | |
| "learning_rate": 1.1181102362204725e-05, | |
| "loss": 3.7881, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.5253831306248692, | |
| "grad_norm": 0.9248081763994404, | |
| "learning_rate": 1.1163604549431321e-05, | |
| "loss": 3.7097, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.526127299365131, | |
| "grad_norm": 0.9256921414384724, | |
| "learning_rate": 1.1146106736657918e-05, | |
| "loss": 3.7275, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.5268714681053929, | |
| "grad_norm": 0.9135316122763312, | |
| "learning_rate": 1.1128608923884517e-05, | |
| "loss": 3.7357, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.5276156368456547, | |
| "grad_norm": 0.9035298722874847, | |
| "learning_rate": 1.1111111111111113e-05, | |
| "loss": 3.7334, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.5283598055859166, | |
| "grad_norm": 0.933760369579414, | |
| "learning_rate": 1.1093613298337708e-05, | |
| "loss": 3.7421, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.5291039743261785, | |
| "grad_norm": 0.925795201730903, | |
| "learning_rate": 1.1076115485564305e-05, | |
| "loss": 3.742, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.5298481430664403, | |
| "grad_norm": 0.9604396246465967, | |
| "learning_rate": 1.1058617672790902e-05, | |
| "loss": 3.738, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.5305923118067022, | |
| "grad_norm": 0.9694216752784578, | |
| "learning_rate": 1.10411198600175e-05, | |
| "loss": 3.776, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.531336480546964, | |
| "grad_norm": 0.9156863865574238, | |
| "learning_rate": 1.1023622047244095e-05, | |
| "loss": 3.7454, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.5320806492872259, | |
| "grad_norm": 0.9658152416493923, | |
| "learning_rate": 1.1006124234470692e-05, | |
| "loss": 3.7479, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.5328248180274877, | |
| "grad_norm": 0.9512479551271446, | |
| "learning_rate": 1.0988626421697289e-05, | |
| "loss": 3.71, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.5335689867677496, | |
| "grad_norm": 0.9684122865854226, | |
| "learning_rate": 1.0971128608923884e-05, | |
| "loss": 3.7535, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.5343131555080114, | |
| "grad_norm": 0.9151195363592776, | |
| "learning_rate": 1.0953630796150482e-05, | |
| "loss": 3.7068, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.5350573242482733, | |
| "grad_norm": 0.9605145845115862, | |
| "learning_rate": 1.0936132983377079e-05, | |
| "loss": 3.7048, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.5358014929885352, | |
| "grad_norm": 0.9481075359354455, | |
| "learning_rate": 1.0918635170603676e-05, | |
| "loss": 3.7233, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.536545661728797, | |
| "grad_norm": 0.9509865541580967, | |
| "learning_rate": 1.0901137357830272e-05, | |
| "loss": 3.7001, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.5372898304690589, | |
| "grad_norm": 0.92881500935682, | |
| "learning_rate": 1.0883639545056867e-05, | |
| "loss": 3.7645, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.5380339992093207, | |
| "grad_norm": 0.9312599300542342, | |
| "learning_rate": 1.0866141732283466e-05, | |
| "loss": 3.7172, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.5387781679495826, | |
| "grad_norm": 0.9302006437950018, | |
| "learning_rate": 1.0848643919510062e-05, | |
| "loss": 3.7163, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.5395223366898444, | |
| "grad_norm": 0.8934779844753681, | |
| "learning_rate": 1.083114610673666e-05, | |
| "loss": 3.7302, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.5402665054301062, | |
| "grad_norm": 0.9365588231773585, | |
| "learning_rate": 1.0813648293963254e-05, | |
| "loss": 3.7007, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.5410106741703681, | |
| "grad_norm": 0.9521219028514314, | |
| "learning_rate": 1.0796150481189853e-05, | |
| "loss": 3.7344, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.54175484291063, | |
| "grad_norm": 0.9004330062872257, | |
| "learning_rate": 1.077865266841645e-05, | |
| "loss": 3.7465, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.5424990116508919, | |
| "grad_norm": 0.8928306756007266, | |
| "learning_rate": 1.0761154855643046e-05, | |
| "loss": 3.6962, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.5432431803911537, | |
| "grad_norm": 0.9153584287292633, | |
| "learning_rate": 1.0743657042869641e-05, | |
| "loss": 3.7406, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5439873491314156, | |
| "grad_norm": 0.9361979806946165, | |
| "learning_rate": 1.0726159230096238e-05, | |
| "loss": 3.7035, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.5447315178716774, | |
| "grad_norm": 0.924966317191546, | |
| "learning_rate": 1.0708661417322836e-05, | |
| "loss": 3.738, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.5454756866119392, | |
| "grad_norm": 0.9289879701835152, | |
| "learning_rate": 1.0691163604549433e-05, | |
| "loss": 3.748, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.5462198553522011, | |
| "grad_norm": 0.9013099195629167, | |
| "learning_rate": 1.067366579177603e-05, | |
| "loss": 3.7406, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.5469640240924629, | |
| "grad_norm": 0.9505895659980985, | |
| "learning_rate": 1.0656167979002625e-05, | |
| "loss": 3.7032, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.5477081928327249, | |
| "grad_norm": 0.902011954635937, | |
| "learning_rate": 1.0638670166229221e-05, | |
| "loss": 3.7291, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.5484523615729867, | |
| "grad_norm": 0.9252747740350574, | |
| "learning_rate": 1.062117235345582e-05, | |
| "loss": 3.7234, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.5491965303132486, | |
| "grad_norm": 0.9363400909948054, | |
| "learning_rate": 1.0603674540682417e-05, | |
| "loss": 3.7371, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.5499406990535104, | |
| "grad_norm": 0.9238195663451927, | |
| "learning_rate": 1.0586176727909012e-05, | |
| "loss": 3.7636, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.5506848677937722, | |
| "grad_norm": 0.9113632981856744, | |
| "learning_rate": 1.0568678915135608e-05, | |
| "loss": 3.749, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5514290365340341, | |
| "grad_norm": 0.8896824758183292, | |
| "learning_rate": 1.0551181102362205e-05, | |
| "loss": 3.7417, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.5521732052742959, | |
| "grad_norm": 0.9539076399850199, | |
| "learning_rate": 1.0533683289588803e-05, | |
| "loss": 3.7455, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.5529173740145578, | |
| "grad_norm": 0.92800474073068, | |
| "learning_rate": 1.0516185476815398e-05, | |
| "loss": 3.705, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.5536615427548196, | |
| "grad_norm": 0.9190704786808062, | |
| "learning_rate": 1.0498687664041995e-05, | |
| "loss": 3.7387, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.5544057114950816, | |
| "grad_norm": 0.9137325400583608, | |
| "learning_rate": 1.0481189851268592e-05, | |
| "loss": 3.7166, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.5551498802353434, | |
| "grad_norm": 0.9353365932683673, | |
| "learning_rate": 1.0463692038495189e-05, | |
| "loss": 3.7242, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.5558940489756052, | |
| "grad_norm": 0.9758849814415886, | |
| "learning_rate": 1.0446194225721787e-05, | |
| "loss": 3.7412, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.5566382177158671, | |
| "grad_norm": 0.9177795598973589, | |
| "learning_rate": 1.0428696412948382e-05, | |
| "loss": 3.7457, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.5573823864561289, | |
| "grad_norm": 0.9551279336408571, | |
| "learning_rate": 1.0411198600174979e-05, | |
| "loss": 3.7107, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.5581265551963908, | |
| "grad_norm": 0.916184904991062, | |
| "learning_rate": 1.0393700787401575e-05, | |
| "loss": 3.7384, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5588707239366526, | |
| "grad_norm": 0.9388673337219728, | |
| "learning_rate": 1.0376202974628174e-05, | |
| "loss": 3.7211, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.5596148926769144, | |
| "grad_norm": 0.9603288434834489, | |
| "learning_rate": 1.0358705161854769e-05, | |
| "loss": 3.7061, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.5603590614171764, | |
| "grad_norm": 0.9749645614168967, | |
| "learning_rate": 1.0341207349081366e-05, | |
| "loss": 3.7262, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.5611032301574382, | |
| "grad_norm": 0.9321688595868419, | |
| "learning_rate": 1.0323709536307962e-05, | |
| "loss": 3.722, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.5618473988977001, | |
| "grad_norm": 0.9007702365377109, | |
| "learning_rate": 1.0306211723534559e-05, | |
| "loss": 3.689, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.5625915676379619, | |
| "grad_norm": 0.9409191582212306, | |
| "learning_rate": 1.0288713910761157e-05, | |
| "loss": 3.7166, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.5633357363782238, | |
| "grad_norm": 0.9344729460263849, | |
| "learning_rate": 1.0271216097987753e-05, | |
| "loss": 3.7301, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.5640799051184856, | |
| "grad_norm": 0.9897715644314247, | |
| "learning_rate": 1.025371828521435e-05, | |
| "loss": 3.7647, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.5648240738587474, | |
| "grad_norm": 0.913176772236781, | |
| "learning_rate": 1.0236220472440946e-05, | |
| "loss": 3.7179, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.5655682425990093, | |
| "grad_norm": 0.9054842771903394, | |
| "learning_rate": 1.0218722659667541e-05, | |
| "loss": 3.7164, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5663124113392711, | |
| "grad_norm": 0.8988217225732424, | |
| "learning_rate": 1.020122484689414e-05, | |
| "loss": 3.7498, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.5670565800795331, | |
| "grad_norm": 0.9239317613898159, | |
| "learning_rate": 1.0183727034120736e-05, | |
| "loss": 3.7142, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.5678007488197949, | |
| "grad_norm": 0.9025486779524126, | |
| "learning_rate": 1.0166229221347333e-05, | |
| "loss": 3.7281, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.5685449175600568, | |
| "grad_norm": 0.9506232124837497, | |
| "learning_rate": 1.0148731408573928e-05, | |
| "loss": 3.7455, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.5692890863003186, | |
| "grad_norm": 0.9229530164265863, | |
| "learning_rate": 1.0131233595800525e-05, | |
| "loss": 3.7407, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.5700332550405804, | |
| "grad_norm": 0.9027536696699909, | |
| "learning_rate": 1.0113735783027123e-05, | |
| "loss": 3.7262, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.5707774237808423, | |
| "grad_norm": 0.9091013235142209, | |
| "learning_rate": 1.009623797025372e-05, | |
| "loss": 3.7083, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.5715215925211041, | |
| "grad_norm": 0.8986314661940508, | |
| "learning_rate": 1.0078740157480316e-05, | |
| "loss": 3.7506, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.572265761261366, | |
| "grad_norm": 0.9042282195089449, | |
| "learning_rate": 1.0061242344706911e-05, | |
| "loss": 3.7413, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.5730099300016279, | |
| "grad_norm": 0.9142227953696717, | |
| "learning_rate": 1.0043744531933508e-05, | |
| "loss": 3.7322, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5737540987418898, | |
| "grad_norm": 0.9436860250164415, | |
| "learning_rate": 1.0026246719160107e-05, | |
| "loss": 3.7175, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.5744982674821516, | |
| "grad_norm": 0.9097465988015286, | |
| "learning_rate": 1.0008748906386703e-05, | |
| "loss": 3.719, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.5752424362224134, | |
| "grad_norm": 0.9208203705406401, | |
| "learning_rate": 9.991251093613298e-06, | |
| "loss": 3.7552, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.5759866049626753, | |
| "grad_norm": 0.8965581633177859, | |
| "learning_rate": 9.973753280839897e-06, | |
| "loss": 3.7528, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.5767307737029371, | |
| "grad_norm": 0.8750757510691709, | |
| "learning_rate": 9.956255468066492e-06, | |
| "loss": 3.7016, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.577474942443199, | |
| "grad_norm": 0.9184794057834814, | |
| "learning_rate": 9.938757655293088e-06, | |
| "loss": 3.6861, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.5782191111834608, | |
| "grad_norm": 0.9192061382939967, | |
| "learning_rate": 9.921259842519685e-06, | |
| "loss": 3.7026, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.5789632799237227, | |
| "grad_norm": 1.0156643634698408, | |
| "learning_rate": 9.903762029746282e-06, | |
| "loss": 3.7396, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.5797074486639846, | |
| "grad_norm": 0.944319479058479, | |
| "learning_rate": 9.88626421697288e-06, | |
| "loss": 3.7333, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.5804516174042464, | |
| "grad_norm": 0.9617821479618149, | |
| "learning_rate": 9.868766404199475e-06, | |
| "loss": 3.7035, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5811957861445083, | |
| "grad_norm": 0.9252939350546395, | |
| "learning_rate": 9.851268591426074e-06, | |
| "loss": 3.7262, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.5819399548847701, | |
| "grad_norm": 0.9668067290521408, | |
| "learning_rate": 9.833770778652669e-06, | |
| "loss": 3.7201, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.582684123625032, | |
| "grad_norm": 0.9303293578382623, | |
| "learning_rate": 9.816272965879266e-06, | |
| "loss": 3.7375, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.5834282923652938, | |
| "grad_norm": 0.8992125004479886, | |
| "learning_rate": 9.798775153105862e-06, | |
| "loss": 3.7378, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.5841724611055557, | |
| "grad_norm": 0.949955010115643, | |
| "learning_rate": 9.781277340332459e-06, | |
| "loss": 3.7568, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.5849166298458175, | |
| "grad_norm": 0.9032055127831073, | |
| "learning_rate": 9.763779527559056e-06, | |
| "loss": 3.7134, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.5856607985860794, | |
| "grad_norm": 0.9294922438216326, | |
| "learning_rate": 9.746281714785652e-06, | |
| "loss": 3.713, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.5864049673263413, | |
| "grad_norm": 0.9527664282364697, | |
| "learning_rate": 9.728783902012249e-06, | |
| "loss": 3.718, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.5871491360666031, | |
| "grad_norm": 0.9601557194278608, | |
| "learning_rate": 9.711286089238846e-06, | |
| "loss": 3.7305, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.587893304806865, | |
| "grad_norm": 0.9002330141376512, | |
| "learning_rate": 9.693788276465443e-06, | |
| "loss": 3.7225, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.5886374735471268, | |
| "grad_norm": 0.9633135192880754, | |
| "learning_rate": 9.67629046369204e-06, | |
| "loss": 3.7016, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.5893816422873887, | |
| "grad_norm": 0.9256761135030225, | |
| "learning_rate": 9.658792650918636e-06, | |
| "loss": 3.705, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.5901258110276505, | |
| "grad_norm": 0.9276707571542945, | |
| "learning_rate": 9.641294838145233e-06, | |
| "loss": 3.7065, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.5908699797679123, | |
| "grad_norm": 0.941998069929586, | |
| "learning_rate": 9.62379702537183e-06, | |
| "loss": 3.704, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.5916141485081742, | |
| "grad_norm": 0.953839816941259, | |
| "learning_rate": 9.606299212598426e-06, | |
| "loss": 3.6879, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.5923583172484361, | |
| "grad_norm": 0.9452355933313026, | |
| "learning_rate": 9.588801399825023e-06, | |
| "loss": 3.7444, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.593102485988698, | |
| "grad_norm": 0.9200215263827904, | |
| "learning_rate": 9.57130358705162e-06, | |
| "loss": 3.6834, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.5938466547289598, | |
| "grad_norm": 0.9115106775758931, | |
| "learning_rate": 9.553805774278216e-06, | |
| "loss": 3.6969, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.5945908234692217, | |
| "grad_norm": 0.911666589182346, | |
| "learning_rate": 9.536307961504813e-06, | |
| "loss": 3.7275, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.5953349922094835, | |
| "grad_norm": 0.959267213400646, | |
| "learning_rate": 9.518810148731408e-06, | |
| "loss": 3.7569, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5960791609497453, | |
| "grad_norm": 0.8869706993026042, | |
| "learning_rate": 9.501312335958006e-06, | |
| "loss": 3.6992, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.5968233296900072, | |
| "grad_norm": 0.9277677538509237, | |
| "learning_rate": 9.483814523184603e-06, | |
| "loss": 3.7346, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.597567498430269, | |
| "grad_norm": 0.8855285985405903, | |
| "learning_rate": 9.4663167104112e-06, | |
| "loss": 3.7058, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.598311667170531, | |
| "grad_norm": 0.9183731553117692, | |
| "learning_rate": 9.448818897637797e-06, | |
| "loss": 3.7138, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.5990558359107928, | |
| "grad_norm": 0.9198156347923907, | |
| "learning_rate": 9.431321084864393e-06, | |
| "loss": 3.7756, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.5998000046510547, | |
| "grad_norm": 0.8947375816975488, | |
| "learning_rate": 9.41382327209099e-06, | |
| "loss": 3.6909, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.6005441733913165, | |
| "grad_norm": 0.930560179638984, | |
| "learning_rate": 9.396325459317585e-06, | |
| "loss": 3.7482, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.6012883421315783, | |
| "grad_norm": 0.9086007636622732, | |
| "learning_rate": 9.378827646544184e-06, | |
| "loss": 3.6865, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.6020325108718402, | |
| "grad_norm": 0.9602548132656676, | |
| "learning_rate": 9.361329833770779e-06, | |
| "loss": 3.7376, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.602776679612102, | |
| "grad_norm": 0.9138073914897307, | |
| "learning_rate": 9.343832020997377e-06, | |
| "loss": 3.7301, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.6035208483523639, | |
| "grad_norm": 0.9472220804487634, | |
| "learning_rate": 9.326334208223972e-06, | |
| "loss": 3.7031, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.6042650170926257, | |
| "grad_norm": 0.909410304705648, | |
| "learning_rate": 9.308836395450569e-06, | |
| "loss": 3.743, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.6050091858328877, | |
| "grad_norm": 0.927232340225224, | |
| "learning_rate": 9.291338582677165e-06, | |
| "loss": 3.7046, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.6057533545731495, | |
| "grad_norm": 0.9102430244774036, | |
| "learning_rate": 9.273840769903762e-06, | |
| "loss": 3.7227, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.6064975233134113, | |
| "grad_norm": 0.9016131647290525, | |
| "learning_rate": 9.25634295713036e-06, | |
| "loss": 3.6795, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.6072416920536732, | |
| "grad_norm": 0.8808541595115517, | |
| "learning_rate": 9.238845144356956e-06, | |
| "loss": 3.7188, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.607985860793935, | |
| "grad_norm": 0.900394365485463, | |
| "learning_rate": 9.221347331583554e-06, | |
| "loss": 3.7, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.6087300295341969, | |
| "grad_norm": 0.9153313204975121, | |
| "learning_rate": 9.203849518810149e-06, | |
| "loss": 3.6639, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.6094741982744587, | |
| "grad_norm": 0.950402062129457, | |
| "learning_rate": 9.186351706036746e-06, | |
| "loss": 3.7226, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.6102183670147205, | |
| "grad_norm": 0.8955657281292623, | |
| "learning_rate": 9.168853893263342e-06, | |
| "loss": 3.7252, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.6109625357549825, | |
| "grad_norm": 0.8935857763328318, | |
| "learning_rate": 9.15135608048994e-06, | |
| "loss": 3.7157, | |
| "step": 821 | |
| }, | |
| { | |
| "epoch": 0.6117067044952443, | |
| "grad_norm": 0.9164256999788148, | |
| "learning_rate": 9.133858267716536e-06, | |
| "loss": 3.6938, | |
| "step": 822 | |
| }, | |
| { | |
| "epoch": 0.6124508732355062, | |
| "grad_norm": 0.9171063167108389, | |
| "learning_rate": 9.116360454943133e-06, | |
| "loss": 3.7206, | |
| "step": 823 | |
| }, | |
| { | |
| "epoch": 0.613195041975768, | |
| "grad_norm": 0.9363022543050231, | |
| "learning_rate": 9.09886264216973e-06, | |
| "loss": 3.7228, | |
| "step": 824 | |
| }, | |
| { | |
| "epoch": 0.6139392107160299, | |
| "grad_norm": 0.9429821932677903, | |
| "learning_rate": 9.081364829396326e-06, | |
| "loss": 3.6802, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.6146833794562917, | |
| "grad_norm": 0.8896368099551419, | |
| "learning_rate": 9.063867016622923e-06, | |
| "loss": 3.755, | |
| "step": 826 | |
| }, | |
| { | |
| "epoch": 0.6154275481965535, | |
| "grad_norm": 0.9283389476882261, | |
| "learning_rate": 9.04636920384952e-06, | |
| "loss": 3.6993, | |
| "step": 827 | |
| }, | |
| { | |
| "epoch": 0.6161717169368154, | |
| "grad_norm": 0.9159281570475902, | |
| "learning_rate": 9.028871391076116e-06, | |
| "loss": 3.7345, | |
| "step": 828 | |
| }, | |
| { | |
| "epoch": 0.6169158856770772, | |
| "grad_norm": 0.8993685615604555, | |
| "learning_rate": 9.011373578302713e-06, | |
| "loss": 3.676, | |
| "step": 829 | |
| }, | |
| { | |
| "epoch": 0.6176600544173392, | |
| "grad_norm": 0.8973158551748408, | |
| "learning_rate": 8.99387576552931e-06, | |
| "loss": 3.7546, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.618404223157601, | |
| "grad_norm": 0.9003350121231588, | |
| "learning_rate": 8.976377952755906e-06, | |
| "loss": 3.7, | |
| "step": 831 | |
| }, | |
| { | |
| "epoch": 0.6191483918978629, | |
| "grad_norm": 0.9433922147833915, | |
| "learning_rate": 8.958880139982503e-06, | |
| "loss": 3.7546, | |
| "step": 832 | |
| }, | |
| { | |
| "epoch": 0.6198925606381247, | |
| "grad_norm": 0.8948142144943422, | |
| "learning_rate": 8.9413823272091e-06, | |
| "loss": 3.6832, | |
| "step": 833 | |
| }, | |
| { | |
| "epoch": 0.6206367293783865, | |
| "grad_norm": 0.8918044707716455, | |
| "learning_rate": 8.923884514435697e-06, | |
| "loss": 3.7072, | |
| "step": 834 | |
| }, | |
| { | |
| "epoch": 0.6213808981186484, | |
| "grad_norm": 0.9018339255968875, | |
| "learning_rate": 8.906386701662293e-06, | |
| "loss": 3.703, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.6221250668589102, | |
| "grad_norm": 0.912807726141547, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 3.6993, | |
| "step": 836 | |
| }, | |
| { | |
| "epoch": 0.6228692355991721, | |
| "grad_norm": 0.8994784062079151, | |
| "learning_rate": 8.871391076115487e-06, | |
| "loss": 3.7215, | |
| "step": 837 | |
| }, | |
| { | |
| "epoch": 0.623613404339434, | |
| "grad_norm": 0.9441616051162164, | |
| "learning_rate": 8.853893263342083e-06, | |
| "loss": 3.7283, | |
| "step": 838 | |
| }, | |
| { | |
| "epoch": 0.6243575730796959, | |
| "grad_norm": 0.9368548513744808, | |
| "learning_rate": 8.83639545056868e-06, | |
| "loss": 3.7068, | |
| "step": 839 | |
| }, | |
| { | |
| "epoch": 0.6251017418199577, | |
| "grad_norm": 0.9295527708448562, | |
| "learning_rate": 8.818897637795277e-06, | |
| "loss": 3.697, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.6258459105602195, | |
| "grad_norm": 0.8990157704648177, | |
| "learning_rate": 8.801399825021874e-06, | |
| "loss": 3.6796, | |
| "step": 841 | |
| }, | |
| { | |
| "epoch": 0.6265900793004814, | |
| "grad_norm": 0.913097259339662, | |
| "learning_rate": 8.78390201224847e-06, | |
| "loss": 3.7228, | |
| "step": 842 | |
| }, | |
| { | |
| "epoch": 0.6273342480407432, | |
| "grad_norm": 0.9082538146873894, | |
| "learning_rate": 8.766404199475065e-06, | |
| "loss": 3.7228, | |
| "step": 843 | |
| }, | |
| { | |
| "epoch": 0.6280784167810051, | |
| "grad_norm": 0.8799771588365104, | |
| "learning_rate": 8.748906386701664e-06, | |
| "loss": 3.7223, | |
| "step": 844 | |
| }, | |
| { | |
| "epoch": 0.6288225855212669, | |
| "grad_norm": 0.9119189917975884, | |
| "learning_rate": 8.731408573928259e-06, | |
| "loss": 3.7286, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.6295667542615287, | |
| "grad_norm": 0.910939599334156, | |
| "learning_rate": 8.713910761154857e-06, | |
| "loss": 3.7229, | |
| "step": 846 | |
| }, | |
| { | |
| "epoch": 0.6303109230017907, | |
| "grad_norm": 0.8728036052752017, | |
| "learning_rate": 8.696412948381452e-06, | |
| "loss": 3.7265, | |
| "step": 847 | |
| }, | |
| { | |
| "epoch": 0.6310550917420525, | |
| "grad_norm": 0.9005241998425194, | |
| "learning_rate": 8.678915135608049e-06, | |
| "loss": 3.6731, | |
| "step": 848 | |
| }, | |
| { | |
| "epoch": 0.6317992604823144, | |
| "grad_norm": 0.9112201219767515, | |
| "learning_rate": 8.661417322834647e-06, | |
| "loss": 3.7169, | |
| "step": 849 | |
| }, | |
| { | |
| "epoch": 0.6325434292225762, | |
| "grad_norm": 0.9076314781080989, | |
| "learning_rate": 8.643919510061242e-06, | |
| "loss": 3.7032, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.6332875979628381, | |
| "grad_norm": 0.9159328937651239, | |
| "learning_rate": 8.62642169728784e-06, | |
| "loss": 3.748, | |
| "step": 851 | |
| }, | |
| { | |
| "epoch": 0.6340317667030999, | |
| "grad_norm": 0.934808762707829, | |
| "learning_rate": 8.608923884514436e-06, | |
| "loss": 3.6878, | |
| "step": 852 | |
| }, | |
| { | |
| "epoch": 0.6347759354433617, | |
| "grad_norm": 0.9431154130053773, | |
| "learning_rate": 8.591426071741034e-06, | |
| "loss": 3.7212, | |
| "step": 853 | |
| }, | |
| { | |
| "epoch": 0.6355201041836236, | |
| "grad_norm": 0.9235193806328037, | |
| "learning_rate": 8.57392825896763e-06, | |
| "loss": 3.7019, | |
| "step": 854 | |
| }, | |
| { | |
| "epoch": 0.6362642729238855, | |
| "grad_norm": 0.9214381157660402, | |
| "learning_rate": 8.556430446194226e-06, | |
| "loss": 3.7135, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.6370084416641474, | |
| "grad_norm": 0.9803703184196864, | |
| "learning_rate": 8.538932633420823e-06, | |
| "loss": 3.7112, | |
| "step": 856 | |
| }, | |
| { | |
| "epoch": 0.6377526104044092, | |
| "grad_norm": 0.8938253887932968, | |
| "learning_rate": 8.52143482064742e-06, | |
| "loss": 3.7152, | |
| "step": 857 | |
| }, | |
| { | |
| "epoch": 0.6384967791446711, | |
| "grad_norm": 0.8874313979194299, | |
| "learning_rate": 8.503937007874016e-06, | |
| "loss": 3.698, | |
| "step": 858 | |
| }, | |
| { | |
| "epoch": 0.6392409478849329, | |
| "grad_norm": 0.9356600895818273, | |
| "learning_rate": 8.486439195100613e-06, | |
| "loss": 3.6936, | |
| "step": 859 | |
| }, | |
| { | |
| "epoch": 0.6399851166251947, | |
| "grad_norm": 0.9012879724167883, | |
| "learning_rate": 8.46894138232721e-06, | |
| "loss": 3.7094, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.6407292853654566, | |
| "grad_norm": 0.9136993292419924, | |
| "learning_rate": 8.451443569553806e-06, | |
| "loss": 3.7233, | |
| "step": 861 | |
| }, | |
| { | |
| "epoch": 0.6414734541057184, | |
| "grad_norm": 0.9076177011954273, | |
| "learning_rate": 8.433945756780403e-06, | |
| "loss": 3.746, | |
| "step": 862 | |
| }, | |
| { | |
| "epoch": 0.6422176228459803, | |
| "grad_norm": 0.9192980176972465, | |
| "learning_rate": 8.416447944007e-06, | |
| "loss": 3.6806, | |
| "step": 863 | |
| }, | |
| { | |
| "epoch": 0.6429617915862422, | |
| "grad_norm": 0.9182289008818338, | |
| "learning_rate": 8.398950131233596e-06, | |
| "loss": 3.6945, | |
| "step": 864 | |
| }, | |
| { | |
| "epoch": 0.6437059603265041, | |
| "grad_norm": 0.925202449216382, | |
| "learning_rate": 8.381452318460193e-06, | |
| "loss": 3.7008, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.6444501290667659, | |
| "grad_norm": 0.9173560248984937, | |
| "learning_rate": 8.36395450568679e-06, | |
| "loss": 3.7266, | |
| "step": 866 | |
| }, | |
| { | |
| "epoch": 0.6451942978070278, | |
| "grad_norm": 0.8975345795506934, | |
| "learning_rate": 8.346456692913387e-06, | |
| "loss": 3.6902, | |
| "step": 867 | |
| }, | |
| { | |
| "epoch": 0.6459384665472896, | |
| "grad_norm": 0.9496601762637903, | |
| "learning_rate": 8.328958880139983e-06, | |
| "loss": 3.7116, | |
| "step": 868 | |
| }, | |
| { | |
| "epoch": 0.6466826352875514, | |
| "grad_norm": 0.9646405934049285, | |
| "learning_rate": 8.31146106736658e-06, | |
| "loss": 3.6975, | |
| "step": 869 | |
| }, | |
| { | |
| "epoch": 0.6474268040278133, | |
| "grad_norm": 0.9160358537967095, | |
| "learning_rate": 8.293963254593177e-06, | |
| "loss": 3.6889, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.6481709727680751, | |
| "grad_norm": 0.930531864832298, | |
| "learning_rate": 8.276465441819773e-06, | |
| "loss": 3.7127, | |
| "step": 871 | |
| }, | |
| { | |
| "epoch": 0.6489151415083371, | |
| "grad_norm": 0.9400347238840481, | |
| "learning_rate": 8.25896762904637e-06, | |
| "loss": 3.7592, | |
| "step": 872 | |
| }, | |
| { | |
| "epoch": 0.6496593102485989, | |
| "grad_norm": 0.9537818082272422, | |
| "learning_rate": 8.241469816272967e-06, | |
| "loss": 3.7461, | |
| "step": 873 | |
| }, | |
| { | |
| "epoch": 0.6504034789888608, | |
| "grad_norm": 0.9222999487881888, | |
| "learning_rate": 8.223972003499564e-06, | |
| "loss": 3.6858, | |
| "step": 874 | |
| }, | |
| { | |
| "epoch": 0.6511476477291226, | |
| "grad_norm": 0.9133646154913314, | |
| "learning_rate": 8.20647419072616e-06, | |
| "loss": 3.6918, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.6518918164693844, | |
| "grad_norm": 0.9423472846878602, | |
| "learning_rate": 8.188976377952757e-06, | |
| "loss": 3.669, | |
| "step": 876 | |
| }, | |
| { | |
| "epoch": 0.6526359852096463, | |
| "grad_norm": 0.9322265033366255, | |
| "learning_rate": 8.171478565179354e-06, | |
| "loss": 3.7628, | |
| "step": 877 | |
| }, | |
| { | |
| "epoch": 0.6533801539499081, | |
| "grad_norm": 0.9379114556667095, | |
| "learning_rate": 8.15398075240595e-06, | |
| "loss": 3.7485, | |
| "step": 878 | |
| }, | |
| { | |
| "epoch": 0.65412432269017, | |
| "grad_norm": 0.9500242930021386, | |
| "learning_rate": 8.136482939632546e-06, | |
| "loss": 3.7286, | |
| "step": 879 | |
| }, | |
| { | |
| "epoch": 0.6548684914304318, | |
| "grad_norm": 0.9424555203013725, | |
| "learning_rate": 8.118985126859144e-06, | |
| "loss": 3.738, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6556126601706938, | |
| "grad_norm": 0.9028126905241344, | |
| "learning_rate": 8.101487314085739e-06, | |
| "loss": 3.7128, | |
| "step": 881 | |
| }, | |
| { | |
| "epoch": 0.6563568289109556, | |
| "grad_norm": 0.9473014956449676, | |
| "learning_rate": 8.083989501312337e-06, | |
| "loss": 3.6902, | |
| "step": 882 | |
| }, | |
| { | |
| "epoch": 0.6571009976512174, | |
| "grad_norm": 0.9448582973983615, | |
| "learning_rate": 8.066491688538932e-06, | |
| "loss": 3.6938, | |
| "step": 883 | |
| }, | |
| { | |
| "epoch": 0.6578451663914793, | |
| "grad_norm": 0.9423506165605101, | |
| "learning_rate": 8.04899387576553e-06, | |
| "loss": 3.7202, | |
| "step": 884 | |
| }, | |
| { | |
| "epoch": 0.6585893351317411, | |
| "grad_norm": 0.9077629427004515, | |
| "learning_rate": 8.031496062992128e-06, | |
| "loss": 3.6962, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.659333503872003, | |
| "grad_norm": 0.8939401637500247, | |
| "learning_rate": 8.013998250218723e-06, | |
| "loss": 3.7175, | |
| "step": 886 | |
| }, | |
| { | |
| "epoch": 0.6600776726122648, | |
| "grad_norm": 0.9116021968357324, | |
| "learning_rate": 7.996500437445321e-06, | |
| "loss": 3.6994, | |
| "step": 887 | |
| }, | |
| { | |
| "epoch": 0.6608218413525266, | |
| "grad_norm": 0.932706571648462, | |
| "learning_rate": 7.979002624671916e-06, | |
| "loss": 3.6955, | |
| "step": 888 | |
| }, | |
| { | |
| "epoch": 0.6615660100927886, | |
| "grad_norm": 0.9533149363124757, | |
| "learning_rate": 7.961504811898514e-06, | |
| "loss": 3.6762, | |
| "step": 889 | |
| }, | |
| { | |
| "epoch": 0.6623101788330504, | |
| "grad_norm": 0.9381809803059193, | |
| "learning_rate": 7.94400699912511e-06, | |
| "loss": 3.7087, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6630543475733123, | |
| "grad_norm": 0.9070013459492722, | |
| "learning_rate": 7.926509186351706e-06, | |
| "loss": 3.6965, | |
| "step": 891 | |
| }, | |
| { | |
| "epoch": 0.6637985163135741, | |
| "grad_norm": 0.9640644563804528, | |
| "learning_rate": 7.909011373578303e-06, | |
| "loss": 3.6849, | |
| "step": 892 | |
| }, | |
| { | |
| "epoch": 0.664542685053836, | |
| "grad_norm": 0.9661069862451881, | |
| "learning_rate": 7.8915135608049e-06, | |
| "loss": 3.7098, | |
| "step": 893 | |
| }, | |
| { | |
| "epoch": 0.6652868537940978, | |
| "grad_norm": 0.9550170674026933, | |
| "learning_rate": 7.874015748031496e-06, | |
| "loss": 3.676, | |
| "step": 894 | |
| }, | |
| { | |
| "epoch": 0.6660310225343596, | |
| "grad_norm": 0.907017227330813, | |
| "learning_rate": 7.856517935258093e-06, | |
| "loss": 3.7112, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.6667751912746215, | |
| "grad_norm": 0.9059006548003531, | |
| "learning_rate": 7.83902012248469e-06, | |
| "loss": 3.7136, | |
| "step": 896 | |
| }, | |
| { | |
| "epoch": 0.6675193600148833, | |
| "grad_norm": 0.891496875553174, | |
| "learning_rate": 7.821522309711287e-06, | |
| "loss": 3.6975, | |
| "step": 897 | |
| }, | |
| { | |
| "epoch": 0.6682635287551453, | |
| "grad_norm": 0.8903390647210839, | |
| "learning_rate": 7.804024496937883e-06, | |
| "loss": 3.7009, | |
| "step": 898 | |
| }, | |
| { | |
| "epoch": 0.6690076974954071, | |
| "grad_norm": 0.9049275522095755, | |
| "learning_rate": 7.78652668416448e-06, | |
| "loss": 3.6741, | |
| "step": 899 | |
| }, | |
| { | |
| "epoch": 0.669751866235669, | |
| "grad_norm": 0.9441811214215331, | |
| "learning_rate": 7.769028871391077e-06, | |
| "loss": 3.6844, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6704960349759308, | |
| "grad_norm": 0.899935784371998, | |
| "learning_rate": 7.751531058617673e-06, | |
| "loss": 3.7074, | |
| "step": 901 | |
| }, | |
| { | |
| "epoch": 0.6712402037161926, | |
| "grad_norm": 0.9143169603326312, | |
| "learning_rate": 7.73403324584427e-06, | |
| "loss": 3.7136, | |
| "step": 902 | |
| }, | |
| { | |
| "epoch": 0.6719843724564545, | |
| "grad_norm": 0.9186773144059467, | |
| "learning_rate": 7.716535433070867e-06, | |
| "loss": 3.707, | |
| "step": 903 | |
| }, | |
| { | |
| "epoch": 0.6727285411967163, | |
| "grad_norm": 0.9385147571391796, | |
| "learning_rate": 7.699037620297464e-06, | |
| "loss": 3.6916, | |
| "step": 904 | |
| }, | |
| { | |
| "epoch": 0.6734727099369782, | |
| "grad_norm": 0.9101099295464483, | |
| "learning_rate": 7.68153980752406e-06, | |
| "loss": 3.7005, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.6742168786772401, | |
| "grad_norm": 0.915677107552123, | |
| "learning_rate": 7.664041994750657e-06, | |
| "loss": 3.7404, | |
| "step": 906 | |
| }, | |
| { | |
| "epoch": 0.674961047417502, | |
| "grad_norm": 0.9257373655675415, | |
| "learning_rate": 7.646544181977254e-06, | |
| "loss": 3.6853, | |
| "step": 907 | |
| }, | |
| { | |
| "epoch": 0.6757052161577638, | |
| "grad_norm": 0.9456973687427489, | |
| "learning_rate": 7.62904636920385e-06, | |
| "loss": 3.7113, | |
| "step": 908 | |
| }, | |
| { | |
| "epoch": 0.6764493848980256, | |
| "grad_norm": 0.8949964264556378, | |
| "learning_rate": 7.611548556430447e-06, | |
| "loss": 3.7091, | |
| "step": 909 | |
| }, | |
| { | |
| "epoch": 0.6771935536382875, | |
| "grad_norm": 0.8798317535615784, | |
| "learning_rate": 7.594050743657043e-06, | |
| "loss": 3.686, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6779377223785493, | |
| "grad_norm": 0.9523746644830995, | |
| "learning_rate": 7.576552930883641e-06, | |
| "loss": 3.7516, | |
| "step": 911 | |
| }, | |
| { | |
| "epoch": 0.6786818911188112, | |
| "grad_norm": 0.9088881541071724, | |
| "learning_rate": 7.5590551181102365e-06, | |
| "loss": 3.7162, | |
| "step": 912 | |
| }, | |
| { | |
| "epoch": 0.679426059859073, | |
| "grad_norm": 0.9379732142803633, | |
| "learning_rate": 7.541557305336834e-06, | |
| "loss": 3.7004, | |
| "step": 913 | |
| }, | |
| { | |
| "epoch": 0.6801702285993348, | |
| "grad_norm": 0.9030256486289328, | |
| "learning_rate": 7.52405949256343e-06, | |
| "loss": 3.6592, | |
| "step": 914 | |
| }, | |
| { | |
| "epoch": 0.6809143973395968, | |
| "grad_norm": 0.933171258469292, | |
| "learning_rate": 7.506561679790027e-06, | |
| "loss": 3.6869, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.6816585660798586, | |
| "grad_norm": 0.9515748533878967, | |
| "learning_rate": 7.489063867016624e-06, | |
| "loss": 3.73, | |
| "step": 916 | |
| }, | |
| { | |
| "epoch": 0.6824027348201205, | |
| "grad_norm": 0.9356530505446903, | |
| "learning_rate": 7.47156605424322e-06, | |
| "loss": 3.6721, | |
| "step": 917 | |
| }, | |
| { | |
| "epoch": 0.6831469035603823, | |
| "grad_norm": 0.8996310937565953, | |
| "learning_rate": 7.454068241469818e-06, | |
| "loss": 3.6905, | |
| "step": 918 | |
| }, | |
| { | |
| "epoch": 0.6838910723006442, | |
| "grad_norm": 0.9109809213427803, | |
| "learning_rate": 7.4365704286964135e-06, | |
| "loss": 3.6829, | |
| "step": 919 | |
| }, | |
| { | |
| "epoch": 0.684635241040906, | |
| "grad_norm": 0.9347124390112883, | |
| "learning_rate": 7.419072615923009e-06, | |
| "loss": 3.6971, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6853794097811678, | |
| "grad_norm": 0.9310710630552124, | |
| "learning_rate": 7.401574803149607e-06, | |
| "loss": 3.702, | |
| "step": 921 | |
| }, | |
| { | |
| "epoch": 0.6861235785214297, | |
| "grad_norm": 0.9690120005909113, | |
| "learning_rate": 7.384076990376204e-06, | |
| "loss": 3.7324, | |
| "step": 922 | |
| }, | |
| { | |
| "epoch": 0.6868677472616916, | |
| "grad_norm": 0.8933746600696765, | |
| "learning_rate": 7.3665791776028e-06, | |
| "loss": 3.6894, | |
| "step": 923 | |
| }, | |
| { | |
| "epoch": 0.6876119160019535, | |
| "grad_norm": 0.9156007687546954, | |
| "learning_rate": 7.349081364829397e-06, | |
| "loss": 3.7008, | |
| "step": 924 | |
| }, | |
| { | |
| "epoch": 0.6883560847422153, | |
| "grad_norm": 0.9522812859805392, | |
| "learning_rate": 7.331583552055994e-06, | |
| "loss": 3.7124, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.6891002534824772, | |
| "grad_norm": 0.9388225777708856, | |
| "learning_rate": 7.3140857392825906e-06, | |
| "loss": 3.7462, | |
| "step": 926 | |
| }, | |
| { | |
| "epoch": 0.689844422222739, | |
| "grad_norm": 0.9084041843316689, | |
| "learning_rate": 7.2965879265091864e-06, | |
| "loss": 3.6756, | |
| "step": 927 | |
| }, | |
| { | |
| "epoch": 0.6905885909630008, | |
| "grad_norm": 0.9142464748793597, | |
| "learning_rate": 7.279090113735784e-06, | |
| "loss": 3.7057, | |
| "step": 928 | |
| }, | |
| { | |
| "epoch": 0.6913327597032627, | |
| "grad_norm": 0.9273341163316667, | |
| "learning_rate": 7.26159230096238e-06, | |
| "loss": 3.6668, | |
| "step": 929 | |
| }, | |
| { | |
| "epoch": 0.6920769284435245, | |
| "grad_norm": 0.9348820453475418, | |
| "learning_rate": 7.2440944881889774e-06, | |
| "loss": 3.6669, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6928210971837864, | |
| "grad_norm": 0.9249343796621757, | |
| "learning_rate": 7.226596675415573e-06, | |
| "loss": 3.6949, | |
| "step": 931 | |
| }, | |
| { | |
| "epoch": 0.6935652659240483, | |
| "grad_norm": 0.9048603309022382, | |
| "learning_rate": 7.20909886264217e-06, | |
| "loss": 3.6574, | |
| "step": 932 | |
| }, | |
| { | |
| "epoch": 0.6943094346643102, | |
| "grad_norm": 0.9902909028207817, | |
| "learning_rate": 7.191601049868768e-06, | |
| "loss": 3.7315, | |
| "step": 933 | |
| }, | |
| { | |
| "epoch": 0.695053603404572, | |
| "grad_norm": 0.9206494108905346, | |
| "learning_rate": 7.1741032370953635e-06, | |
| "loss": 3.6958, | |
| "step": 934 | |
| }, | |
| { | |
| "epoch": 0.6957977721448338, | |
| "grad_norm": 0.9312790984769943, | |
| "learning_rate": 7.156605424321961e-06, | |
| "loss": 3.7223, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.6965419408850957, | |
| "grad_norm": 0.947129257830227, | |
| "learning_rate": 7.139107611548557e-06, | |
| "loss": 3.6977, | |
| "step": 936 | |
| }, | |
| { | |
| "epoch": 0.6972861096253575, | |
| "grad_norm": 0.9790523982982996, | |
| "learning_rate": 7.1216097987751545e-06, | |
| "loss": 3.7113, | |
| "step": 937 | |
| }, | |
| { | |
| "epoch": 0.6980302783656194, | |
| "grad_norm": 0.9332728301033003, | |
| "learning_rate": 7.10411198600175e-06, | |
| "loss": 3.6987, | |
| "step": 938 | |
| }, | |
| { | |
| "epoch": 0.6987744471058812, | |
| "grad_norm": 0.9290097129098147, | |
| "learning_rate": 7.086614173228347e-06, | |
| "loss": 3.7326, | |
| "step": 939 | |
| }, | |
| { | |
| "epoch": 0.6995186158461432, | |
| "grad_norm": 1.0037322369989579, | |
| "learning_rate": 7.069116360454944e-06, | |
| "loss": 3.6755, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.700262784586405, | |
| "grad_norm": 0.9668736552737901, | |
| "learning_rate": 7.0516185476815405e-06, | |
| "loss": 3.7092, | |
| "step": 941 | |
| }, | |
| { | |
| "epoch": 0.7010069533266668, | |
| "grad_norm": 0.9236919932990034, | |
| "learning_rate": 7.034120734908137e-06, | |
| "loss": 3.6962, | |
| "step": 942 | |
| }, | |
| { | |
| "epoch": 0.7017511220669287, | |
| "grad_norm": 0.9239328887863173, | |
| "learning_rate": 7.016622922134734e-06, | |
| "loss": 3.6502, | |
| "step": 943 | |
| }, | |
| { | |
| "epoch": 0.7024952908071905, | |
| "grad_norm": 0.9225042932088281, | |
| "learning_rate": 6.99912510936133e-06, | |
| "loss": 3.6686, | |
| "step": 944 | |
| }, | |
| { | |
| "epoch": 0.7032394595474524, | |
| "grad_norm": 0.9742239858590739, | |
| "learning_rate": 6.981627296587927e-06, | |
| "loss": 3.6951, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.7039836282877142, | |
| "grad_norm": 0.9696542602632232, | |
| "learning_rate": 6.964129483814523e-06, | |
| "loss": 3.694, | |
| "step": 946 | |
| }, | |
| { | |
| "epoch": 0.704727797027976, | |
| "grad_norm": 0.931590229084351, | |
| "learning_rate": 6.946631671041121e-06, | |
| "loss": 3.6642, | |
| "step": 947 | |
| }, | |
| { | |
| "epoch": 0.7054719657682379, | |
| "grad_norm": 0.9206777307342459, | |
| "learning_rate": 6.929133858267717e-06, | |
| "loss": 3.6985, | |
| "step": 948 | |
| }, | |
| { | |
| "epoch": 0.7062161345084998, | |
| "grad_norm": 0.9324283795605197, | |
| "learning_rate": 6.911636045494314e-06, | |
| "loss": 3.691, | |
| "step": 949 | |
| }, | |
| { | |
| "epoch": 0.7069603032487617, | |
| "grad_norm": 0.8914879803104745, | |
| "learning_rate": 6.89413823272091e-06, | |
| "loss": 3.6942, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.7077044719890235, | |
| "grad_norm": 0.9261189633433973, | |
| "learning_rate": 6.876640419947507e-06, | |
| "loss": 3.7017, | |
| "step": 951 | |
| }, | |
| { | |
| "epoch": 0.7084486407292854, | |
| "grad_norm": 0.9513328457699453, | |
| "learning_rate": 6.859142607174104e-06, | |
| "loss": 3.7089, | |
| "step": 952 | |
| }, | |
| { | |
| "epoch": 0.7091928094695472, | |
| "grad_norm": 0.8778981087236783, | |
| "learning_rate": 6.8416447944007e-06, | |
| "loss": 3.6786, | |
| "step": 953 | |
| }, | |
| { | |
| "epoch": 0.709936978209809, | |
| "grad_norm": 0.9502403934807211, | |
| "learning_rate": 6.824146981627298e-06, | |
| "loss": 3.7271, | |
| "step": 954 | |
| }, | |
| { | |
| "epoch": 0.7106811469500709, | |
| "grad_norm": 0.8958984727986383, | |
| "learning_rate": 6.806649168853894e-06, | |
| "loss": 3.7078, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.7114253156903327, | |
| "grad_norm": 0.8760937718163647, | |
| "learning_rate": 6.78915135608049e-06, | |
| "loss": 3.6442, | |
| "step": 956 | |
| }, | |
| { | |
| "epoch": 0.7121694844305947, | |
| "grad_norm": 0.8745863809731302, | |
| "learning_rate": 6.771653543307087e-06, | |
| "loss": 3.6824, | |
| "step": 957 | |
| }, | |
| { | |
| "epoch": 0.7129136531708565, | |
| "grad_norm": 0.8890110093995057, | |
| "learning_rate": 6.754155730533684e-06, | |
| "loss": 3.7098, | |
| "step": 958 | |
| }, | |
| { | |
| "epoch": 0.7136578219111184, | |
| "grad_norm": 0.9130913853527122, | |
| "learning_rate": 6.736657917760281e-06, | |
| "loss": 3.7309, | |
| "step": 959 | |
| }, | |
| { | |
| "epoch": 0.7144019906513802, | |
| "grad_norm": 0.9195646545602905, | |
| "learning_rate": 6.719160104986877e-06, | |
| "loss": 3.6701, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.7151461593916421, | |
| "grad_norm": 0.8998344591344706, | |
| "learning_rate": 6.701662292213474e-06, | |
| "loss": 3.6874, | |
| "step": 961 | |
| }, | |
| { | |
| "epoch": 0.7158903281319039, | |
| "grad_norm": 0.92308329636997, | |
| "learning_rate": 6.684164479440071e-06, | |
| "loss": 3.7041, | |
| "step": 962 | |
| }, | |
| { | |
| "epoch": 0.7166344968721657, | |
| "grad_norm": 0.8901012530182424, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 3.694, | |
| "step": 963 | |
| }, | |
| { | |
| "epoch": 0.7173786656124276, | |
| "grad_norm": 0.9254855362526868, | |
| "learning_rate": 6.649168853893264e-06, | |
| "loss": 3.6852, | |
| "step": 964 | |
| }, | |
| { | |
| "epoch": 0.7181228343526894, | |
| "grad_norm": 0.9817845423622017, | |
| "learning_rate": 6.63167104111986e-06, | |
| "loss": 3.7007, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.7188670030929514, | |
| "grad_norm": 0.9032555737995551, | |
| "learning_rate": 6.614173228346458e-06, | |
| "loss": 3.6791, | |
| "step": 966 | |
| }, | |
| { | |
| "epoch": 0.7196111718332132, | |
| "grad_norm": 0.9342993376183282, | |
| "learning_rate": 6.5966754155730535e-06, | |
| "loss": 3.7225, | |
| "step": 967 | |
| }, | |
| { | |
| "epoch": 0.7203553405734751, | |
| "grad_norm": 0.9237164505173904, | |
| "learning_rate": 6.57917760279965e-06, | |
| "loss": 3.6896, | |
| "step": 968 | |
| }, | |
| { | |
| "epoch": 0.7210995093137369, | |
| "grad_norm": 0.9267459790055574, | |
| "learning_rate": 6.561679790026248e-06, | |
| "loss": 3.698, | |
| "step": 969 | |
| }, | |
| { | |
| "epoch": 0.7218436780539987, | |
| "grad_norm": 0.9277212031392768, | |
| "learning_rate": 6.544181977252844e-06, | |
| "loss": 3.6635, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.7225878467942606, | |
| "grad_norm": 0.9312505458065482, | |
| "learning_rate": 6.526684164479441e-06, | |
| "loss": 3.685, | |
| "step": 971 | |
| }, | |
| { | |
| "epoch": 0.7233320155345224, | |
| "grad_norm": 0.9137614553076302, | |
| "learning_rate": 6.509186351706037e-06, | |
| "loss": 3.729, | |
| "step": 972 | |
| }, | |
| { | |
| "epoch": 0.7240761842747843, | |
| "grad_norm": 0.9023050113086767, | |
| "learning_rate": 6.491688538932633e-06, | |
| "loss": 3.7043, | |
| "step": 973 | |
| }, | |
| { | |
| "epoch": 0.7248203530150462, | |
| "grad_norm": 0.9208072664394867, | |
| "learning_rate": 6.4741907261592306e-06, | |
| "loss": 3.6889, | |
| "step": 974 | |
| }, | |
| { | |
| "epoch": 0.7255645217553081, | |
| "grad_norm": 0.9579663934697336, | |
| "learning_rate": 6.456692913385827e-06, | |
| "loss": 3.7355, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.7263086904955699, | |
| "grad_norm": 0.946575514644256, | |
| "learning_rate": 6.439195100612424e-06, | |
| "loss": 3.7199, | |
| "step": 976 | |
| }, | |
| { | |
| "epoch": 0.7270528592358317, | |
| "grad_norm": 0.922807187915385, | |
| "learning_rate": 6.421697287839021e-06, | |
| "loss": 3.7097, | |
| "step": 977 | |
| }, | |
| { | |
| "epoch": 0.7277970279760936, | |
| "grad_norm": 0.8933258157928686, | |
| "learning_rate": 6.4041994750656174e-06, | |
| "loss": 3.6932, | |
| "step": 978 | |
| }, | |
| { | |
| "epoch": 0.7285411967163554, | |
| "grad_norm": 0.8940497366286886, | |
| "learning_rate": 6.386701662292214e-06, | |
| "loss": 3.7081, | |
| "step": 979 | |
| }, | |
| { | |
| "epoch": 0.7292853654566173, | |
| "grad_norm": 0.8998138911593359, | |
| "learning_rate": 6.36920384951881e-06, | |
| "loss": 3.6968, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.7300295341968791, | |
| "grad_norm": 0.9446577990413411, | |
| "learning_rate": 6.351706036745408e-06, | |
| "loss": 3.7015, | |
| "step": 981 | |
| }, | |
| { | |
| "epoch": 0.730773702937141, | |
| "grad_norm": 0.897362083027042, | |
| "learning_rate": 6.3342082239720035e-06, | |
| "loss": 3.6764, | |
| "step": 982 | |
| }, | |
| { | |
| "epoch": 0.7315178716774029, | |
| "grad_norm": 0.9278366576100482, | |
| "learning_rate": 6.316710411198601e-06, | |
| "loss": 3.7092, | |
| "step": 983 | |
| }, | |
| { | |
| "epoch": 0.7322620404176647, | |
| "grad_norm": 0.8852600153217219, | |
| "learning_rate": 6.299212598425197e-06, | |
| "loss": 3.7286, | |
| "step": 984 | |
| }, | |
| { | |
| "epoch": 0.7330062091579266, | |
| "grad_norm": 0.9191168850628386, | |
| "learning_rate": 6.281714785651794e-06, | |
| "loss": 3.69, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.7337503778981884, | |
| "grad_norm": 0.9240928803964832, | |
| "learning_rate": 6.264216972878391e-06, | |
| "loss": 3.7406, | |
| "step": 986 | |
| }, | |
| { | |
| "epoch": 0.7344945466384503, | |
| "grad_norm": 0.8974105083011322, | |
| "learning_rate": 6.246719160104987e-06, | |
| "loss": 3.6627, | |
| "step": 987 | |
| }, | |
| { | |
| "epoch": 0.7352387153787121, | |
| "grad_norm": 0.984862977557596, | |
| "learning_rate": 6.229221347331585e-06, | |
| "loss": 3.7054, | |
| "step": 988 | |
| }, | |
| { | |
| "epoch": 0.735982884118974, | |
| "grad_norm": 0.9394903372756057, | |
| "learning_rate": 6.2117235345581805e-06, | |
| "loss": 3.731, | |
| "step": 989 | |
| }, | |
| { | |
| "epoch": 0.7367270528592358, | |
| "grad_norm": 0.8879424452882946, | |
| "learning_rate": 6.194225721784778e-06, | |
| "loss": 3.692, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.7374712215994977, | |
| "grad_norm": 0.9145608055530033, | |
| "learning_rate": 6.176727909011374e-06, | |
| "loss": 3.7094, | |
| "step": 991 | |
| }, | |
| { | |
| "epoch": 0.7382153903397596, | |
| "grad_norm": 0.9334398120700705, | |
| "learning_rate": 6.159230096237971e-06, | |
| "loss": 3.6564, | |
| "step": 992 | |
| }, | |
| { | |
| "epoch": 0.7389595590800214, | |
| "grad_norm": 0.9259744742614276, | |
| "learning_rate": 6.141732283464567e-06, | |
| "loss": 3.7358, | |
| "step": 993 | |
| }, | |
| { | |
| "epoch": 0.7397037278202833, | |
| "grad_norm": 0.9312689880471816, | |
| "learning_rate": 6.124234470691164e-06, | |
| "loss": 3.7082, | |
| "step": 994 | |
| }, | |
| { | |
| "epoch": 0.7404478965605451, | |
| "grad_norm": 0.9212027328507567, | |
| "learning_rate": 6.106736657917761e-06, | |
| "loss": 3.6862, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.741192065300807, | |
| "grad_norm": 0.9425227208252673, | |
| "learning_rate": 6.0892388451443576e-06, | |
| "loss": 3.6924, | |
| "step": 996 | |
| }, | |
| { | |
| "epoch": 0.7419362340410688, | |
| "grad_norm": 0.9429946142368746, | |
| "learning_rate": 6.0717410323709534e-06, | |
| "loss": 3.7221, | |
| "step": 997 | |
| }, | |
| { | |
| "epoch": 0.7426804027813306, | |
| "grad_norm": 0.9071005094871947, | |
| "learning_rate": 6.054243219597551e-06, | |
| "loss": 3.6626, | |
| "step": 998 | |
| }, | |
| { | |
| "epoch": 0.7434245715215925, | |
| "grad_norm": 0.8883401471559458, | |
| "learning_rate": 6.036745406824147e-06, | |
| "loss": 3.7226, | |
| "step": 999 | |
| }, | |
| { | |
| "epoch": 0.7441687402618544, | |
| "grad_norm": 0.8929512251063004, | |
| "learning_rate": 6.0192475940507444e-06, | |
| "loss": 3.6982, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.7449129090021163, | |
| "grad_norm": 0.9235155678042534, | |
| "learning_rate": 6.00174978127734e-06, | |
| "loss": 3.6679, | |
| "step": 1001 | |
| }, | |
| { | |
| "epoch": 0.7456570777423781, | |
| "grad_norm": 0.9306885497258314, | |
| "learning_rate": 5.984251968503938e-06, | |
| "loss": 3.7339, | |
| "step": 1002 | |
| }, | |
| { | |
| "epoch": 0.74640124648264, | |
| "grad_norm": 0.9146060890779841, | |
| "learning_rate": 5.966754155730535e-06, | |
| "loss": 3.6822, | |
| "step": 1003 | |
| }, | |
| { | |
| "epoch": 0.7471454152229018, | |
| "grad_norm": 0.8804874279225268, | |
| "learning_rate": 5.9492563429571305e-06, | |
| "loss": 3.704, | |
| "step": 1004 | |
| }, | |
| { | |
| "epoch": 0.7478895839631636, | |
| "grad_norm": 0.924663247916108, | |
| "learning_rate": 5.931758530183728e-06, | |
| "loss": 3.6909, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 0.7486337527034255, | |
| "grad_norm": 0.8873740912944477, | |
| "learning_rate": 5.914260717410324e-06, | |
| "loss": 3.6715, | |
| "step": 1006 | |
| }, | |
| { | |
| "epoch": 0.7493779214436873, | |
| "grad_norm": 0.9137356367290378, | |
| "learning_rate": 5.8967629046369215e-06, | |
| "loss": 3.7109, | |
| "step": 1007 | |
| }, | |
| { | |
| "epoch": 0.7501220901839493, | |
| "grad_norm": 0.880682610186265, | |
| "learning_rate": 5.879265091863517e-06, | |
| "loss": 3.6754, | |
| "step": 1008 | |
| }, | |
| { | |
| "epoch": 0.7508662589242111, | |
| "grad_norm": 0.9108364302621635, | |
| "learning_rate": 5.861767279090114e-06, | |
| "loss": 3.7197, | |
| "step": 1009 | |
| }, | |
| { | |
| "epoch": 0.751610427664473, | |
| "grad_norm": 0.9466000490470623, | |
| "learning_rate": 5.844269466316711e-06, | |
| "loss": 3.7305, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.7523545964047348, | |
| "grad_norm": 0.9013882969881057, | |
| "learning_rate": 5.8267716535433075e-06, | |
| "loss": 3.6922, | |
| "step": 1011 | |
| }, | |
| { | |
| "epoch": 0.7530987651449966, | |
| "grad_norm": 0.9387709098201181, | |
| "learning_rate": 5.809273840769904e-06, | |
| "loss": 3.6984, | |
| "step": 1012 | |
| }, | |
| { | |
| "epoch": 0.7538429338852585, | |
| "grad_norm": 0.9286264409917633, | |
| "learning_rate": 5.791776027996501e-06, | |
| "loss": 3.7386, | |
| "step": 1013 | |
| }, | |
| { | |
| "epoch": 0.7545871026255203, | |
| "grad_norm": 0.9128026273551847, | |
| "learning_rate": 5.774278215223098e-06, | |
| "loss": 3.6904, | |
| "step": 1014 | |
| }, | |
| { | |
| "epoch": 0.7553312713657822, | |
| "grad_norm": 0.917933760745015, | |
| "learning_rate": 5.756780402449694e-06, | |
| "loss": 3.7108, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 0.756075440106044, | |
| "grad_norm": 0.917467836487481, | |
| "learning_rate": 5.73928258967629e-06, | |
| "loss": 3.6532, | |
| "step": 1016 | |
| }, | |
| { | |
| "epoch": 0.756819608846306, | |
| "grad_norm": 0.919116253792959, | |
| "learning_rate": 5.721784776902888e-06, | |
| "loss": 3.651, | |
| "step": 1017 | |
| }, | |
| { | |
| "epoch": 0.7575637775865678, | |
| "grad_norm": 0.9897461404098725, | |
| "learning_rate": 5.704286964129484e-06, | |
| "loss": 3.6685, | |
| "step": 1018 | |
| }, | |
| { | |
| "epoch": 0.7583079463268296, | |
| "grad_norm": 0.9274466887949088, | |
| "learning_rate": 5.686789151356081e-06, | |
| "loss": 3.6916, | |
| "step": 1019 | |
| }, | |
| { | |
| "epoch": 0.7590521150670915, | |
| "grad_norm": 0.8965748883439177, | |
| "learning_rate": 5.669291338582677e-06, | |
| "loss": 3.7003, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.7597962838073533, | |
| "grad_norm": 0.8933349755303948, | |
| "learning_rate": 5.651793525809274e-06, | |
| "loss": 3.6736, | |
| "step": 1021 | |
| }, | |
| { | |
| "epoch": 0.7605404525476152, | |
| "grad_norm": 0.9493321823158284, | |
| "learning_rate": 5.6342957130358714e-06, | |
| "loss": 3.7198, | |
| "step": 1022 | |
| }, | |
| { | |
| "epoch": 0.761284621287877, | |
| "grad_norm": 0.890365667689091, | |
| "learning_rate": 5.616797900262467e-06, | |
| "loss": 3.6732, | |
| "step": 1023 | |
| }, | |
| { | |
| "epoch": 0.7620287900281388, | |
| "grad_norm": 0.8995326940203813, | |
| "learning_rate": 5.599300087489065e-06, | |
| "loss": 3.6485, | |
| "step": 1024 | |
| }, | |
| { | |
| "epoch": 0.7627729587684008, | |
| "grad_norm": 0.8977690503463777, | |
| "learning_rate": 5.581802274715661e-06, | |
| "loss": 3.7202, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.7635171275086626, | |
| "grad_norm": 0.9017890661463822, | |
| "learning_rate": 5.564304461942258e-06, | |
| "loss": 3.663, | |
| "step": 1026 | |
| }, | |
| { | |
| "epoch": 0.7642612962489245, | |
| "grad_norm": 0.9019454066551111, | |
| "learning_rate": 5.546806649168854e-06, | |
| "loss": 3.6553, | |
| "step": 1027 | |
| }, | |
| { | |
| "epoch": 0.7650054649891863, | |
| "grad_norm": 0.9155044104911286, | |
| "learning_rate": 5.529308836395451e-06, | |
| "loss": 3.667, | |
| "step": 1028 | |
| }, | |
| { | |
| "epoch": 0.7657496337294482, | |
| "grad_norm": 0.9504085259875951, | |
| "learning_rate": 5.511811023622048e-06, | |
| "loss": 3.6719, | |
| "step": 1029 | |
| }, | |
| { | |
| "epoch": 0.76649380246971, | |
| "grad_norm": 0.8968659590573693, | |
| "learning_rate": 5.494313210848644e-06, | |
| "loss": 3.6697, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7672379712099718, | |
| "grad_norm": 0.9372866523771236, | |
| "learning_rate": 5.476815398075241e-06, | |
| "loss": 3.6944, | |
| "step": 1031 | |
| }, | |
| { | |
| "epoch": 0.7679821399502337, | |
| "grad_norm": 0.9107862521671581, | |
| "learning_rate": 5.459317585301838e-06, | |
| "loss": 3.727, | |
| "step": 1032 | |
| }, | |
| { | |
| "epoch": 0.7687263086904955, | |
| "grad_norm": 0.8965638909243643, | |
| "learning_rate": 5.441819772528434e-06, | |
| "loss": 3.7018, | |
| "step": 1033 | |
| }, | |
| { | |
| "epoch": 0.7694704774307575, | |
| "grad_norm": 0.9181574174648887, | |
| "learning_rate": 5.424321959755031e-06, | |
| "loss": 3.6907, | |
| "step": 1034 | |
| }, | |
| { | |
| "epoch": 0.7702146461710193, | |
| "grad_norm": 0.9023926360467333, | |
| "learning_rate": 5.406824146981627e-06, | |
| "loss": 3.6594, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 0.7709588149112812, | |
| "grad_norm": 0.9416711274277582, | |
| "learning_rate": 5.389326334208225e-06, | |
| "loss": 3.6528, | |
| "step": 1036 | |
| }, | |
| { | |
| "epoch": 0.771702983651543, | |
| "grad_norm": 0.8843842653895319, | |
| "learning_rate": 5.3718285214348205e-06, | |
| "loss": 3.6947, | |
| "step": 1037 | |
| }, | |
| { | |
| "epoch": 0.7724471523918048, | |
| "grad_norm": 0.9306105548250855, | |
| "learning_rate": 5.354330708661418e-06, | |
| "loss": 3.667, | |
| "step": 1038 | |
| }, | |
| { | |
| "epoch": 0.7731913211320667, | |
| "grad_norm": 0.9018062320271113, | |
| "learning_rate": 5.336832895888015e-06, | |
| "loss": 3.6073, | |
| "step": 1039 | |
| }, | |
| { | |
| "epoch": 0.7739354898723285, | |
| "grad_norm": 0.9126786271859934, | |
| "learning_rate": 5.319335083114611e-06, | |
| "loss": 3.6598, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7746796586125904, | |
| "grad_norm": 0.939112081199725, | |
| "learning_rate": 5.301837270341208e-06, | |
| "loss": 3.6804, | |
| "step": 1041 | |
| }, | |
| { | |
| "epoch": 0.7754238273528523, | |
| "grad_norm": 0.9128402254201701, | |
| "learning_rate": 5.284339457567804e-06, | |
| "loss": 3.6844, | |
| "step": 1042 | |
| }, | |
| { | |
| "epoch": 0.7761679960931142, | |
| "grad_norm": 0.8895020117926487, | |
| "learning_rate": 5.266841644794402e-06, | |
| "loss": 3.698, | |
| "step": 1043 | |
| }, | |
| { | |
| "epoch": 0.776912164833376, | |
| "grad_norm": 0.9097903364003814, | |
| "learning_rate": 5.2493438320209976e-06, | |
| "loss": 3.7027, | |
| "step": 1044 | |
| }, | |
| { | |
| "epoch": 0.7776563335736378, | |
| "grad_norm": 0.9158809770741764, | |
| "learning_rate": 5.231846019247594e-06, | |
| "loss": 3.6713, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 0.7784005023138997, | |
| "grad_norm": 0.8919700795037542, | |
| "learning_rate": 5.214348206474191e-06, | |
| "loss": 3.7119, | |
| "step": 1046 | |
| }, | |
| { | |
| "epoch": 0.7791446710541615, | |
| "grad_norm": 0.8878872383166743, | |
| "learning_rate": 5.196850393700788e-06, | |
| "loss": 3.6804, | |
| "step": 1047 | |
| }, | |
| { | |
| "epoch": 0.7798888397944234, | |
| "grad_norm": 0.9380292613080534, | |
| "learning_rate": 5.1793525809273845e-06, | |
| "loss": 3.6807, | |
| "step": 1048 | |
| }, | |
| { | |
| "epoch": 0.7806330085346852, | |
| "grad_norm": 0.8801612394176171, | |
| "learning_rate": 5.161854768153981e-06, | |
| "loss": 3.7002, | |
| "step": 1049 | |
| }, | |
| { | |
| "epoch": 0.781377177274947, | |
| "grad_norm": 0.8847150763430098, | |
| "learning_rate": 5.144356955380579e-06, | |
| "loss": 3.6961, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.782121346015209, | |
| "grad_norm": 0.9131943427615435, | |
| "learning_rate": 5.126859142607175e-06, | |
| "loss": 3.6935, | |
| "step": 1051 | |
| }, | |
| { | |
| "epoch": 0.7828655147554708, | |
| "grad_norm": 0.867491636710469, | |
| "learning_rate": 5.1093613298337705e-06, | |
| "loss": 3.6458, | |
| "step": 1052 | |
| }, | |
| { | |
| "epoch": 0.7836096834957327, | |
| "grad_norm": 0.9395079887530251, | |
| "learning_rate": 5.091863517060368e-06, | |
| "loss": 3.7019, | |
| "step": 1053 | |
| }, | |
| { | |
| "epoch": 0.7843538522359945, | |
| "grad_norm": 0.9309171711730478, | |
| "learning_rate": 5.074365704286964e-06, | |
| "loss": 3.6783, | |
| "step": 1054 | |
| }, | |
| { | |
| "epoch": 0.7850980209762564, | |
| "grad_norm": 0.925187423081186, | |
| "learning_rate": 5.0568678915135615e-06, | |
| "loss": 3.6905, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 0.7858421897165182, | |
| "grad_norm": 0.8919592728360723, | |
| "learning_rate": 5.039370078740158e-06, | |
| "loss": 3.7043, | |
| "step": 1056 | |
| }, | |
| { | |
| "epoch": 0.78658635845678, | |
| "grad_norm": 0.8897203379806394, | |
| "learning_rate": 5.021872265966754e-06, | |
| "loss": 3.6978, | |
| "step": 1057 | |
| }, | |
| { | |
| "epoch": 0.7873305271970419, | |
| "grad_norm": 0.9028572184397602, | |
| "learning_rate": 5.004374453193352e-06, | |
| "loss": 3.6692, | |
| "step": 1058 | |
| }, | |
| { | |
| "epoch": 0.7880746959373038, | |
| "grad_norm": 0.8998993222277701, | |
| "learning_rate": 4.986876640419948e-06, | |
| "loss": 3.6482, | |
| "step": 1059 | |
| }, | |
| { | |
| "epoch": 0.7888188646775657, | |
| "grad_norm": 0.9370569558401803, | |
| "learning_rate": 4.969378827646544e-06, | |
| "loss": 3.7304, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7895630334178275, | |
| "grad_norm": 0.917734983018238, | |
| "learning_rate": 4.951881014873141e-06, | |
| "loss": 3.7117, | |
| "step": 1061 | |
| }, | |
| { | |
| "epoch": 0.7903072021580894, | |
| "grad_norm": 0.9101069872503845, | |
| "learning_rate": 4.934383202099738e-06, | |
| "loss": 3.6886, | |
| "step": 1062 | |
| }, | |
| { | |
| "epoch": 0.7910513708983512, | |
| "grad_norm": 0.8889106532199818, | |
| "learning_rate": 4.916885389326334e-06, | |
| "loss": 3.6924, | |
| "step": 1063 | |
| }, | |
| { | |
| "epoch": 0.791795539638613, | |
| "grad_norm": 0.9147783760556952, | |
| "learning_rate": 4.899387576552931e-06, | |
| "loss": 3.6895, | |
| "step": 1064 | |
| }, | |
| { | |
| "epoch": 0.7925397083788749, | |
| "grad_norm": 0.9505841071427902, | |
| "learning_rate": 4.881889763779528e-06, | |
| "loss": 3.6724, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 0.7932838771191367, | |
| "grad_norm": 0.9457483189914331, | |
| "learning_rate": 4.8643919510061246e-06, | |
| "loss": 3.6899, | |
| "step": 1066 | |
| }, | |
| { | |
| "epoch": 0.7940280458593986, | |
| "grad_norm": 0.8922495692856779, | |
| "learning_rate": 4.846894138232721e-06, | |
| "loss": 3.701, | |
| "step": 1067 | |
| }, | |
| { | |
| "epoch": 0.7947722145996605, | |
| "grad_norm": 0.9283422231932675, | |
| "learning_rate": 4.829396325459318e-06, | |
| "loss": 3.6489, | |
| "step": 1068 | |
| }, | |
| { | |
| "epoch": 0.7955163833399224, | |
| "grad_norm": 0.9307883639699536, | |
| "learning_rate": 4.811898512685915e-06, | |
| "loss": 3.6646, | |
| "step": 1069 | |
| }, | |
| { | |
| "epoch": 0.7962605520801842, | |
| "grad_norm": 0.9070940509958437, | |
| "learning_rate": 4.7944006999125114e-06, | |
| "loss": 3.6782, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.797004720820446, | |
| "grad_norm": 0.9030930739886565, | |
| "learning_rate": 4.776902887139108e-06, | |
| "loss": 3.7002, | |
| "step": 1071 | |
| }, | |
| { | |
| "epoch": 0.7977488895607079, | |
| "grad_norm": 0.9145752896525304, | |
| "learning_rate": 4.759405074365704e-06, | |
| "loss": 3.7049, | |
| "step": 1072 | |
| }, | |
| { | |
| "epoch": 0.7984930583009697, | |
| "grad_norm": 0.9087632481896102, | |
| "learning_rate": 4.741907261592302e-06, | |
| "loss": 3.722, | |
| "step": 1073 | |
| }, | |
| { | |
| "epoch": 0.7992372270412316, | |
| "grad_norm": 0.8780885451319033, | |
| "learning_rate": 4.724409448818898e-06, | |
| "loss": 3.6655, | |
| "step": 1074 | |
| }, | |
| { | |
| "epoch": 0.7999813957814934, | |
| "grad_norm": 0.8923735660074177, | |
| "learning_rate": 4.706911636045495e-06, | |
| "loss": 3.7023, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.8007255645217554, | |
| "grad_norm": 0.9090759131193389, | |
| "learning_rate": 4.689413823272092e-06, | |
| "loss": 3.6959, | |
| "step": 1076 | |
| }, | |
| { | |
| "epoch": 0.8014697332620172, | |
| "grad_norm": 0.9034296148412665, | |
| "learning_rate": 4.6719160104986885e-06, | |
| "loss": 3.6792, | |
| "step": 1077 | |
| }, | |
| { | |
| "epoch": 0.802213902002279, | |
| "grad_norm": 0.871318232113845, | |
| "learning_rate": 4.654418197725284e-06, | |
| "loss": 3.6766, | |
| "step": 1078 | |
| }, | |
| { | |
| "epoch": 0.8029580707425409, | |
| "grad_norm": 0.9112708256092108, | |
| "learning_rate": 4.636920384951881e-06, | |
| "loss": 3.702, | |
| "step": 1079 | |
| }, | |
| { | |
| "epoch": 0.8037022394828027, | |
| "grad_norm": 0.9188574288444337, | |
| "learning_rate": 4.619422572178478e-06, | |
| "loss": 3.6584, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.8044464082230646, | |
| "grad_norm": 0.9409788387682897, | |
| "learning_rate": 4.6019247594050745e-06, | |
| "loss": 3.6488, | |
| "step": 1081 | |
| }, | |
| { | |
| "epoch": 0.8051905769633264, | |
| "grad_norm": 0.9181643014852878, | |
| "learning_rate": 4.584426946631671e-06, | |
| "loss": 3.6863, | |
| "step": 1082 | |
| }, | |
| { | |
| "epoch": 0.8059347457035883, | |
| "grad_norm": 0.8991387463624163, | |
| "learning_rate": 4.566929133858268e-06, | |
| "loss": 3.6711, | |
| "step": 1083 | |
| }, | |
| { | |
| "epoch": 0.8066789144438501, | |
| "grad_norm": 0.9015869560576475, | |
| "learning_rate": 4.549431321084865e-06, | |
| "loss": 3.6813, | |
| "step": 1084 | |
| }, | |
| { | |
| "epoch": 0.807423083184112, | |
| "grad_norm": 0.9433740215699435, | |
| "learning_rate": 4.531933508311461e-06, | |
| "loss": 3.6112, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 0.8081672519243739, | |
| "grad_norm": 0.9389671096239891, | |
| "learning_rate": 4.514435695538058e-06, | |
| "loss": 3.6631, | |
| "step": 1086 | |
| }, | |
| { | |
| "epoch": 0.8089114206646357, | |
| "grad_norm": 0.910617680415671, | |
| "learning_rate": 4.496937882764655e-06, | |
| "loss": 3.6732, | |
| "step": 1087 | |
| }, | |
| { | |
| "epoch": 0.8096555894048976, | |
| "grad_norm": 0.8985008973190096, | |
| "learning_rate": 4.4794400699912516e-06, | |
| "loss": 3.6557, | |
| "step": 1088 | |
| }, | |
| { | |
| "epoch": 0.8103997581451594, | |
| "grad_norm": 0.9467480415006052, | |
| "learning_rate": 4.461942257217848e-06, | |
| "loss": 3.6638, | |
| "step": 1089 | |
| }, | |
| { | |
| "epoch": 0.8111439268854213, | |
| "grad_norm": 0.9109824042847842, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 3.6992, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.8118880956256831, | |
| "grad_norm": 0.9280648867136734, | |
| "learning_rate": 4.426946631671042e-06, | |
| "loss": 3.6925, | |
| "step": 1091 | |
| }, | |
| { | |
| "epoch": 0.8126322643659449, | |
| "grad_norm": 0.932815002986764, | |
| "learning_rate": 4.4094488188976384e-06, | |
| "loss": 3.6933, | |
| "step": 1092 | |
| }, | |
| { | |
| "epoch": 0.8133764331062069, | |
| "grad_norm": 0.9154156832839634, | |
| "learning_rate": 4.391951006124235e-06, | |
| "loss": 3.6675, | |
| "step": 1093 | |
| }, | |
| { | |
| "epoch": 0.8141206018464687, | |
| "grad_norm": 0.9044240134322926, | |
| "learning_rate": 4.374453193350832e-06, | |
| "loss": 3.6876, | |
| "step": 1094 | |
| }, | |
| { | |
| "epoch": 0.8148647705867306, | |
| "grad_norm": 0.9225410948165761, | |
| "learning_rate": 4.356955380577429e-06, | |
| "loss": 3.6708, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 0.8156089393269924, | |
| "grad_norm": 0.9466280230310901, | |
| "learning_rate": 4.3394575678040245e-06, | |
| "loss": 3.7154, | |
| "step": 1096 | |
| }, | |
| { | |
| "epoch": 0.8163531080672543, | |
| "grad_norm": 0.8946164176040977, | |
| "learning_rate": 4.321959755030621e-06, | |
| "loss": 3.6588, | |
| "step": 1097 | |
| }, | |
| { | |
| "epoch": 0.8170972768075161, | |
| "grad_norm": 0.888683744952802, | |
| "learning_rate": 4.304461942257218e-06, | |
| "loss": 3.716, | |
| "step": 1098 | |
| }, | |
| { | |
| "epoch": 0.8178414455477779, | |
| "grad_norm": 0.9268796422586117, | |
| "learning_rate": 4.286964129483815e-06, | |
| "loss": 3.6771, | |
| "step": 1099 | |
| }, | |
| { | |
| "epoch": 0.8185856142880398, | |
| "grad_norm": 0.9268092709157334, | |
| "learning_rate": 4.269466316710411e-06, | |
| "loss": 3.6846, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.8193297830283016, | |
| "grad_norm": 0.9994861196175193, | |
| "learning_rate": 4.251968503937008e-06, | |
| "loss": 3.6926, | |
| "step": 1101 | |
| }, | |
| { | |
| "epoch": 0.8200739517685636, | |
| "grad_norm": 0.9037611991765405, | |
| "learning_rate": 4.234470691163605e-06, | |
| "loss": 3.6764, | |
| "step": 1102 | |
| }, | |
| { | |
| "epoch": 0.8208181205088254, | |
| "grad_norm": 0.9021477511833556, | |
| "learning_rate": 4.2169728783902015e-06, | |
| "loss": 3.6998, | |
| "step": 1103 | |
| }, | |
| { | |
| "epoch": 0.8215622892490873, | |
| "grad_norm": 0.9202265322455968, | |
| "learning_rate": 4.199475065616798e-06, | |
| "loss": 3.6649, | |
| "step": 1104 | |
| }, | |
| { | |
| "epoch": 0.8223064579893491, | |
| "grad_norm": 0.9143258646734084, | |
| "learning_rate": 4.181977252843395e-06, | |
| "loss": 3.6928, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 0.8230506267296109, | |
| "grad_norm": 0.907440223645327, | |
| "learning_rate": 4.164479440069992e-06, | |
| "loss": 3.6538, | |
| "step": 1106 | |
| }, | |
| { | |
| "epoch": 0.8237947954698728, | |
| "grad_norm": 0.8728040618393719, | |
| "learning_rate": 4.146981627296588e-06, | |
| "loss": 3.6707, | |
| "step": 1107 | |
| }, | |
| { | |
| "epoch": 0.8245389642101346, | |
| "grad_norm": 0.9399683521116524, | |
| "learning_rate": 4.129483814523185e-06, | |
| "loss": 3.6644, | |
| "step": 1108 | |
| }, | |
| { | |
| "epoch": 0.8252831329503965, | |
| "grad_norm": 0.89167182668972, | |
| "learning_rate": 4.111986001749782e-06, | |
| "loss": 3.6624, | |
| "step": 1109 | |
| }, | |
| { | |
| "epoch": 0.8260273016906584, | |
| "grad_norm": 0.8921511012430581, | |
| "learning_rate": 4.0944881889763785e-06, | |
| "loss": 3.681, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.8267714704309203, | |
| "grad_norm": 0.9140327949690479, | |
| "learning_rate": 4.076990376202975e-06, | |
| "loss": 3.6714, | |
| "step": 1111 | |
| }, | |
| { | |
| "epoch": 0.8275156391711821, | |
| "grad_norm": 0.9008316079101544, | |
| "learning_rate": 4.059492563429572e-06, | |
| "loss": 3.6251, | |
| "step": 1112 | |
| }, | |
| { | |
| "epoch": 0.8282598079114439, | |
| "grad_norm": 0.8869153128498294, | |
| "learning_rate": 4.041994750656169e-06, | |
| "loss": 3.6746, | |
| "step": 1113 | |
| }, | |
| { | |
| "epoch": 0.8290039766517058, | |
| "grad_norm": 0.9232094126907991, | |
| "learning_rate": 4.024496937882765e-06, | |
| "loss": 3.6636, | |
| "step": 1114 | |
| }, | |
| { | |
| "epoch": 0.8297481453919676, | |
| "grad_norm": 0.8918099177467479, | |
| "learning_rate": 4.006999125109361e-06, | |
| "loss": 3.6883, | |
| "step": 1115 | |
| }, | |
| { | |
| "epoch": 0.8304923141322295, | |
| "grad_norm": 0.9395857647274934, | |
| "learning_rate": 3.989501312335958e-06, | |
| "loss": 3.6799, | |
| "step": 1116 | |
| }, | |
| { | |
| "epoch": 0.8312364828724913, | |
| "grad_norm": 0.8878031381874515, | |
| "learning_rate": 3.972003499562555e-06, | |
| "loss": 3.6895, | |
| "step": 1117 | |
| }, | |
| { | |
| "epoch": 0.8319806516127531, | |
| "grad_norm": 0.8614633087856474, | |
| "learning_rate": 3.9545056867891515e-06, | |
| "loss": 3.6942, | |
| "step": 1118 | |
| }, | |
| { | |
| "epoch": 0.8327248203530151, | |
| "grad_norm": 0.9077671011406638, | |
| "learning_rate": 3.937007874015748e-06, | |
| "loss": 3.6744, | |
| "step": 1119 | |
| }, | |
| { | |
| "epoch": 0.8334689890932769, | |
| "grad_norm": 0.9350147869240537, | |
| "learning_rate": 3.919510061242345e-06, | |
| "loss": 3.7048, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.8342131578335388, | |
| "grad_norm": 0.8665352755278202, | |
| "learning_rate": 3.902012248468942e-06, | |
| "loss": 3.6664, | |
| "step": 1121 | |
| }, | |
| { | |
| "epoch": 0.8349573265738006, | |
| "grad_norm": 0.9327060019299036, | |
| "learning_rate": 3.884514435695538e-06, | |
| "loss": 3.6833, | |
| "step": 1122 | |
| }, | |
| { | |
| "epoch": 0.8357014953140625, | |
| "grad_norm": 0.8988127638770026, | |
| "learning_rate": 3.867016622922135e-06, | |
| "loss": 3.7169, | |
| "step": 1123 | |
| }, | |
| { | |
| "epoch": 0.8364456640543243, | |
| "grad_norm": 0.9163573204164044, | |
| "learning_rate": 3.849518810148732e-06, | |
| "loss": 3.6694, | |
| "step": 1124 | |
| }, | |
| { | |
| "epoch": 0.8371898327945861, | |
| "grad_norm": 0.9024717850822429, | |
| "learning_rate": 3.8320209973753285e-06, | |
| "loss": 3.6866, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.837934001534848, | |
| "grad_norm": 0.8850032178563563, | |
| "learning_rate": 3.814523184601925e-06, | |
| "loss": 3.6588, | |
| "step": 1126 | |
| }, | |
| { | |
| "epoch": 0.8386781702751099, | |
| "grad_norm": 0.8721787099994606, | |
| "learning_rate": 3.7970253718285215e-06, | |
| "loss": 3.7072, | |
| "step": 1127 | |
| }, | |
| { | |
| "epoch": 0.8394223390153718, | |
| "grad_norm": 0.9081182520908622, | |
| "learning_rate": 3.7795275590551182e-06, | |
| "loss": 3.6689, | |
| "step": 1128 | |
| }, | |
| { | |
| "epoch": 0.8401665077556336, | |
| "grad_norm": 0.8496102510423653, | |
| "learning_rate": 3.762029746281715e-06, | |
| "loss": 3.6812, | |
| "step": 1129 | |
| }, | |
| { | |
| "epoch": 0.8409106764958955, | |
| "grad_norm": 0.8920090495588396, | |
| "learning_rate": 3.744531933508312e-06, | |
| "loss": 3.6671, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.8416548452361573, | |
| "grad_norm": 0.87790943860311, | |
| "learning_rate": 3.727034120734909e-06, | |
| "loss": 3.7011, | |
| "step": 1131 | |
| }, | |
| { | |
| "epoch": 0.8423990139764191, | |
| "grad_norm": 0.8952630001091338, | |
| "learning_rate": 3.7095363079615047e-06, | |
| "loss": 3.6616, | |
| "step": 1132 | |
| }, | |
| { | |
| "epoch": 0.843143182716681, | |
| "grad_norm": 0.8954767661382252, | |
| "learning_rate": 3.692038495188102e-06, | |
| "loss": 3.6917, | |
| "step": 1133 | |
| }, | |
| { | |
| "epoch": 0.8438873514569428, | |
| "grad_norm": 0.8977063823118526, | |
| "learning_rate": 3.6745406824146986e-06, | |
| "loss": 3.6919, | |
| "step": 1134 | |
| }, | |
| { | |
| "epoch": 0.8446315201972047, | |
| "grad_norm": 0.9024377182626876, | |
| "learning_rate": 3.6570428696412953e-06, | |
| "loss": 3.6886, | |
| "step": 1135 | |
| }, | |
| { | |
| "epoch": 0.8453756889374666, | |
| "grad_norm": 0.8887965765782548, | |
| "learning_rate": 3.639545056867892e-06, | |
| "loss": 3.6873, | |
| "step": 1136 | |
| }, | |
| { | |
| "epoch": 0.8461198576777285, | |
| "grad_norm": 0.891033194112756, | |
| "learning_rate": 3.6220472440944887e-06, | |
| "loss": 3.6566, | |
| "step": 1137 | |
| }, | |
| { | |
| "epoch": 0.8468640264179903, | |
| "grad_norm": 0.8812604388661017, | |
| "learning_rate": 3.604549431321085e-06, | |
| "loss": 3.7167, | |
| "step": 1138 | |
| }, | |
| { | |
| "epoch": 0.8476081951582521, | |
| "grad_norm": 0.904601523114026, | |
| "learning_rate": 3.5870516185476817e-06, | |
| "loss": 3.6819, | |
| "step": 1139 | |
| }, | |
| { | |
| "epoch": 0.848352363898514, | |
| "grad_norm": 0.8771966207563247, | |
| "learning_rate": 3.5695538057742785e-06, | |
| "loss": 3.6797, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.8490965326387758, | |
| "grad_norm": 0.8619287045102585, | |
| "learning_rate": 3.552055993000875e-06, | |
| "loss": 3.6185, | |
| "step": 1141 | |
| }, | |
| { | |
| "epoch": 0.8498407013790377, | |
| "grad_norm": 0.8915198019034517, | |
| "learning_rate": 3.534558180227472e-06, | |
| "loss": 3.6669, | |
| "step": 1142 | |
| }, | |
| { | |
| "epoch": 0.8505848701192995, | |
| "grad_norm": 0.8849708878056508, | |
| "learning_rate": 3.5170603674540686e-06, | |
| "loss": 3.678, | |
| "step": 1143 | |
| }, | |
| { | |
| "epoch": 0.8513290388595615, | |
| "grad_norm": 0.9120921538153527, | |
| "learning_rate": 3.499562554680665e-06, | |
| "loss": 3.6657, | |
| "step": 1144 | |
| }, | |
| { | |
| "epoch": 0.8520732075998233, | |
| "grad_norm": 0.8844433302011494, | |
| "learning_rate": 3.4820647419072616e-06, | |
| "loss": 3.6916, | |
| "step": 1145 | |
| }, | |
| { | |
| "epoch": 0.8528173763400851, | |
| "grad_norm": 0.9198938813798748, | |
| "learning_rate": 3.4645669291338583e-06, | |
| "loss": 3.6519, | |
| "step": 1146 | |
| }, | |
| { | |
| "epoch": 0.853561545080347, | |
| "grad_norm": 0.8878386077559169, | |
| "learning_rate": 3.447069116360455e-06, | |
| "loss": 3.6805, | |
| "step": 1147 | |
| }, | |
| { | |
| "epoch": 0.8543057138206088, | |
| "grad_norm": 0.8987398584105286, | |
| "learning_rate": 3.429571303587052e-06, | |
| "loss": 3.6967, | |
| "step": 1148 | |
| }, | |
| { | |
| "epoch": 0.8550498825608707, | |
| "grad_norm": 0.9069146734610795, | |
| "learning_rate": 3.412073490813649e-06, | |
| "loss": 3.6536, | |
| "step": 1149 | |
| }, | |
| { | |
| "epoch": 0.8557940513011325, | |
| "grad_norm": 0.9148097440447495, | |
| "learning_rate": 3.394575678040245e-06, | |
| "loss": 3.6476, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.8565382200413943, | |
| "grad_norm": 0.8932606202763849, | |
| "learning_rate": 3.377077865266842e-06, | |
| "loss": 3.697, | |
| "step": 1151 | |
| }, | |
| { | |
| "epoch": 0.8572823887816562, | |
| "grad_norm": 0.902909623816088, | |
| "learning_rate": 3.3595800524934387e-06, | |
| "loss": 3.6884, | |
| "step": 1152 | |
| }, | |
| { | |
| "epoch": 0.8580265575219181, | |
| "grad_norm": 0.9324965074726095, | |
| "learning_rate": 3.3420822397200354e-06, | |
| "loss": 3.6712, | |
| "step": 1153 | |
| }, | |
| { | |
| "epoch": 0.85877072626218, | |
| "grad_norm": 0.9165872510594605, | |
| "learning_rate": 3.324584426946632e-06, | |
| "loss": 3.6631, | |
| "step": 1154 | |
| }, | |
| { | |
| "epoch": 0.8595148950024418, | |
| "grad_norm": 0.8846215741742113, | |
| "learning_rate": 3.307086614173229e-06, | |
| "loss": 3.6747, | |
| "step": 1155 | |
| }, | |
| { | |
| "epoch": 0.8602590637427037, | |
| "grad_norm": 0.9052758220803853, | |
| "learning_rate": 3.289588801399825e-06, | |
| "loss": 3.675, | |
| "step": 1156 | |
| }, | |
| { | |
| "epoch": 0.8610032324829655, | |
| "grad_norm": 0.9188105737152846, | |
| "learning_rate": 3.272090988626422e-06, | |
| "loss": 3.6592, | |
| "step": 1157 | |
| }, | |
| { | |
| "epoch": 0.8617474012232273, | |
| "grad_norm": 0.8850438699636951, | |
| "learning_rate": 3.2545931758530186e-06, | |
| "loss": 3.6783, | |
| "step": 1158 | |
| }, | |
| { | |
| "epoch": 0.8624915699634892, | |
| "grad_norm": 0.8764235454239341, | |
| "learning_rate": 3.2370953630796153e-06, | |
| "loss": 3.6619, | |
| "step": 1159 | |
| }, | |
| { | |
| "epoch": 0.863235738703751, | |
| "grad_norm": 0.8857631993800366, | |
| "learning_rate": 3.219597550306212e-06, | |
| "loss": 3.6564, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.863979907444013, | |
| "grad_norm": 0.9043986766007083, | |
| "learning_rate": 3.2020997375328087e-06, | |
| "loss": 3.685, | |
| "step": 1161 | |
| }, | |
| { | |
| "epoch": 0.8647240761842748, | |
| "grad_norm": 0.9116985760977426, | |
| "learning_rate": 3.184601924759405e-06, | |
| "loss": 3.6622, | |
| "step": 1162 | |
| }, | |
| { | |
| "epoch": 0.8654682449245367, | |
| "grad_norm": 0.8691809010202386, | |
| "learning_rate": 3.1671041119860017e-06, | |
| "loss": 3.6758, | |
| "step": 1163 | |
| }, | |
| { | |
| "epoch": 0.8662124136647985, | |
| "grad_norm": 0.8864027951829367, | |
| "learning_rate": 3.1496062992125985e-06, | |
| "loss": 3.6648, | |
| "step": 1164 | |
| }, | |
| { | |
| "epoch": 0.8669565824050603, | |
| "grad_norm": 0.9217383413396099, | |
| "learning_rate": 3.1321084864391956e-06, | |
| "loss": 3.6919, | |
| "step": 1165 | |
| }, | |
| { | |
| "epoch": 0.8677007511453222, | |
| "grad_norm": 0.8792878560681597, | |
| "learning_rate": 3.1146106736657923e-06, | |
| "loss": 3.6695, | |
| "step": 1166 | |
| }, | |
| { | |
| "epoch": 0.868444919885584, | |
| "grad_norm": 0.9037278922238513, | |
| "learning_rate": 3.097112860892389e-06, | |
| "loss": 3.6771, | |
| "step": 1167 | |
| }, | |
| { | |
| "epoch": 0.8691890886258459, | |
| "grad_norm": 0.9258405037994375, | |
| "learning_rate": 3.0796150481189853e-06, | |
| "loss": 3.6662, | |
| "step": 1168 | |
| }, | |
| { | |
| "epoch": 0.8699332573661077, | |
| "grad_norm": 0.8766370057465983, | |
| "learning_rate": 3.062117235345582e-06, | |
| "loss": 3.6768, | |
| "step": 1169 | |
| }, | |
| { | |
| "epoch": 0.8706774261063697, | |
| "grad_norm": 0.8749800933927286, | |
| "learning_rate": 3.0446194225721788e-06, | |
| "loss": 3.683, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8714215948466315, | |
| "grad_norm": 0.9012138388011587, | |
| "learning_rate": 3.0271216097987755e-06, | |
| "loss": 3.6947, | |
| "step": 1171 | |
| }, | |
| { | |
| "epoch": 0.8721657635868934, | |
| "grad_norm": 0.8637483701109303, | |
| "learning_rate": 3.0096237970253722e-06, | |
| "loss": 3.6495, | |
| "step": 1172 | |
| }, | |
| { | |
| "epoch": 0.8729099323271552, | |
| "grad_norm": 0.8954274316198473, | |
| "learning_rate": 2.992125984251969e-06, | |
| "loss": 3.6535, | |
| "step": 1173 | |
| }, | |
| { | |
| "epoch": 0.873654101067417, | |
| "grad_norm": 0.8939345970807205, | |
| "learning_rate": 2.9746281714785652e-06, | |
| "loss": 3.6662, | |
| "step": 1174 | |
| }, | |
| { | |
| "epoch": 0.8743982698076789, | |
| "grad_norm": 0.8895803082996133, | |
| "learning_rate": 2.957130358705162e-06, | |
| "loss": 3.6718, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.8751424385479407, | |
| "grad_norm": 0.8699256017153515, | |
| "learning_rate": 2.9396325459317587e-06, | |
| "loss": 3.6791, | |
| "step": 1176 | |
| }, | |
| { | |
| "epoch": 0.8758866072882026, | |
| "grad_norm": 0.8767096298511755, | |
| "learning_rate": 2.9221347331583554e-06, | |
| "loss": 3.6649, | |
| "step": 1177 | |
| }, | |
| { | |
| "epoch": 0.8766307760284645, | |
| "grad_norm": 0.8952069928601892, | |
| "learning_rate": 2.904636920384952e-06, | |
| "loss": 3.6496, | |
| "step": 1178 | |
| }, | |
| { | |
| "epoch": 0.8773749447687264, | |
| "grad_norm": 0.909823423223946, | |
| "learning_rate": 2.887139107611549e-06, | |
| "loss": 3.6534, | |
| "step": 1179 | |
| }, | |
| { | |
| "epoch": 0.8781191135089882, | |
| "grad_norm": 0.9240178825526391, | |
| "learning_rate": 2.869641294838145e-06, | |
| "loss": 3.6242, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.87886328224925, | |
| "grad_norm": 0.8866253455689194, | |
| "learning_rate": 2.852143482064742e-06, | |
| "loss": 3.6155, | |
| "step": 1181 | |
| }, | |
| { | |
| "epoch": 0.8796074509895119, | |
| "grad_norm": 0.9073926018919111, | |
| "learning_rate": 2.8346456692913386e-06, | |
| "loss": 3.6527, | |
| "step": 1182 | |
| }, | |
| { | |
| "epoch": 0.8803516197297737, | |
| "grad_norm": 0.8889913702252497, | |
| "learning_rate": 2.8171478565179357e-06, | |
| "loss": 3.6897, | |
| "step": 1183 | |
| }, | |
| { | |
| "epoch": 0.8810957884700356, | |
| "grad_norm": 0.9191661618575031, | |
| "learning_rate": 2.7996500437445324e-06, | |
| "loss": 3.6523, | |
| "step": 1184 | |
| }, | |
| { | |
| "epoch": 0.8818399572102974, | |
| "grad_norm": 0.9139224840651652, | |
| "learning_rate": 2.782152230971129e-06, | |
| "loss": 3.6479, | |
| "step": 1185 | |
| }, | |
| { | |
| "epoch": 0.8825841259505592, | |
| "grad_norm": 0.9184555638449092, | |
| "learning_rate": 2.7646544181977255e-06, | |
| "loss": 3.6746, | |
| "step": 1186 | |
| }, | |
| { | |
| "epoch": 0.8833282946908212, | |
| "grad_norm": 0.886959238850157, | |
| "learning_rate": 2.747156605424322e-06, | |
| "loss": 3.6967, | |
| "step": 1187 | |
| }, | |
| { | |
| "epoch": 0.884072463431083, | |
| "grad_norm": 0.8940358336810101, | |
| "learning_rate": 2.729658792650919e-06, | |
| "loss": 3.6816, | |
| "step": 1188 | |
| }, | |
| { | |
| "epoch": 0.8848166321713449, | |
| "grad_norm": 0.9244831788898189, | |
| "learning_rate": 2.7121609798775156e-06, | |
| "loss": 3.6522, | |
| "step": 1189 | |
| }, | |
| { | |
| "epoch": 0.8855608009116067, | |
| "grad_norm": 0.8917865534365214, | |
| "learning_rate": 2.6946631671041123e-06, | |
| "loss": 3.6726, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8863049696518686, | |
| "grad_norm": 0.8958232010969741, | |
| "learning_rate": 2.677165354330709e-06, | |
| "loss": 3.6843, | |
| "step": 1191 | |
| }, | |
| { | |
| "epoch": 0.8870491383921304, | |
| "grad_norm": 0.8810789249367214, | |
| "learning_rate": 2.6596675415573053e-06, | |
| "loss": 3.673, | |
| "step": 1192 | |
| }, | |
| { | |
| "epoch": 0.8877933071323922, | |
| "grad_norm": 0.887451089458014, | |
| "learning_rate": 2.642169728783902e-06, | |
| "loss": 3.672, | |
| "step": 1193 | |
| }, | |
| { | |
| "epoch": 0.8885374758726541, | |
| "grad_norm": 0.8873732927255216, | |
| "learning_rate": 2.6246719160104988e-06, | |
| "loss": 3.6572, | |
| "step": 1194 | |
| }, | |
| { | |
| "epoch": 0.889281644612916, | |
| "grad_norm": 0.8859542366321156, | |
| "learning_rate": 2.6071741032370955e-06, | |
| "loss": 3.6489, | |
| "step": 1195 | |
| }, | |
| { | |
| "epoch": 0.8900258133531779, | |
| "grad_norm": 0.9050680889254548, | |
| "learning_rate": 2.5896762904636922e-06, | |
| "loss": 3.7016, | |
| "step": 1196 | |
| }, | |
| { | |
| "epoch": 0.8907699820934397, | |
| "grad_norm": 0.8879565767427062, | |
| "learning_rate": 2.5721784776902894e-06, | |
| "loss": 3.6515, | |
| "step": 1197 | |
| }, | |
| { | |
| "epoch": 0.8915141508337016, | |
| "grad_norm": 0.8992677569365484, | |
| "learning_rate": 2.5546806649168852e-06, | |
| "loss": 3.7102, | |
| "step": 1198 | |
| }, | |
| { | |
| "epoch": 0.8922583195739634, | |
| "grad_norm": 0.8823127650314926, | |
| "learning_rate": 2.537182852143482e-06, | |
| "loss": 3.7032, | |
| "step": 1199 | |
| }, | |
| { | |
| "epoch": 0.8930024883142252, | |
| "grad_norm": 0.9119477355413133, | |
| "learning_rate": 2.519685039370079e-06, | |
| "loss": 3.6695, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8937466570544871, | |
| "grad_norm": 0.8988221343691756, | |
| "learning_rate": 2.502187226596676e-06, | |
| "loss": 3.6734, | |
| "step": 1201 | |
| }, | |
| { | |
| "epoch": 0.8944908257947489, | |
| "grad_norm": 0.8658187626423547, | |
| "learning_rate": 2.484689413823272e-06, | |
| "loss": 3.6417, | |
| "step": 1202 | |
| }, | |
| { | |
| "epoch": 0.8952349945350108, | |
| "grad_norm": 0.8770446838213599, | |
| "learning_rate": 2.467191601049869e-06, | |
| "loss": 3.7121, | |
| "step": 1203 | |
| }, | |
| { | |
| "epoch": 0.8959791632752727, | |
| "grad_norm": 0.8881052108255448, | |
| "learning_rate": 2.4496937882764656e-06, | |
| "loss": 3.7138, | |
| "step": 1204 | |
| }, | |
| { | |
| "epoch": 0.8967233320155346, | |
| "grad_norm": 0.9035670356237357, | |
| "learning_rate": 2.4321959755030623e-06, | |
| "loss": 3.6447, | |
| "step": 1205 | |
| }, | |
| { | |
| "epoch": 0.8974675007557964, | |
| "grad_norm": 0.8763761785059622, | |
| "learning_rate": 2.414698162729659e-06, | |
| "loss": 3.639, | |
| "step": 1206 | |
| }, | |
| { | |
| "epoch": 0.8982116694960582, | |
| "grad_norm": 0.901893219740008, | |
| "learning_rate": 2.3972003499562557e-06, | |
| "loss": 3.6935, | |
| "step": 1207 | |
| }, | |
| { | |
| "epoch": 0.8989558382363201, | |
| "grad_norm": 0.9194824374635306, | |
| "learning_rate": 2.379702537182852e-06, | |
| "loss": 3.6879, | |
| "step": 1208 | |
| }, | |
| { | |
| "epoch": 0.8997000069765819, | |
| "grad_norm": 0.8562778514313033, | |
| "learning_rate": 2.362204724409449e-06, | |
| "loss": 3.642, | |
| "step": 1209 | |
| }, | |
| { | |
| "epoch": 0.9004441757168438, | |
| "grad_norm": 0.8770009981574711, | |
| "learning_rate": 2.344706911636046e-06, | |
| "loss": 3.6671, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.9011883444571056, | |
| "grad_norm": 0.889955846295838, | |
| "learning_rate": 2.327209098862642e-06, | |
| "loss": 3.6699, | |
| "step": 1211 | |
| }, | |
| { | |
| "epoch": 0.9019325131973676, | |
| "grad_norm": 0.8787680231696078, | |
| "learning_rate": 2.309711286089239e-06, | |
| "loss": 3.6533, | |
| "step": 1212 | |
| }, | |
| { | |
| "epoch": 0.9026766819376294, | |
| "grad_norm": 0.897158830776579, | |
| "learning_rate": 2.2922134733158356e-06, | |
| "loss": 3.6709, | |
| "step": 1213 | |
| }, | |
| { | |
| "epoch": 0.9034208506778912, | |
| "grad_norm": 0.9555502429723852, | |
| "learning_rate": 2.2747156605424323e-06, | |
| "loss": 3.6452, | |
| "step": 1214 | |
| }, | |
| { | |
| "epoch": 0.9041650194181531, | |
| "grad_norm": 0.8887303869970781, | |
| "learning_rate": 2.257217847769029e-06, | |
| "loss": 3.6472, | |
| "step": 1215 | |
| }, | |
| { | |
| "epoch": 0.9049091881584149, | |
| "grad_norm": 0.9262657031154059, | |
| "learning_rate": 2.2397200349956258e-06, | |
| "loss": 3.6971, | |
| "step": 1216 | |
| }, | |
| { | |
| "epoch": 0.9056533568986768, | |
| "grad_norm": 0.9076122206136279, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 3.6826, | |
| "step": 1217 | |
| }, | |
| { | |
| "epoch": 0.9063975256389386, | |
| "grad_norm": 0.9231192517524169, | |
| "learning_rate": 2.2047244094488192e-06, | |
| "loss": 3.7132, | |
| "step": 1218 | |
| }, | |
| { | |
| "epoch": 0.9071416943792004, | |
| "grad_norm": 0.8648106402758707, | |
| "learning_rate": 2.187226596675416e-06, | |
| "loss": 3.6562, | |
| "step": 1219 | |
| }, | |
| { | |
| "epoch": 0.9078858631194623, | |
| "grad_norm": 0.90799607069842, | |
| "learning_rate": 2.1697287839020122e-06, | |
| "loss": 3.6905, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.9086300318597242, | |
| "grad_norm": 0.9323750191794503, | |
| "learning_rate": 2.152230971128609e-06, | |
| "loss": 3.6531, | |
| "step": 1221 | |
| }, | |
| { | |
| "epoch": 0.9093742005999861, | |
| "grad_norm": 0.9144538594705779, | |
| "learning_rate": 2.1347331583552057e-06, | |
| "loss": 3.6681, | |
| "step": 1222 | |
| }, | |
| { | |
| "epoch": 0.9101183693402479, | |
| "grad_norm": 0.8760433336155817, | |
| "learning_rate": 2.1172353455818024e-06, | |
| "loss": 3.6614, | |
| "step": 1223 | |
| }, | |
| { | |
| "epoch": 0.9108625380805098, | |
| "grad_norm": 0.8804460584318855, | |
| "learning_rate": 2.099737532808399e-06, | |
| "loss": 3.6433, | |
| "step": 1224 | |
| }, | |
| { | |
| "epoch": 0.9116067068207716, | |
| "grad_norm": 0.886546183809884, | |
| "learning_rate": 2.082239720034996e-06, | |
| "loss": 3.6558, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.9123508755610334, | |
| "grad_norm": 0.9237111970673366, | |
| "learning_rate": 2.0647419072615926e-06, | |
| "loss": 3.6782, | |
| "step": 1226 | |
| }, | |
| { | |
| "epoch": 0.9130950443012953, | |
| "grad_norm": 0.9040867989770951, | |
| "learning_rate": 2.0472440944881893e-06, | |
| "loss": 3.6406, | |
| "step": 1227 | |
| }, | |
| { | |
| "epoch": 0.9138392130415571, | |
| "grad_norm": 0.9008271800304214, | |
| "learning_rate": 2.029746281714786e-06, | |
| "loss": 3.657, | |
| "step": 1228 | |
| }, | |
| { | |
| "epoch": 0.9145833817818191, | |
| "grad_norm": 0.8971076418109339, | |
| "learning_rate": 2.0122484689413823e-06, | |
| "loss": 3.6861, | |
| "step": 1229 | |
| }, | |
| { | |
| "epoch": 0.9153275505220809, | |
| "grad_norm": 0.8844912806833802, | |
| "learning_rate": 1.994750656167979e-06, | |
| "loss": 3.6576, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.9160717192623428, | |
| "grad_norm": 0.9012273055340636, | |
| "learning_rate": 1.9772528433945757e-06, | |
| "loss": 3.6356, | |
| "step": 1231 | |
| }, | |
| { | |
| "epoch": 0.9168158880026046, | |
| "grad_norm": 0.8907075719836548, | |
| "learning_rate": 1.9597550306211725e-06, | |
| "loss": 3.6619, | |
| "step": 1232 | |
| }, | |
| { | |
| "epoch": 0.9175600567428664, | |
| "grad_norm": 0.8977368963137614, | |
| "learning_rate": 1.942257217847769e-06, | |
| "loss": 3.6539, | |
| "step": 1233 | |
| }, | |
| { | |
| "epoch": 0.9183042254831283, | |
| "grad_norm": 0.9037418400907917, | |
| "learning_rate": 1.924759405074366e-06, | |
| "loss": 3.6579, | |
| "step": 1234 | |
| }, | |
| { | |
| "epoch": 0.9190483942233901, | |
| "grad_norm": 0.8742157776384799, | |
| "learning_rate": 1.9072615923009624e-06, | |
| "loss": 3.6397, | |
| "step": 1235 | |
| }, | |
| { | |
| "epoch": 0.919792562963652, | |
| "grad_norm": 0.9322227116692695, | |
| "learning_rate": 1.8897637795275591e-06, | |
| "loss": 3.6906, | |
| "step": 1236 | |
| }, | |
| { | |
| "epoch": 0.9205367317039138, | |
| "grad_norm": 0.8846717393950376, | |
| "learning_rate": 1.872265966754156e-06, | |
| "loss": 3.6837, | |
| "step": 1237 | |
| }, | |
| { | |
| "epoch": 0.9212809004441758, | |
| "grad_norm": 0.8972065215518901, | |
| "learning_rate": 1.8547681539807523e-06, | |
| "loss": 3.669, | |
| "step": 1238 | |
| }, | |
| { | |
| "epoch": 0.9220250691844376, | |
| "grad_norm": 0.9034199090087826, | |
| "learning_rate": 1.8372703412073493e-06, | |
| "loss": 3.6891, | |
| "step": 1239 | |
| }, | |
| { | |
| "epoch": 0.9227692379246994, | |
| "grad_norm": 0.8848498740215638, | |
| "learning_rate": 1.819772528433946e-06, | |
| "loss": 3.6384, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.9235134066649613, | |
| "grad_norm": 0.8893439357127124, | |
| "learning_rate": 1.8022747156605425e-06, | |
| "loss": 3.6639, | |
| "step": 1241 | |
| }, | |
| { | |
| "epoch": 0.9242575754052231, | |
| "grad_norm": 0.913738366216189, | |
| "learning_rate": 1.7847769028871392e-06, | |
| "loss": 3.6928, | |
| "step": 1242 | |
| }, | |
| { | |
| "epoch": 0.925001744145485, | |
| "grad_norm": 0.8904789537908234, | |
| "learning_rate": 1.767279090113736e-06, | |
| "loss": 3.7113, | |
| "step": 1243 | |
| }, | |
| { | |
| "epoch": 0.9257459128857468, | |
| "grad_norm": 0.8943787374205897, | |
| "learning_rate": 1.7497812773403325e-06, | |
| "loss": 3.6555, | |
| "step": 1244 | |
| }, | |
| { | |
| "epoch": 0.9264900816260087, | |
| "grad_norm": 0.9022408191456482, | |
| "learning_rate": 1.7322834645669292e-06, | |
| "loss": 3.6556, | |
| "step": 1245 | |
| }, | |
| { | |
| "epoch": 0.9272342503662706, | |
| "grad_norm": 0.8883746159124616, | |
| "learning_rate": 1.714785651793526e-06, | |
| "loss": 3.6685, | |
| "step": 1246 | |
| }, | |
| { | |
| "epoch": 0.9279784191065324, | |
| "grad_norm": 0.8944684704851439, | |
| "learning_rate": 1.6972878390201224e-06, | |
| "loss": 3.7053, | |
| "step": 1247 | |
| }, | |
| { | |
| "epoch": 0.9287225878467943, | |
| "grad_norm": 0.8891288058879199, | |
| "learning_rate": 1.6797900262467193e-06, | |
| "loss": 3.6422, | |
| "step": 1248 | |
| }, | |
| { | |
| "epoch": 0.9294667565870561, | |
| "grad_norm": 0.9140418350837658, | |
| "learning_rate": 1.662292213473316e-06, | |
| "loss": 3.6533, | |
| "step": 1249 | |
| }, | |
| { | |
| "epoch": 0.930210925327318, | |
| "grad_norm": 0.8927113531702113, | |
| "learning_rate": 1.6447944006999126e-06, | |
| "loss": 3.6575, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.9309550940675798, | |
| "grad_norm": 0.9066218376813425, | |
| "learning_rate": 1.6272965879265093e-06, | |
| "loss": 3.7052, | |
| "step": 1251 | |
| }, | |
| { | |
| "epoch": 0.9316992628078417, | |
| "grad_norm": 0.9150251398932568, | |
| "learning_rate": 1.609798775153106e-06, | |
| "loss": 3.6716, | |
| "step": 1252 | |
| }, | |
| { | |
| "epoch": 0.9324434315481035, | |
| "grad_norm": 0.9272896341703473, | |
| "learning_rate": 1.5923009623797025e-06, | |
| "loss": 3.674, | |
| "step": 1253 | |
| }, | |
| { | |
| "epoch": 0.9331876002883653, | |
| "grad_norm": 0.8976457892841331, | |
| "learning_rate": 1.5748031496062992e-06, | |
| "loss": 3.6581, | |
| "step": 1254 | |
| }, | |
| { | |
| "epoch": 0.9339317690286273, | |
| "grad_norm": 0.8570212562703146, | |
| "learning_rate": 1.5573053368328962e-06, | |
| "loss": 3.6563, | |
| "step": 1255 | |
| }, | |
| { | |
| "epoch": 0.9346759377688891, | |
| "grad_norm": 0.9040238934066118, | |
| "learning_rate": 1.5398075240594927e-06, | |
| "loss": 3.6902, | |
| "step": 1256 | |
| }, | |
| { | |
| "epoch": 0.935420106509151, | |
| "grad_norm": 0.9107287216958391, | |
| "learning_rate": 1.5223097112860894e-06, | |
| "loss": 3.6863, | |
| "step": 1257 | |
| }, | |
| { | |
| "epoch": 0.9361642752494128, | |
| "grad_norm": 0.8914015610994018, | |
| "learning_rate": 1.5048118985126861e-06, | |
| "loss": 3.6637, | |
| "step": 1258 | |
| }, | |
| { | |
| "epoch": 0.9369084439896747, | |
| "grad_norm": 0.8760972892942224, | |
| "learning_rate": 1.4873140857392826e-06, | |
| "loss": 3.6992, | |
| "step": 1259 | |
| }, | |
| { | |
| "epoch": 0.9376526127299365, | |
| "grad_norm": 0.9095736252593042, | |
| "learning_rate": 1.4698162729658793e-06, | |
| "loss": 3.6539, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.9383967814701983, | |
| "grad_norm": 0.8660339004664301, | |
| "learning_rate": 1.452318460192476e-06, | |
| "loss": 3.6615, | |
| "step": 1261 | |
| }, | |
| { | |
| "epoch": 0.9391409502104602, | |
| "grad_norm": 0.9037567850566451, | |
| "learning_rate": 1.4348206474190726e-06, | |
| "loss": 3.6935, | |
| "step": 1262 | |
| }, | |
| { | |
| "epoch": 0.9398851189507221, | |
| "grad_norm": 0.8852258936690676, | |
| "learning_rate": 1.4173228346456693e-06, | |
| "loss": 3.6321, | |
| "step": 1263 | |
| }, | |
| { | |
| "epoch": 0.940629287690984, | |
| "grad_norm": 0.863952211047765, | |
| "learning_rate": 1.3998250218722662e-06, | |
| "loss": 3.6658, | |
| "step": 1264 | |
| }, | |
| { | |
| "epoch": 0.9413734564312458, | |
| "grad_norm": 0.8895691169244767, | |
| "learning_rate": 1.3823272090988627e-06, | |
| "loss": 3.6515, | |
| "step": 1265 | |
| }, | |
| { | |
| "epoch": 0.9421176251715077, | |
| "grad_norm": 1.6282146438498717, | |
| "learning_rate": 1.3648293963254594e-06, | |
| "loss": 3.6747, | |
| "step": 1266 | |
| }, | |
| { | |
| "epoch": 0.9428617939117695, | |
| "grad_norm": 0.9090492637919169, | |
| "learning_rate": 1.3473315835520562e-06, | |
| "loss": 3.6965, | |
| "step": 1267 | |
| }, | |
| { | |
| "epoch": 0.9436059626520313, | |
| "grad_norm": 0.8943477810509683, | |
| "learning_rate": 1.3298337707786527e-06, | |
| "loss": 3.7002, | |
| "step": 1268 | |
| }, | |
| { | |
| "epoch": 0.9443501313922932, | |
| "grad_norm": 0.8692926017296629, | |
| "learning_rate": 1.3123359580052494e-06, | |
| "loss": 3.6216, | |
| "step": 1269 | |
| }, | |
| { | |
| "epoch": 0.945094300132555, | |
| "grad_norm": 0.8946006158678829, | |
| "learning_rate": 1.2948381452318461e-06, | |
| "loss": 3.6732, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.9458384688728169, | |
| "grad_norm": 0.9060193834580261, | |
| "learning_rate": 1.2773403324584426e-06, | |
| "loss": 3.6825, | |
| "step": 1271 | |
| }, | |
| { | |
| "epoch": 0.9465826376130788, | |
| "grad_norm": 0.8887761927456032, | |
| "learning_rate": 1.2598425196850396e-06, | |
| "loss": 3.6817, | |
| "step": 1272 | |
| }, | |
| { | |
| "epoch": 0.9473268063533407, | |
| "grad_norm": 0.8867291945212894, | |
| "learning_rate": 1.242344706911636e-06, | |
| "loss": 3.6742, | |
| "step": 1273 | |
| }, | |
| { | |
| "epoch": 0.9480709750936025, | |
| "grad_norm": 0.8720971026452644, | |
| "learning_rate": 1.2248468941382328e-06, | |
| "loss": 3.6555, | |
| "step": 1274 | |
| }, | |
| { | |
| "epoch": 0.9488151438338643, | |
| "grad_norm": 1.0108968892085608, | |
| "learning_rate": 1.2073490813648295e-06, | |
| "loss": 3.6604, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.9495593125741262, | |
| "grad_norm": 0.9026377387331183, | |
| "learning_rate": 1.189851268591426e-06, | |
| "loss": 3.6477, | |
| "step": 1276 | |
| }, | |
| { | |
| "epoch": 0.950303481314388, | |
| "grad_norm": 0.8821479238356694, | |
| "learning_rate": 1.172353455818023e-06, | |
| "loss": 3.6578, | |
| "step": 1277 | |
| }, | |
| { | |
| "epoch": 0.9510476500546499, | |
| "grad_norm": 0.9102098830912364, | |
| "learning_rate": 1.1548556430446194e-06, | |
| "loss": 3.666, | |
| "step": 1278 | |
| }, | |
| { | |
| "epoch": 0.9517918187949117, | |
| "grad_norm": 0.8878494367276946, | |
| "learning_rate": 1.1373578302712162e-06, | |
| "loss": 3.6786, | |
| "step": 1279 | |
| }, | |
| { | |
| "epoch": 0.9525359875351737, | |
| "grad_norm": 0.9374269650922404, | |
| "learning_rate": 1.1198600174978129e-06, | |
| "loss": 3.692, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.9532801562754355, | |
| "grad_norm": 0.8710137643227259, | |
| "learning_rate": 1.1023622047244096e-06, | |
| "loss": 3.649, | |
| "step": 1281 | |
| }, | |
| { | |
| "epoch": 0.9540243250156973, | |
| "grad_norm": 0.894314111824436, | |
| "learning_rate": 1.0848643919510061e-06, | |
| "loss": 3.6795, | |
| "step": 1282 | |
| }, | |
| { | |
| "epoch": 0.9547684937559592, | |
| "grad_norm": 0.8965829210846336, | |
| "learning_rate": 1.0673665791776028e-06, | |
| "loss": 3.6662, | |
| "step": 1283 | |
| }, | |
| { | |
| "epoch": 0.955512662496221, | |
| "grad_norm": 0.9110169368458857, | |
| "learning_rate": 1.0498687664041996e-06, | |
| "loss": 3.6524, | |
| "step": 1284 | |
| }, | |
| { | |
| "epoch": 0.9562568312364829, | |
| "grad_norm": 0.9003553654134852, | |
| "learning_rate": 1.0323709536307963e-06, | |
| "loss": 3.6549, | |
| "step": 1285 | |
| }, | |
| { | |
| "epoch": 0.9570009999767447, | |
| "grad_norm": 0.8840532130415062, | |
| "learning_rate": 1.014873140857393e-06, | |
| "loss": 3.6558, | |
| "step": 1286 | |
| }, | |
| { | |
| "epoch": 0.9577451687170065, | |
| "grad_norm": 0.8883477888667453, | |
| "learning_rate": 9.973753280839895e-07, | |
| "loss": 3.6995, | |
| "step": 1287 | |
| }, | |
| { | |
| "epoch": 0.9584893374572684, | |
| "grad_norm": 0.8703843149191622, | |
| "learning_rate": 9.798775153105862e-07, | |
| "loss": 3.6484, | |
| "step": 1288 | |
| }, | |
| { | |
| "epoch": 0.9592335061975303, | |
| "grad_norm": 0.9088831124003299, | |
| "learning_rate": 9.62379702537183e-07, | |
| "loss": 3.6681, | |
| "step": 1289 | |
| }, | |
| { | |
| "epoch": 0.9599776749377922, | |
| "grad_norm": 0.8997417826254724, | |
| "learning_rate": 9.448818897637796e-07, | |
| "loss": 3.7126, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.960721843678054, | |
| "grad_norm": 0.9161195980748456, | |
| "learning_rate": 9.273840769903762e-07, | |
| "loss": 3.6476, | |
| "step": 1291 | |
| }, | |
| { | |
| "epoch": 0.9614660124183159, | |
| "grad_norm": 0.9041220125435347, | |
| "learning_rate": 9.09886264216973e-07, | |
| "loss": 3.671, | |
| "step": 1292 | |
| }, | |
| { | |
| "epoch": 0.9622101811585777, | |
| "grad_norm": 0.8790824823958725, | |
| "learning_rate": 8.923884514435696e-07, | |
| "loss": 3.6356, | |
| "step": 1293 | |
| }, | |
| { | |
| "epoch": 0.9629543498988395, | |
| "grad_norm": 0.8898803262515801, | |
| "learning_rate": 8.748906386701662e-07, | |
| "loss": 3.6533, | |
| "step": 1294 | |
| }, | |
| { | |
| "epoch": 0.9636985186391014, | |
| "grad_norm": 0.9000530135977408, | |
| "learning_rate": 8.57392825896763e-07, | |
| "loss": 3.6635, | |
| "step": 1295 | |
| }, | |
| { | |
| "epoch": 0.9644426873793632, | |
| "grad_norm": 0.8740224567043798, | |
| "learning_rate": 8.398950131233597e-07, | |
| "loss": 3.6558, | |
| "step": 1296 | |
| }, | |
| { | |
| "epoch": 0.9651868561196251, | |
| "grad_norm": 0.8769600344544728, | |
| "learning_rate": 8.223972003499563e-07, | |
| "loss": 3.68, | |
| "step": 1297 | |
| }, | |
| { | |
| "epoch": 0.965931024859887, | |
| "grad_norm": 0.9147412594233745, | |
| "learning_rate": 8.04899387576553e-07, | |
| "loss": 3.6828, | |
| "step": 1298 | |
| }, | |
| { | |
| "epoch": 0.9666751936001489, | |
| "grad_norm": 0.9166492316715372, | |
| "learning_rate": 7.874015748031496e-07, | |
| "loss": 3.6495, | |
| "step": 1299 | |
| }, | |
| { | |
| "epoch": 0.9674193623404107, | |
| "grad_norm": 0.8920538856956253, | |
| "learning_rate": 7.699037620297463e-07, | |
| "loss": 3.6417, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.9681635310806725, | |
| "grad_norm": 0.9206554733983152, | |
| "learning_rate": 7.524059492563431e-07, | |
| "loss": 3.6676, | |
| "step": 1301 | |
| }, | |
| { | |
| "epoch": 0.9689076998209344, | |
| "grad_norm": 0.8883910583491523, | |
| "learning_rate": 7.349081364829397e-07, | |
| "loss": 3.7301, | |
| "step": 1302 | |
| }, | |
| { | |
| "epoch": 0.9696518685611962, | |
| "grad_norm": 0.9122900357813901, | |
| "learning_rate": 7.174103237095363e-07, | |
| "loss": 3.6507, | |
| "step": 1303 | |
| }, | |
| { | |
| "epoch": 0.9703960373014581, | |
| "grad_norm": 0.9067191755450721, | |
| "learning_rate": 6.999125109361331e-07, | |
| "loss": 3.6446, | |
| "step": 1304 | |
| }, | |
| { | |
| "epoch": 0.9711402060417199, | |
| "grad_norm": 0.8957121057492408, | |
| "learning_rate": 6.824146981627297e-07, | |
| "loss": 3.6718, | |
| "step": 1305 | |
| }, | |
| { | |
| "epoch": 0.9718843747819819, | |
| "grad_norm": 0.8754169349296737, | |
| "learning_rate": 6.649168853893263e-07, | |
| "loss": 3.6971, | |
| "step": 1306 | |
| }, | |
| { | |
| "epoch": 0.9726285435222437, | |
| "grad_norm": 0.9129030475091213, | |
| "learning_rate": 6.474190726159231e-07, | |
| "loss": 3.6625, | |
| "step": 1307 | |
| }, | |
| { | |
| "epoch": 0.9733727122625055, | |
| "grad_norm": 0.9597341571289971, | |
| "learning_rate": 6.299212598425198e-07, | |
| "loss": 3.6765, | |
| "step": 1308 | |
| }, | |
| { | |
| "epoch": 0.9741168810027674, | |
| "grad_norm": 0.8855990806767573, | |
| "learning_rate": 6.124234470691164e-07, | |
| "loss": 3.6738, | |
| "step": 1309 | |
| }, | |
| { | |
| "epoch": 0.9748610497430292, | |
| "grad_norm": 0.8731697162397332, | |
| "learning_rate": 5.94925634295713e-07, | |
| "loss": 3.671, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.9756052184832911, | |
| "grad_norm": 0.889076027278804, | |
| "learning_rate": 5.774278215223097e-07, | |
| "loss": 3.6786, | |
| "step": 1311 | |
| }, | |
| { | |
| "epoch": 0.9763493872235529, | |
| "grad_norm": 0.9123012537459776, | |
| "learning_rate": 5.599300087489064e-07, | |
| "loss": 3.6755, | |
| "step": 1312 | |
| }, | |
| { | |
| "epoch": 0.9770935559638148, | |
| "grad_norm": 0.8503967788672535, | |
| "learning_rate": 5.424321959755031e-07, | |
| "loss": 3.6612, | |
| "step": 1313 | |
| }, | |
| { | |
| "epoch": 0.9778377247040766, | |
| "grad_norm": 0.8679172387534995, | |
| "learning_rate": 5.249343832020998e-07, | |
| "loss": 3.6644, | |
| "step": 1314 | |
| }, | |
| { | |
| "epoch": 0.9785818934443385, | |
| "grad_norm": 0.8930197520021473, | |
| "learning_rate": 5.074365704286965e-07, | |
| "loss": 3.71, | |
| "step": 1315 | |
| }, | |
| { | |
| "epoch": 0.9793260621846004, | |
| "grad_norm": 0.8925851814973809, | |
| "learning_rate": 4.899387576552931e-07, | |
| "loss": 3.653, | |
| "step": 1316 | |
| }, | |
| { | |
| "epoch": 0.9800702309248622, | |
| "grad_norm": 0.8976139574296512, | |
| "learning_rate": 4.724409448818898e-07, | |
| "loss": 3.6783, | |
| "step": 1317 | |
| }, | |
| { | |
| "epoch": 0.9808143996651241, | |
| "grad_norm": 0.894305941806524, | |
| "learning_rate": 4.549431321084865e-07, | |
| "loss": 3.66, | |
| "step": 1318 | |
| }, | |
| { | |
| "epoch": 0.9815585684053859, | |
| "grad_norm": 0.8913965941291434, | |
| "learning_rate": 4.374453193350831e-07, | |
| "loss": 3.6885, | |
| "step": 1319 | |
| }, | |
| { | |
| "epoch": 0.9823027371456478, | |
| "grad_norm": 0.9064247768086454, | |
| "learning_rate": 4.1994750656167983e-07, | |
| "loss": 3.6737, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.9830469058859096, | |
| "grad_norm": 0.9534593131044328, | |
| "learning_rate": 4.024496937882765e-07, | |
| "loss": 3.7017, | |
| "step": 1321 | |
| }, | |
| { | |
| "epoch": 0.9837910746261714, | |
| "grad_norm": 0.8855451507419254, | |
| "learning_rate": 3.8495188101487317e-07, | |
| "loss": 3.6624, | |
| "step": 1322 | |
| }, | |
| { | |
| "epoch": 0.9845352433664334, | |
| "grad_norm": 0.8863824522059905, | |
| "learning_rate": 3.6745406824146983e-07, | |
| "loss": 3.6863, | |
| "step": 1323 | |
| }, | |
| { | |
| "epoch": 0.9852794121066952, | |
| "grad_norm": 0.8767736624513971, | |
| "learning_rate": 3.4995625546806655e-07, | |
| "loss": 3.6447, | |
| "step": 1324 | |
| }, | |
| { | |
| "epoch": 0.9860235808469571, | |
| "grad_norm": 0.8637176295149417, | |
| "learning_rate": 3.3245844269466317e-07, | |
| "loss": 3.6862, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.9867677495872189, | |
| "grad_norm": 0.9004029354072568, | |
| "learning_rate": 3.149606299212599e-07, | |
| "loss": 3.706, | |
| "step": 1326 | |
| }, | |
| { | |
| "epoch": 0.9875119183274808, | |
| "grad_norm": 0.8869101669443029, | |
| "learning_rate": 2.974628171478565e-07, | |
| "loss": 3.7066, | |
| "step": 1327 | |
| }, | |
| { | |
| "epoch": 0.9882560870677426, | |
| "grad_norm": 0.8821238548579676, | |
| "learning_rate": 2.799650043744532e-07, | |
| "loss": 3.6487, | |
| "step": 1328 | |
| }, | |
| { | |
| "epoch": 0.9890002558080044, | |
| "grad_norm": 0.8802028357307297, | |
| "learning_rate": 2.624671916010499e-07, | |
| "loss": 3.6742, | |
| "step": 1329 | |
| }, | |
| { | |
| "epoch": 0.9897444245482663, | |
| "grad_norm": 0.8607889513879532, | |
| "learning_rate": 2.4496937882764656e-07, | |
| "loss": 3.6474, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.9904885932885281, | |
| "grad_norm": 0.8876516845941499, | |
| "learning_rate": 2.2747156605424325e-07, | |
| "loss": 3.6543, | |
| "step": 1331 | |
| }, | |
| { | |
| "epoch": 0.9912327620287901, | |
| "grad_norm": 0.8758331824727327, | |
| "learning_rate": 2.0997375328083992e-07, | |
| "loss": 3.6654, | |
| "step": 1332 | |
| }, | |
| { | |
| "epoch": 0.9919769307690519, | |
| "grad_norm": 0.8520552905672313, | |
| "learning_rate": 1.9247594050743658e-07, | |
| "loss": 3.6452, | |
| "step": 1333 | |
| }, | |
| { | |
| "epoch": 0.9927210995093138, | |
| "grad_norm": 0.879611269163845, | |
| "learning_rate": 1.7497812773403328e-07, | |
| "loss": 3.666, | |
| "step": 1334 | |
| }, | |
| { | |
| "epoch": 0.9934652682495756, | |
| "grad_norm": 0.8908683185953267, | |
| "learning_rate": 1.5748031496062994e-07, | |
| "loss": 3.6408, | |
| "step": 1335 | |
| }, | |
| { | |
| "epoch": 0.9942094369898374, | |
| "grad_norm": 0.933343748325911, | |
| "learning_rate": 1.399825021872266e-07, | |
| "loss": 3.731, | |
| "step": 1336 | |
| }, | |
| { | |
| "epoch": 0.9949536057300993, | |
| "grad_norm": 0.8777242550161365, | |
| "learning_rate": 1.2248468941382328e-07, | |
| "loss": 3.6323, | |
| "step": 1337 | |
| }, | |
| { | |
| "epoch": 0.9956977744703611, | |
| "grad_norm": 0.9040382145591546, | |
| "learning_rate": 1.0498687664041996e-07, | |
| "loss": 3.6591, | |
| "step": 1338 | |
| }, | |
| { | |
| "epoch": 0.996441943210623, | |
| "grad_norm": 0.8966348611662436, | |
| "learning_rate": 8.748906386701664e-08, | |
| "loss": 3.6302, | |
| "step": 1339 | |
| }, | |
| { | |
| "epoch": 0.9971861119508849, | |
| "grad_norm": 0.9178574429401175, | |
| "learning_rate": 6.99912510936133e-08, | |
| "loss": 3.6943, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9979302806911468, | |
| "grad_norm": 0.8830349088112303, | |
| "learning_rate": 5.249343832020998e-08, | |
| "loss": 3.6792, | |
| "step": 1341 | |
| }, | |
| { | |
| "epoch": 0.9986744494314086, | |
| "grad_norm": 0.9211530735866087, | |
| "learning_rate": 3.499562554680665e-08, | |
| "loss": 3.6992, | |
| "step": 1342 | |
| }, | |
| { | |
| "epoch": 0.9994186181716704, | |
| "grad_norm": 0.9173525784935835, | |
| "learning_rate": 1.7497812773403326e-08, | |
| "loss": 3.6519, | |
| "step": 1343 | |
| }, | |
| { | |
| "epoch": 0.9994186181716704, | |
| "step": 1343, | |
| "total_flos": 7026499732897792.0, | |
| "train_loss": 3.791093183469169, | |
| "train_runtime": 74483.0839, | |
| "train_samples_per_second": 9.237, | |
| "train_steps_per_second": 0.018 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 1343, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 200, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 7026499732897792.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |