| { | |
| "best_metric": 0.046638909727334976, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.054745774310545406, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.000273728871552727, | |
| "grad_norm": 0.6941263675689697, | |
| "learning_rate": 1e-05, | |
| "loss": 0.8293, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000273728871552727, | |
| "eval_loss": 2.9798033237457275, | |
| "eval_runtime": 261.1764, | |
| "eval_samples_per_second": 23.559, | |
| "eval_steps_per_second": 5.893, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000547457743105454, | |
| "grad_norm": 1.2184916734695435, | |
| "learning_rate": 2e-05, | |
| "loss": 1.3028, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.000821186614658181, | |
| "grad_norm": 1.0150651931762695, | |
| "learning_rate": 3e-05, | |
| "loss": 1.0134, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.001094915486210908, | |
| "grad_norm": 0.9276098012924194, | |
| "learning_rate": 4e-05, | |
| "loss": 0.9233, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0013686443577636351, | |
| "grad_norm": 1.511775255203247, | |
| "learning_rate": 5e-05, | |
| "loss": 1.4922, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.001642373229316362, | |
| "grad_norm": 1.647055983543396, | |
| "learning_rate": 6e-05, | |
| "loss": 1.6422, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0019161021008690892, | |
| "grad_norm": 1.019641399383545, | |
| "learning_rate": 7e-05, | |
| "loss": 0.8935, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.002189830972421816, | |
| "grad_norm": 1.17073392868042, | |
| "learning_rate": 8e-05, | |
| "loss": 0.6967, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.002463559843974543, | |
| "grad_norm": 2.7586605548858643, | |
| "learning_rate": 9e-05, | |
| "loss": 0.8508, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0027372887155272703, | |
| "grad_norm": 0.6207762360572815, | |
| "learning_rate": 0.0001, | |
| "loss": 0.3636, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0030110175870799974, | |
| "grad_norm": 0.7272778153419495, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.3618, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.003284746458632724, | |
| "grad_norm": 0.5148478746414185, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.3207, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.003558475330185451, | |
| "grad_norm": 0.8237785696983337, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.2985, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0038322042017381783, | |
| "grad_norm": 0.8858790397644043, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.2782, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0041059330732909054, | |
| "grad_norm": 0.8292993903160095, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.2169, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.004379661944843632, | |
| "grad_norm": 0.8542286157608032, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.1566, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.00465339081639636, | |
| "grad_norm": 0.533530592918396, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.1346, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.004927119687949086, | |
| "grad_norm": 0.6270493865013123, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.1376, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.005200848559501813, | |
| "grad_norm": 0.8777332305908203, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.1615, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.005474577431054541, | |
| "grad_norm": 0.9948918223381042, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.1516, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.005748306302607267, | |
| "grad_norm": 0.8603919148445129, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.1271, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.006022035174159995, | |
| "grad_norm": 0.7687948942184448, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.0525, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0062957640457127215, | |
| "grad_norm": 0.4885317087173462, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.0714, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.006569492917265448, | |
| "grad_norm": 0.7037376761436462, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.0932, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.006843221788818176, | |
| "grad_norm": 0.9735738039016724, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.08, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.007116950660370902, | |
| "grad_norm": 0.37204796075820923, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.0417, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.00739067953192363, | |
| "grad_norm": 0.650737464427948, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.0683, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.007664408403476357, | |
| "grad_norm": 0.5781377553939819, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.0341, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.007938137275029084, | |
| "grad_norm": 0.590464174747467, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.0542, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.008211866146581811, | |
| "grad_norm": 0.4975374639034271, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.0437, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.008485595018134538, | |
| "grad_norm": 0.7869663834571838, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.18, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.008759323889687264, | |
| "grad_norm": 1.1055541038513184, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.306, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.009033052761239991, | |
| "grad_norm": 0.43376174569129944, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.0721, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.00930678163279272, | |
| "grad_norm": 0.772247850894928, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.0681, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.009580510504345446, | |
| "grad_norm": 0.5541563630104065, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.0682, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.009854239375898173, | |
| "grad_norm": 0.297048419713974, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.0344, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.0101279682474509, | |
| "grad_norm": 0.13589619100093842, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0167, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.010401697119003626, | |
| "grad_norm": 0.17115527391433716, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.0157, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.010675425990556354, | |
| "grad_norm": 0.17680104076862335, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.0225, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.010949154862109081, | |
| "grad_norm": 0.14002275466918945, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0156, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.011222883733661808, | |
| "grad_norm": 0.3309665322303772, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.047, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.011496612605214535, | |
| "grad_norm": 0.15639910101890564, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.0148, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.011770341476767261, | |
| "grad_norm": 0.17438195645809174, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.0191, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.01204407034831999, | |
| "grad_norm": 0.09357629716396332, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.0044, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.012317799219872716, | |
| "grad_norm": 0.16031214594841003, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.0091, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.012591528091425443, | |
| "grad_norm": 0.07358153164386749, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.003, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.01286525696297817, | |
| "grad_norm": 0.08584200590848923, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0039, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.013138985834530896, | |
| "grad_norm": 0.06349136680364609, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.0023, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.013412714706083625, | |
| "grad_norm": 0.0750914067029953, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0033, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.013686443577636351, | |
| "grad_norm": 0.20852030813694, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0114, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.013686443577636351, | |
| "eval_loss": 0.10318926721811295, | |
| "eval_runtime": 263.2495, | |
| "eval_samples_per_second": 23.373, | |
| "eval_steps_per_second": 5.846, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.013960172449189078, | |
| "grad_norm": 1.7688506841659546, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.4599, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.014233901320741805, | |
| "grad_norm": 0.6852141618728638, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.1715, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.014507630192294532, | |
| "grad_norm": 0.64682537317276, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.1896, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.01478135906384726, | |
| "grad_norm": 0.4129149615764618, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.1405, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.015055087935399987, | |
| "grad_norm": 0.39000850915908813, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.0792, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.015328816806952713, | |
| "grad_norm": 0.7193068265914917, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.1492, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.01560254567850544, | |
| "grad_norm": 0.2947554588317871, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.0963, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.01587627455005817, | |
| "grad_norm": 0.28656435012817383, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0775, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.016150003421610895, | |
| "grad_norm": 0.28851979970932007, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.1041, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.016423732293163622, | |
| "grad_norm": 0.5464861989021301, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.0947, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.01669746116471635, | |
| "grad_norm": 0.39874088764190674, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.1459, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.016971190036269075, | |
| "grad_norm": 0.39730069041252136, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0872, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.017244918907821802, | |
| "grad_norm": 0.3217560052871704, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.1013, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.01751864777937453, | |
| "grad_norm": 0.36995869874954224, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0794, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.017792376650927255, | |
| "grad_norm": 0.29235830903053284, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0753, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.018066105522479982, | |
| "grad_norm": 0.40605628490448, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.1249, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.018339834394032712, | |
| "grad_norm": 0.24858100712299347, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0819, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.01861356326558544, | |
| "grad_norm": 0.30691301822662354, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0559, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.018887292137138165, | |
| "grad_norm": 0.27972522377967834, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.0858, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.019161021008690892, | |
| "grad_norm": 0.28502631187438965, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.0782, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.01943474988024362, | |
| "grad_norm": 0.27517464756965637, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.1164, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.019708478751796345, | |
| "grad_norm": 0.2742948830127716, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0812, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.019982207623349072, | |
| "grad_norm": 0.2698785364627838, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.0526, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.0202559364949018, | |
| "grad_norm": 0.1924898773431778, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.0653, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.020529665366454525, | |
| "grad_norm": 0.24038903415203094, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0571, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.020803394238007252, | |
| "grad_norm": 0.2771959602832794, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0678, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.021077123109559982, | |
| "grad_norm": 0.3027803599834442, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0824, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.02135085198111271, | |
| "grad_norm": 0.34813180565834045, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0428, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.021624580852665436, | |
| "grad_norm": 0.18736878037452698, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0164, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.021898309724218162, | |
| "grad_norm": 0.1618126630783081, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.0252, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.02217203859577089, | |
| "grad_norm": 0.18529681861400604, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0256, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.022445767467323616, | |
| "grad_norm": 0.5240387916564941, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0771, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.022719496338876342, | |
| "grad_norm": 0.3223040699958801, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.1034, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.02299322521042907, | |
| "grad_norm": 0.342072993516922, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.0586, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.023266954081981796, | |
| "grad_norm": 0.2215188443660736, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0384, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.023540682953534522, | |
| "grad_norm": 0.1570575088262558, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0218, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.023814411825087253, | |
| "grad_norm": 0.19102266430854797, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0424, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.02408814069663998, | |
| "grad_norm": 0.1055581122636795, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0121, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.024361869568192706, | |
| "grad_norm": 0.3091917037963867, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.079, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.024635598439745433, | |
| "grad_norm": 0.1684788465499878, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.0372, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.02490932731129816, | |
| "grad_norm": 0.25582513213157654, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0406, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.025183056182850886, | |
| "grad_norm": 0.10428531467914581, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0058, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.025456785054403613, | |
| "grad_norm": 0.25681793689727783, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0333, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.02573051392595634, | |
| "grad_norm": 0.09947719424962997, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0055, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.026004242797509066, | |
| "grad_norm": 0.05656822770833969, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0031, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.026277971669061793, | |
| "grad_norm": 0.10895723849534988, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0061, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.026551700540614523, | |
| "grad_norm": 0.30041539669036865, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0417, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.02682542941216725, | |
| "grad_norm": 0.20105572044849396, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0128, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.027099158283719976, | |
| "grad_norm": 0.09207609295845032, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.005, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.027372887155272703, | |
| "grad_norm": 0.07330376654863358, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0041, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.027372887155272703, | |
| "eval_loss": 0.0689723864197731, | |
| "eval_runtime": 263.1747, | |
| "eval_samples_per_second": 23.38, | |
| "eval_steps_per_second": 5.848, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.02764661602682543, | |
| "grad_norm": 1.0368256568908691, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.2054, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.027920344898378156, | |
| "grad_norm": 0.5863410234451294, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.1558, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.028194073769930883, | |
| "grad_norm": 0.5034375190734863, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.1396, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.02846780264148361, | |
| "grad_norm": 0.5163214802742004, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.1619, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.028741531513036336, | |
| "grad_norm": 0.4672788977622986, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1527, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.029015260384589063, | |
| "grad_norm": 0.35526421666145325, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.1102, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.029288989256141793, | |
| "grad_norm": 0.30567991733551025, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0776, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.02956271812769452, | |
| "grad_norm": 0.4598815143108368, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.1627, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.029836446999247247, | |
| "grad_norm": 0.2589607238769531, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0687, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.030110175870799973, | |
| "grad_norm": 0.3864089846611023, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.1135, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0303839047423527, | |
| "grad_norm": 0.25004953145980835, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.1004, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.030657633613905427, | |
| "grad_norm": 0.25639253854751587, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0793, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.030931362485458153, | |
| "grad_norm": 0.36456888914108276, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0842, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.03120509135701088, | |
| "grad_norm": 0.39386072754859924, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0656, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.03147882022856361, | |
| "grad_norm": 0.2991383373737335, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0856, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.03175254910011634, | |
| "grad_norm": 0.21862885355949402, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0452, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.032026277971669063, | |
| "grad_norm": 0.21904170513153076, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.051, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.03230000684322179, | |
| "grad_norm": 0.2709364593029022, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0849, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.03257373571477452, | |
| "grad_norm": 0.27606716752052307, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0689, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.032847464586327244, | |
| "grad_norm": 0.3288068175315857, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0845, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.03312119345787997, | |
| "grad_norm": 0.2838031053543091, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0412, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0333949223294327, | |
| "grad_norm": 0.18428663909435272, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0595, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.033668651200985424, | |
| "grad_norm": 0.37800440192222595, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0672, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.03394238007253815, | |
| "grad_norm": 0.22834280133247375, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0346, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.03421610894409088, | |
| "grad_norm": 0.2601318955421448, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.1222, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.034489837815643604, | |
| "grad_norm": 0.1850673258304596, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0388, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.03476356668719633, | |
| "grad_norm": 0.3141416311264038, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0646, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.03503729555874906, | |
| "grad_norm": 0.18401959538459778, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.034, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.035311024430301784, | |
| "grad_norm": 0.15586012601852417, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0326, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.03558475330185451, | |
| "grad_norm": 0.22381485998630524, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0342, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.03585848217340724, | |
| "grad_norm": 0.3787159025669098, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0816, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.036132211044959964, | |
| "grad_norm": 0.14897434413433075, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0296, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.0364059399165127, | |
| "grad_norm": 0.2271733433008194, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0287, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.036679668788065424, | |
| "grad_norm": 0.13798674941062927, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0292, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.03695339765961815, | |
| "grad_norm": 0.16890235245227814, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0242, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.03722712653117088, | |
| "grad_norm": 0.11271921545267105, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.008, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.037500855402723604, | |
| "grad_norm": 0.11944177746772766, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0147, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.03777458427427633, | |
| "grad_norm": 0.14850112795829773, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0178, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.03804831314582906, | |
| "grad_norm": 0.15633860230445862, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0207, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.038322042017381784, | |
| "grad_norm": 0.17218001186847687, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0389, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.03859577088893451, | |
| "grad_norm": 0.08486124128103256, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0054, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.03886949976048724, | |
| "grad_norm": 0.13307073712348938, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0183, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.039143228632039964, | |
| "grad_norm": 0.12460210174322128, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.0074, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.03941695750359269, | |
| "grad_norm": 0.1307774931192398, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0074, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.03969068637514542, | |
| "grad_norm": 0.09602323919534683, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0062, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.039964415246698144, | |
| "grad_norm": 0.09614124894142151, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0177, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.04023814411825087, | |
| "grad_norm": 0.04025377333164215, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0025, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.0405118729898036, | |
| "grad_norm": 0.4157078266143799, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0212, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.040785601861356324, | |
| "grad_norm": 0.08781462162733078, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0055, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.04105933073290905, | |
| "grad_norm": 0.16040568053722382, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0173, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04105933073290905, | |
| "eval_loss": 0.053889013826847076, | |
| "eval_runtime": 262.788, | |
| "eval_samples_per_second": 23.414, | |
| "eval_steps_per_second": 5.856, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.04133305960446178, | |
| "grad_norm": 0.42668816447257996, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.1353, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.041606788476014504, | |
| "grad_norm": 0.4605340361595154, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.1332, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.04188051734756724, | |
| "grad_norm": 0.4513793885707855, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.1305, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.042154246219119965, | |
| "grad_norm": 0.4168568253517151, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.1478, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.04242797509067269, | |
| "grad_norm": 0.508091151714325, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.1776, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.04270170396222542, | |
| "grad_norm": 0.33139148354530334, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0916, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.042975432833778145, | |
| "grad_norm": 0.39131301641464233, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.1118, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.04324916170533087, | |
| "grad_norm": 0.46013709902763367, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.1722, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.0435228905768836, | |
| "grad_norm": 0.2845227122306824, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0652, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.043796619448436325, | |
| "grad_norm": 0.3384856879711151, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0876, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.04407034831998905, | |
| "grad_norm": 0.27916601300239563, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0621, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.04434407719154178, | |
| "grad_norm": 0.3159334063529968, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.065, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.044617806063094505, | |
| "grad_norm": 0.507485032081604, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.1085, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.04489153493464723, | |
| "grad_norm": 0.3902236521244049, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.1489, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.04516526380619996, | |
| "grad_norm": 0.350129097700119, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0553, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.045438992677752685, | |
| "grad_norm": 0.20551259815692902, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0344, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.04571272154930541, | |
| "grad_norm": 0.2562514841556549, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0609, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.04598645042085814, | |
| "grad_norm": 0.34810352325439453, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0453, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.046260179292410865, | |
| "grad_norm": 0.21661634743213654, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0491, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.04653390816396359, | |
| "grad_norm": 0.2689395248889923, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0456, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.04680763703551632, | |
| "grad_norm": 0.23819881677627563, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.072, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.047081365907069045, | |
| "grad_norm": 0.1767212599515915, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0364, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.04735509477862178, | |
| "grad_norm": 0.16702032089233398, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.043, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.047628823650174505, | |
| "grad_norm": 0.21438774466514587, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0423, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.04790255252172723, | |
| "grad_norm": 0.3000297248363495, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0489, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.04817628139327996, | |
| "grad_norm": 0.2725224792957306, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0834, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.048450010264832685, | |
| "grad_norm": 0.2468566596508026, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0363, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.04872373913638541, | |
| "grad_norm": 0.1893908977508545, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0391, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.04899746800793814, | |
| "grad_norm": 0.2877200245857239, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0735, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.049271196879490865, | |
| "grad_norm": 0.18520237505435944, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0423, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.04954492575104359, | |
| "grad_norm": 0.5646597743034363, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1359, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.04981865462259632, | |
| "grad_norm": 0.1474703848361969, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0177, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.050092383494149045, | |
| "grad_norm": 0.15733028948307037, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0356, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.05036611236570177, | |
| "grad_norm": 0.17028987407684326, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0353, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.0506398412372545, | |
| "grad_norm": 0.12482590973377228, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0185, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.050913570108807225, | |
| "grad_norm": 0.10851741582155228, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0127, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.05118729898035995, | |
| "grad_norm": 0.13794581592082977, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0211, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.05146102785191268, | |
| "grad_norm": 0.22493144869804382, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0414, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.051734756723465405, | |
| "grad_norm": 0.17623072862625122, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.017, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.05200848559501813, | |
| "grad_norm": 0.23774200677871704, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0199, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.05228221446657086, | |
| "grad_norm": 0.14420565962791443, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0218, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.052555943338123585, | |
| "grad_norm": 0.25365015864372253, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0411, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.05282967220967631, | |
| "grad_norm": 0.15551206469535828, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.014, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.053103401081229046, | |
| "grad_norm": 0.1488182693719864, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0111, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.05337712995278177, | |
| "grad_norm": 0.12060103565454483, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0213, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0536508588243345, | |
| "grad_norm": 0.1427120417356491, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0113, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.053924587695887226, | |
| "grad_norm": 0.23374320566654205, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.018, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.05419831656743995, | |
| "grad_norm": 0.268144816160202, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0359, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.05447204543899268, | |
| "grad_norm": 0.24813376367092133, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0154, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.054745774310545406, | |
| "grad_norm": 0.30383244156837463, | |
| "learning_rate": 0.0, | |
| "loss": 0.0217, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.054745774310545406, | |
| "eval_loss": 0.046638909727334976, | |
| "eval_runtime": 262.4816, | |
| "eval_samples_per_second": 23.442, | |
| "eval_steps_per_second": 5.863, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "EarlyStoppingCallback": { | |
| "args": { | |
| "early_stopping_patience": 5, | |
| "early_stopping_threshold": 0.0 | |
| }, | |
| "attributes": { | |
| "early_stopping_patience_counter": 0 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.3504381150298112e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |