| { | |
| "best_metric": 0.35439521074295044, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.012984905047881837, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 6.492452523940918e-05, | |
| "grad_norm": 4.540472507476807, | |
| "learning_rate": 1e-05, | |
| "loss": 2.1793, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 6.492452523940918e-05, | |
| "eval_loss": 0.558050274848938, | |
| "eval_runtime": 2718.0417, | |
| "eval_samples_per_second": 9.544, | |
| "eval_steps_per_second": 2.386, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00012984905047881836, | |
| "grad_norm": 6.050493240356445, | |
| "learning_rate": 2e-05, | |
| "loss": 2.7294, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.00019477357571822757, | |
| "grad_norm": 4.454987525939941, | |
| "learning_rate": 3e-05, | |
| "loss": 2.1744, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0002596981009576367, | |
| "grad_norm": 3.7849512100219727, | |
| "learning_rate": 4e-05, | |
| "loss": 2.436, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.00032462262619704593, | |
| "grad_norm": 4.17337703704834, | |
| "learning_rate": 5e-05, | |
| "loss": 1.824, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.00038954715143645513, | |
| "grad_norm": 3.64642071723938, | |
| "learning_rate": 6e-05, | |
| "loss": 2.1666, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0004544716766758643, | |
| "grad_norm": 3.3155298233032227, | |
| "learning_rate": 7e-05, | |
| "loss": 2.1527, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0005193962019152734, | |
| "grad_norm": 3.41972279548645, | |
| "learning_rate": 8e-05, | |
| "loss": 2.037, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0005843207271546826, | |
| "grad_norm": 3.2290754318237305, | |
| "learning_rate": 9e-05, | |
| "loss": 1.7392, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0006492452523940919, | |
| "grad_norm": 2.8902621269226074, | |
| "learning_rate": 0.0001, | |
| "loss": 1.9498, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0007141697776335011, | |
| "grad_norm": 3.157255172729492, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.6533, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0007790943028729103, | |
| "grad_norm": 2.9064061641693115, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.8565, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0008440188281123195, | |
| "grad_norm": 2.6529300212860107, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.5706, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0009089433533517286, | |
| "grad_norm": 3.223670482635498, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.6391, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.0009738678785911378, | |
| "grad_norm": 2.7879912853240967, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.4823, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0010387924038305469, | |
| "grad_norm": 2.6725378036499023, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.4706, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0011037169290699562, | |
| "grad_norm": 2.8224666118621826, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.8116, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0011686414543093653, | |
| "grad_norm": 3.0412087440490723, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.5247, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.0012335659795487746, | |
| "grad_norm": 2.8261916637420654, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.4624, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.0012984905047881837, | |
| "grad_norm": 2.813544988632202, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.5645, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.001363415030027593, | |
| "grad_norm": 2.84627103805542, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.7431, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0014283395552670021, | |
| "grad_norm": 2.620612144470215, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 1.5353, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0014932640805064112, | |
| "grad_norm": 2.7557530403137207, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.5994, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.0015581886057458205, | |
| "grad_norm": 2.5682120323181152, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.4973, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0016231131309852296, | |
| "grad_norm": 2.783432960510254, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.652, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.001688037656224639, | |
| "grad_norm": 2.9067819118499756, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.4789, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.001752962181464048, | |
| "grad_norm": 2.970757484436035, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 1.5241, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0018178867067034571, | |
| "grad_norm": 2.8803842067718506, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.6238, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0018828112319428665, | |
| "grad_norm": 2.8296127319335938, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.5477, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.0019477357571822756, | |
| "grad_norm": 2.8743557929992676, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.5855, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.002012660282421685, | |
| "grad_norm": 3.8598484992980957, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.4768, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0020775848076610938, | |
| "grad_norm": 4.713948726654053, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.458, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.002142509332900503, | |
| "grad_norm": 2.804553270339966, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.3238, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.0022074338581399124, | |
| "grad_norm": 3.3731460571289062, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.7495, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.0022723583833793217, | |
| "grad_norm": 3.1557204723358154, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.4697, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0023372829086187306, | |
| "grad_norm": 3.188981056213379, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.4911, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.00240220743385814, | |
| "grad_norm": 3.202531337738037, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.3946, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.0024671319590975492, | |
| "grad_norm": 3.1800291538238525, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.5396, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.002532056484336958, | |
| "grad_norm": 3.344865322113037, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 1.5526, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0025969810095763674, | |
| "grad_norm": 3.3695080280303955, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.577, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0026619055348157767, | |
| "grad_norm": 3.3032450675964355, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.4475, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.002726830060055186, | |
| "grad_norm": 3.5895912647247314, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.5896, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.002791754585294595, | |
| "grad_norm": 3.4149351119995117, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 1.3504, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0028566791105340042, | |
| "grad_norm": 3.534778118133545, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.3705, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0029216036357734136, | |
| "grad_norm": 3.6278538703918457, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.6635, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0029865281610128224, | |
| "grad_norm": 3.4827487468719482, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.4033, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.0030514526862522318, | |
| "grad_norm": 3.8299813270568848, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.5587, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.003116377211491641, | |
| "grad_norm": 4.618851184844971, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.4306, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.00318130173673105, | |
| "grad_norm": 4.444945812225342, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.5862, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.0032462262619704593, | |
| "grad_norm": 6.026591777801514, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.8037, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0032462262619704593, | |
| "eval_loss": 0.42139238119125366, | |
| "eval_runtime": 2729.0429, | |
| "eval_samples_per_second": 9.506, | |
| "eval_steps_per_second": 2.377, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0033111507872098686, | |
| "grad_norm": 4.079866409301758, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 2.0916, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.003376075312449278, | |
| "grad_norm": 3.098527669906616, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 2.2287, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.003440999837688687, | |
| "grad_norm": 2.1626501083374023, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 1.7575, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.003505924362928096, | |
| "grad_norm": 1.966941475868225, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 1.8485, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.0035708488881675054, | |
| "grad_norm": 3.7294154167175293, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 1.728, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0036357734134069143, | |
| "grad_norm": 2.359339952468872, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.7742, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0037006979386463236, | |
| "grad_norm": 2.1914119720458984, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.5247, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.003765622463885733, | |
| "grad_norm": 1.9947062730789185, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.5345, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.003830546989125142, | |
| "grad_norm": 2.044597864151001, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.6942, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.003895471514364551, | |
| "grad_norm": 2.267122507095337, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.7808, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0039603960396039604, | |
| "grad_norm": 2.0266075134277344, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.6411, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.00402532056484337, | |
| "grad_norm": 2.1180357933044434, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 1.7271, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.004090245090082779, | |
| "grad_norm": 2.1186745166778564, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 1.4394, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.0041551696153221875, | |
| "grad_norm": 2.276301145553589, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.5767, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.004220094140561597, | |
| "grad_norm": 2.281355857849121, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 1.8338, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.004285018665801006, | |
| "grad_norm": 2.0677578449249268, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 1.2292, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.0043499431910404155, | |
| "grad_norm": 2.4492828845977783, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.5271, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.004414867716279825, | |
| "grad_norm": 2.246931314468384, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.4539, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.004479792241519234, | |
| "grad_norm": 2.3499512672424316, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 1.6166, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.004544716766758643, | |
| "grad_norm": 2.2578580379486084, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.3928, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.004609641291998052, | |
| "grad_norm": 2.352757692337036, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 1.5195, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.004674565817237461, | |
| "grad_norm": 2.227308988571167, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.4991, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0047394903424768705, | |
| "grad_norm": 2.5206069946289062, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.6782, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.00480441486771628, | |
| "grad_norm": 2.4039251804351807, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 1.357, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.004869339392955689, | |
| "grad_norm": 2.5262157917022705, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.4331, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.0049342639181950984, | |
| "grad_norm": 2.4065189361572266, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.5405, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.004999188443434508, | |
| "grad_norm": 2.6150152683258057, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 1.5466, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.005064112968673916, | |
| "grad_norm": 2.3654088973999023, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 1.3768, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.0051290374939133255, | |
| "grad_norm": 2.36637544631958, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.5675, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.005193962019152735, | |
| "grad_norm": 2.498779535293579, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.5529, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.005258886544392144, | |
| "grad_norm": 2.4871792793273926, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.5426, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.0053238110696315535, | |
| "grad_norm": 2.462024211883545, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.2875, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.005388735594870963, | |
| "grad_norm": 2.650010108947754, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.4617, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.005453660120110372, | |
| "grad_norm": 2.7433016300201416, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.4802, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.0055185846453497805, | |
| "grad_norm": 2.771566867828369, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 1.5809, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.00558350917058919, | |
| "grad_norm": 2.490809440612793, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.2319, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.005648433695828599, | |
| "grad_norm": 2.7578866481781006, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 1.4479, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.0057133582210680085, | |
| "grad_norm": 2.8412978649139404, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.4386, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.005778282746307418, | |
| "grad_norm": 2.8536527156829834, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 1.3228, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.005843207271546827, | |
| "grad_norm": 2.8993990421295166, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.453, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.005908131796786236, | |
| "grad_norm": 3.1611733436584473, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 1.5174, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.005973056322025645, | |
| "grad_norm": 2.999743700027466, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.3884, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.006037980847265054, | |
| "grad_norm": 3.1500251293182373, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.3779, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.0061029053725044635, | |
| "grad_norm": 3.3458006381988525, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.5905, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.006167829897743873, | |
| "grad_norm": 3.151742935180664, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.3328, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.006232754422983282, | |
| "grad_norm": 3.4480724334716797, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.4021, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.0062976789482226915, | |
| "grad_norm": 3.1579878330230713, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.2967, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.0063626034734621, | |
| "grad_norm": 3.8392672538757324, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.42, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.006427527998701509, | |
| "grad_norm": 3.4535746574401855, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.1299, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.0064924525239409185, | |
| "grad_norm": 5.015627861022949, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.5899, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0064924525239409185, | |
| "eval_loss": 0.39287006855010986, | |
| "eval_runtime": 2725.3783, | |
| "eval_samples_per_second": 9.518, | |
| "eval_steps_per_second": 2.38, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.006557377049180328, | |
| "grad_norm": 3.354386329650879, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 2.0273, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.006622301574419737, | |
| "grad_norm": 2.5337774753570557, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.5614, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.0066872260996591465, | |
| "grad_norm": 2.057063341140747, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 1.8339, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.006752150624898556, | |
| "grad_norm": 1.8221439123153687, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 1.55, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.006817075150137964, | |
| "grad_norm": 1.7888840436935425, | |
| "learning_rate": 5e-05, | |
| "loss": 1.8628, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.006881999675377374, | |
| "grad_norm": 1.8267186880111694, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.5607, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.006946924200616783, | |
| "grad_norm": 1.9880849123001099, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.6143, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.007011848725856192, | |
| "grad_norm": 1.8925626277923584, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 1.5868, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.0070767732510956015, | |
| "grad_norm": 2.010892152786255, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 1.6561, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.007141697776335011, | |
| "grad_norm": 1.9386076927185059, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 1.2943, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.00720662230157442, | |
| "grad_norm": 1.9109869003295898, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.4862, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.007271546826813829, | |
| "grad_norm": 2.005671501159668, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.534, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.007336471352053238, | |
| "grad_norm": 2.043527603149414, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 1.7753, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.007401395877292647, | |
| "grad_norm": 2.0597848892211914, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 1.6781, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.0074663204025320565, | |
| "grad_norm": 2.056081771850586, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.6857, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.007531244927771466, | |
| "grad_norm": 1.9409863948822021, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 1.3481, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.007596169453010875, | |
| "grad_norm": 2.038977861404419, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 1.3336, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.007661093978250284, | |
| "grad_norm": 1.9617688655853271, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 1.3789, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.007726018503489693, | |
| "grad_norm": 2.061816692352295, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.4228, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.007790943028729102, | |
| "grad_norm": 2.1589889526367188, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.3424, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.007855867553968512, | |
| "grad_norm": 2.100494384765625, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.4676, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.007920792079207921, | |
| "grad_norm": 2.2070977687835693, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 1.6314, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.00798571660444733, | |
| "grad_norm": 2.2676262855529785, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.45, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.00805064112968674, | |
| "grad_norm": 2.1959469318389893, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.3145, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.008115565654926148, | |
| "grad_norm": 2.2538602352142334, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.4348, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.008180490180165558, | |
| "grad_norm": 2.3079259395599365, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.364, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.008245414705404967, | |
| "grad_norm": 2.487854242324829, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 1.5644, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.008310339230644375, | |
| "grad_norm": 2.407414674758911, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.3611, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.008375263755883785, | |
| "grad_norm": 2.470609426498413, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.4366, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.008440188281123194, | |
| "grad_norm": 2.462022304534912, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 1.2699, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.008505112806362604, | |
| "grad_norm": 2.4126827716827393, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.3328, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.008570037331602012, | |
| "grad_norm": 2.437617063522339, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.3357, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.008634961856841422, | |
| "grad_norm": 2.4791176319122314, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 1.4554, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.008699886382080831, | |
| "grad_norm": 2.4162757396698, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.371, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.00876481090732024, | |
| "grad_norm": 2.373032569885254, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 1.1934, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.00882973543255965, | |
| "grad_norm": 2.6824629306793213, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.5817, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.008894659957799058, | |
| "grad_norm": 2.456705093383789, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 1.1805, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.008959584483038468, | |
| "grad_norm": 2.6049578189849854, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.2282, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.009024509008277877, | |
| "grad_norm": 2.7366414070129395, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 1.2222, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.009089433533517287, | |
| "grad_norm": 2.865466594696045, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.3415, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.009154358058756695, | |
| "grad_norm": 3.0013866424560547, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.3793, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.009219282583996104, | |
| "grad_norm": 2.955798625946045, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.4376, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.009284207109235514, | |
| "grad_norm": 2.8766069412231445, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 1.3984, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.009349131634474922, | |
| "grad_norm": 2.926100254058838, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.3543, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.009414056159714333, | |
| "grad_norm": 2.953508138656616, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.3205, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.009478980684953741, | |
| "grad_norm": 3.3534350395202637, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.3131, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.009543905210193151, | |
| "grad_norm": 3.56013560295105, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.5387, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.00960882973543256, | |
| "grad_norm": 3.473095655441284, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.313, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.009673754260671968, | |
| "grad_norm": 3.595202922821045, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.2718, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.009738678785911378, | |
| "grad_norm": 4.7592339515686035, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.5611, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.009738678785911378, | |
| "eval_loss": 0.3620188534259796, | |
| "eval_runtime": 2727.0957, | |
| "eval_samples_per_second": 9.512, | |
| "eval_steps_per_second": 2.378, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.009803603311150787, | |
| "grad_norm": 1.852523922920227, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 1.42, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.009868527836390197, | |
| "grad_norm": 1.725640892982483, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 1.3733, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.009933452361629605, | |
| "grad_norm": 1.6250057220458984, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 1.3382, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.009998376886869016, | |
| "grad_norm": 1.7621614933013916, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 1.6195, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.010063301412108424, | |
| "grad_norm": 1.7789777517318726, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 1.4922, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.010128225937347832, | |
| "grad_norm": 1.781243920326233, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 1.6918, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.010193150462587243, | |
| "grad_norm": 1.7903107404708862, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 1.6324, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.010258074987826651, | |
| "grad_norm": 1.7299500703811646, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 1.4214, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.010322999513066061, | |
| "grad_norm": 1.8235970735549927, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 1.4777, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.01038792403830547, | |
| "grad_norm": 1.8660005331039429, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 1.7239, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01045284856354488, | |
| "grad_norm": 1.848463773727417, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.5992, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.010517773088784288, | |
| "grad_norm": 1.7882180213928223, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 1.48, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.010582697614023697, | |
| "grad_norm": 1.8663733005523682, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 1.4401, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.010647622139263107, | |
| "grad_norm": 1.9330668449401855, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 1.5743, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.010712546664502515, | |
| "grad_norm": 1.853959083557129, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.2428, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.010777471189741926, | |
| "grad_norm": 1.909489393234253, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.2757, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.010842395714981334, | |
| "grad_norm": 2.0268399715423584, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 1.5055, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.010907320240220744, | |
| "grad_norm": 1.9924023151397705, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.3611, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.010972244765460153, | |
| "grad_norm": 2.1283366680145264, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.6103, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.011037169290699561, | |
| "grad_norm": 2.1066362857818604, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 1.47, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.011102093815938971, | |
| "grad_norm": 2.1233506202697754, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.478, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.01116701834117838, | |
| "grad_norm": 2.2296395301818848, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 1.3507, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.01123194286641779, | |
| "grad_norm": 2.076498031616211, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.3114, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.011296867391657198, | |
| "grad_norm": 2.3783020973205566, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.5188, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.011361791916896609, | |
| "grad_norm": 2.3257882595062256, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 1.5652, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.011426716442136017, | |
| "grad_norm": 2.2322728633880615, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.3695, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.011491640967375425, | |
| "grad_norm": 2.375216484069824, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.4681, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.011556565492614836, | |
| "grad_norm": 2.3548824787139893, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.453, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.011621490017854244, | |
| "grad_norm": 2.362833261489868, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.3901, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.011686414543093654, | |
| "grad_norm": 2.3238675594329834, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 1.2833, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.011751339068333063, | |
| "grad_norm": 2.2882113456726074, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.3704, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.011816263593572471, | |
| "grad_norm": 2.5618388652801514, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 1.4673, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.011881188118811881, | |
| "grad_norm": 2.5349512100219727, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.3702, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.01194611264405129, | |
| "grad_norm": 2.7572004795074463, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 1.3737, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.0120110371692907, | |
| "grad_norm": 2.699723720550537, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.4351, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.012075961694530108, | |
| "grad_norm": 2.928107738494873, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.446, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.012140886219769519, | |
| "grad_norm": 2.8252618312835693, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 1.4771, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.012205810745008927, | |
| "grad_norm": 2.6466662883758545, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 1.2121, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.012270735270248335, | |
| "grad_norm": 2.8890042304992676, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 1.3702, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.012335659795487746, | |
| "grad_norm": 2.6212081909179688, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 1.2293, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.012400584320727154, | |
| "grad_norm": 3.002800464630127, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.4438, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.012465508845966564, | |
| "grad_norm": 2.90425968170166, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.4433, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.012530433371205973, | |
| "grad_norm": 3.0153472423553467, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.1824, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.012595357896445383, | |
| "grad_norm": 2.9073667526245117, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 1.3127, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.012660282421684791, | |
| "grad_norm": 3.1661148071289062, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.3205, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.0127252069469242, | |
| "grad_norm": 3.3867547512054443, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.4784, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.01279013147216361, | |
| "grad_norm": 3.4859659671783447, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.2741, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.012855055997403018, | |
| "grad_norm": 3.688143014907837, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.4224, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.012919980522642429, | |
| "grad_norm": 4.207131862640381, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.2046, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.012984905047881837, | |
| "grad_norm": 4.53595495223999, | |
| "learning_rate": 0.0, | |
| "loss": 1.4319, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.012984905047881837, | |
| "eval_loss": 0.35439521074295044, | |
| "eval_runtime": 2723.7858, | |
| "eval_samples_per_second": 9.524, | |
| "eval_steps_per_second": 2.381, | |
| "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": 3.0516032541661594e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |