| { | |
| "best_metric": 0.14146259427070618, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.023303233323623652, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00011651616661811826, | |
| "grad_norm": 0.9599093794822693, | |
| "learning_rate": 1e-05, | |
| "loss": 0.462, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00011651616661811826, | |
| "eval_loss": 1.1012016534805298, | |
| "eval_runtime": 1245.7683, | |
| "eval_samples_per_second": 11.603, | |
| "eval_steps_per_second": 2.901, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.00023303233323623653, | |
| "grad_norm": 1.3655271530151367, | |
| "learning_rate": 2e-05, | |
| "loss": 0.8663, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0003495484998543548, | |
| "grad_norm": 1.5218656063079834, | |
| "learning_rate": 3e-05, | |
| "loss": 1.1569, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.00046606466647247305, | |
| "grad_norm": 1.7400411367416382, | |
| "learning_rate": 4e-05, | |
| "loss": 1.1806, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0005825808330905913, | |
| "grad_norm": 2.2830474376678467, | |
| "learning_rate": 5e-05, | |
| "loss": 1.5318, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0006990969997087095, | |
| "grad_norm": 1.6540509462356567, | |
| "learning_rate": 6e-05, | |
| "loss": 1.15, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0008156131663268278, | |
| "grad_norm": 1.304975986480713, | |
| "learning_rate": 7e-05, | |
| "loss": 0.6866, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.0009321293329449461, | |
| "grad_norm": 1.2826467752456665, | |
| "learning_rate": 8e-05, | |
| "loss": 0.5037, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0010486454995630643, | |
| "grad_norm": 1.8179422616958618, | |
| "learning_rate": 9e-05, | |
| "loss": 0.6308, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0011651616661811825, | |
| "grad_norm": 1.5292041301727295, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4277, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0012816778327993008, | |
| "grad_norm": 1.4343953132629395, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.345, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.001398193999417419, | |
| "grad_norm": 1.1318269968032837, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.2948, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.0015147101660355374, | |
| "grad_norm": 3.134321451187134, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.5487, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.0016312263326536556, | |
| "grad_norm": 2.8082644939422607, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.4675, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.001747742499271774, | |
| "grad_norm": 2.0587656497955322, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.2904, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.0018642586658898922, | |
| "grad_norm": 2.1597909927368164, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.2897, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0019807748325080107, | |
| "grad_norm": 1.486000895500183, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.3515, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0020972909991261285, | |
| "grad_norm": 1.3509650230407715, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.3656, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.002213807165744247, | |
| "grad_norm": 1.1281728744506836, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.3172, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.002330323332362365, | |
| "grad_norm": 0.9763940572738647, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.3182, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0024468394989804834, | |
| "grad_norm": 0.8626099228858948, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.2651, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0025633556655986016, | |
| "grad_norm": 0.8070346713066101, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.2176, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.00267987183221672, | |
| "grad_norm": 0.9028394222259521, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.2222, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.002796387998834838, | |
| "grad_norm": 0.8324002027511597, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.2513, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0029129041654529565, | |
| "grad_norm": 0.8881924748420715, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.1499, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.0030294203320710747, | |
| "grad_norm": 1.127667784690857, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.2031, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.003145936498689193, | |
| "grad_norm": 0.712140679359436, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.108, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.0032624526653073113, | |
| "grad_norm": 0.8020001649856567, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.0924, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.0033789688319254296, | |
| "grad_norm": 0.8230465054512024, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.1676, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.003495484998543548, | |
| "grad_norm": 0.6456973552703857, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.1302, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.003612001165161666, | |
| "grad_norm": 0.47665271162986755, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.0856, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.0037285173317797844, | |
| "grad_norm": 0.6971763372421265, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.1459, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0038450334983979027, | |
| "grad_norm": 0.9844892621040344, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.2129, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.003961549665016021, | |
| "grad_norm": 0.6267403960227966, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.109, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.00407806583163414, | |
| "grad_norm": 0.8642484545707703, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.1943, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.004194581998252257, | |
| "grad_norm": 0.570564329624176, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.1107, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.004311098164870375, | |
| "grad_norm": 0.6440751552581787, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.1152, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.004427614331488494, | |
| "grad_norm": 0.5586627721786499, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.1089, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.004544130498106612, | |
| "grad_norm": 0.5665009617805481, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.128, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.00466064666472473, | |
| "grad_norm": 0.5694087147712708, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0767, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0047771628313428485, | |
| "grad_norm": 0.7201205492019653, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.1527, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.004893678997960967, | |
| "grad_norm": 0.9461737275123596, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.1251, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.005010195164579085, | |
| "grad_norm": 0.6246421933174133, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.1458, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.005126711331197203, | |
| "grad_norm": 0.6179465651512146, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.1298, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0052432274978153216, | |
| "grad_norm": 0.6541555523872375, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.1124, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.00535974366443344, | |
| "grad_norm": 0.5699607729911804, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.1506, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.005476259831051558, | |
| "grad_norm": 0.7160690426826477, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.1171, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.005592775997669676, | |
| "grad_norm": 0.9568198323249817, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.1088, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.005709292164287795, | |
| "grad_norm": 0.6204705834388733, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.1044, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.005825808330905913, | |
| "grad_norm": 0.565091073513031, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0737, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005825808330905913, | |
| "eval_loss": 0.3085353672504425, | |
| "eval_runtime": 1251.0029, | |
| "eval_samples_per_second": 11.555, | |
| "eval_steps_per_second": 2.889, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.005942324497524031, | |
| "grad_norm": 1.1472634077072144, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.3805, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.0060588406641421495, | |
| "grad_norm": 1.1541748046875, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.572, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.006175356830760268, | |
| "grad_norm": 1.2070544958114624, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8035, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.006291872997378386, | |
| "grad_norm": 1.1297117471694946, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.7949, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.006408389163996504, | |
| "grad_norm": 1.1752163171768188, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.7666, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.006524905330614623, | |
| "grad_norm": 1.1247031688690186, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.6017, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.006641421497232741, | |
| "grad_norm": 0.9158157706260681, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.2493, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.006757937663850859, | |
| "grad_norm": 0.9923712015151978, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.3598, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.006874453830468977, | |
| "grad_norm": 1.1763262748718262, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3467, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.006990969997087096, | |
| "grad_norm": 1.0084041357040405, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.1882, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.007107486163705214, | |
| "grad_norm": 0.9700591564178467, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.2153, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.007224002330323332, | |
| "grad_norm": 1.3720237016677856, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.2899, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0073405184969414505, | |
| "grad_norm": 1.7217190265655518, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.3145, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.007457034663559569, | |
| "grad_norm": 1.0699461698532104, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.1848, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.007573550830177687, | |
| "grad_norm": 1.2412887811660767, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.176, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.007690066996795805, | |
| "grad_norm": 0.9800851345062256, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.1664, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.007806583163413924, | |
| "grad_norm": 1.0461195707321167, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.266, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.007923099330032043, | |
| "grad_norm": 0.8845111131668091, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.197, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.00803961549665016, | |
| "grad_norm": 1.0253078937530518, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.2117, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.00815613166326828, | |
| "grad_norm": 0.6442228555679321, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.1293, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.008272647829886397, | |
| "grad_norm": 0.9206815958023071, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.1381, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.008389163996504514, | |
| "grad_norm": 0.7148730158805847, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.1213, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.008505680163122633, | |
| "grad_norm": 0.8817028403282166, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.1738, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.00862219632974075, | |
| "grad_norm": 0.736078679561615, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.1344, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.00873871249635887, | |
| "grad_norm": 0.795749306678772, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.1484, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.008855228662976987, | |
| "grad_norm": 0.5856538414955139, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0935, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.008971744829595106, | |
| "grad_norm": 0.5711079239845276, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0987, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.009088260996213224, | |
| "grad_norm": 0.684877336025238, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.1224, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.009204777162831343, | |
| "grad_norm": 0.6529368162155151, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.1063, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.00932129332944946, | |
| "grad_norm": 0.7334063649177551, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.1374, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00943780949606758, | |
| "grad_norm": 0.5875141024589539, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0998, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.009554325662685697, | |
| "grad_norm": 0.6797360181808472, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.1177, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.009670841829303816, | |
| "grad_norm": 0.6950648427009583, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.1503, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.009787357995921933, | |
| "grad_norm": 0.9285181164741516, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.1699, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.009903874162540053, | |
| "grad_norm": 0.55496746301651, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0838, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.01002039032915817, | |
| "grad_norm": 0.41909873485565186, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0876, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.01013690649577629, | |
| "grad_norm": 0.543818473815918, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0812, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.010253422662394407, | |
| "grad_norm": 0.4426962435245514, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.067, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.010369938829012526, | |
| "grad_norm": 0.4855525493621826, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.081, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.010486454995630643, | |
| "grad_norm": 0.46776020526885986, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.093, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.010602971162248762, | |
| "grad_norm": 0.7187413573265076, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.1916, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.01071948732886688, | |
| "grad_norm": 0.4713269770145416, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0677, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.010836003495484999, | |
| "grad_norm": 0.9408530592918396, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.1043, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.010952519662103116, | |
| "grad_norm": 0.6162998676300049, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.1069, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.011069035828721235, | |
| "grad_norm": 0.5137761235237122, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0767, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.011185551995339353, | |
| "grad_norm": 0.58188396692276, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0924, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.011302068161957472, | |
| "grad_norm": 0.699849009513855, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0969, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.01141858432857559, | |
| "grad_norm": 0.7289067506790161, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.1335, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.011535100495193708, | |
| "grad_norm": 0.48988938331604004, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0576, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.011651616661811826, | |
| "grad_norm": 0.41000425815582275, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0462, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011651616661811826, | |
| "eval_loss": 0.22562910616397858, | |
| "eval_runtime": 1249.8378, | |
| "eval_samples_per_second": 11.566, | |
| "eval_steps_per_second": 2.892, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011768132828429945, | |
| "grad_norm": 4.617305278778076, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.3589, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.011884648995048062, | |
| "grad_norm": 0.9132931232452393, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.3753, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.012001165161666182, | |
| "grad_norm": 1.0184781551361084, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.697, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.012117681328284299, | |
| "grad_norm": 1.0063551664352417, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.6997, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.012234197494902418, | |
| "grad_norm": 1.130664587020874, | |
| "learning_rate": 5e-05, | |
| "loss": 0.7572, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.012350713661520536, | |
| "grad_norm": 1.0476655960083008, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.5017, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.012467229828138655, | |
| "grad_norm": 2.3411855697631836, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.3705, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.012583745994756772, | |
| "grad_norm": 1.0491865873336792, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.1829, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.012700262161374891, | |
| "grad_norm": 1.1174777746200562, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.2456, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.012816778327993009, | |
| "grad_norm": 0.9330149292945862, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.2296, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012933294494611128, | |
| "grad_norm": 1.139375925064087, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.192, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.013049810661229245, | |
| "grad_norm": 0.7400180101394653, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.1205, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.013166326827847364, | |
| "grad_norm": 1.5136195421218872, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.2404, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.013282842994465482, | |
| "grad_norm": 1.1875526905059814, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.1462, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.013399359161083601, | |
| "grad_norm": 0.9687106609344482, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.168, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.013515875327701718, | |
| "grad_norm": 1.085702896118164, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.223, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.013632391494319837, | |
| "grad_norm": 0.8963435888290405, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.24, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.013748907660937955, | |
| "grad_norm": 1.128997564315796, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.2415, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.013865423827556074, | |
| "grad_norm": 0.8026189804077148, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.1587, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.013981939994174191, | |
| "grad_norm": 0.6406410336494446, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.1178, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.01409845616079231, | |
| "grad_norm": 0.6451287269592285, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.109, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.014214972327410428, | |
| "grad_norm": 0.5956007838249207, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0962, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.014331488494028547, | |
| "grad_norm": 0.6382842659950256, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.1277, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.014448004660646665, | |
| "grad_norm": 0.5531885623931885, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0928, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.014564520827264784, | |
| "grad_norm": 0.3945748209953308, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0775, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.014681036993882901, | |
| "grad_norm": 0.45172953605651855, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0824, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.01479755316050102, | |
| "grad_norm": 0.3784625828266144, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0659, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.014914069327119138, | |
| "grad_norm": 0.5193028450012207, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.1038, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.015030585493737257, | |
| "grad_norm": 0.46165239810943604, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0642, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.015147101660355374, | |
| "grad_norm": 0.5272737145423889, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.1047, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.015263617826973493, | |
| "grad_norm": 0.4919346570968628, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0751, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.01538013399359161, | |
| "grad_norm": 0.599928617477417, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.1229, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.01549665016020973, | |
| "grad_norm": 0.523377001285553, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0933, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.015613166326827847, | |
| "grad_norm": 0.5800530314445496, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.103, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.015729682493445966, | |
| "grad_norm": 0.5339798331260681, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0647, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.015846198660064086, | |
| "grad_norm": 0.6347007751464844, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0807, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.0159627148266822, | |
| "grad_norm": 0.6524521708488464, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.1489, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.01607923099330032, | |
| "grad_norm": 0.3986034095287323, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0672, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.01619574715991844, | |
| "grad_norm": 0.4242058992385864, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0605, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.01631226332653656, | |
| "grad_norm": 0.6749107241630554, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.1153, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.016428779493154674, | |
| "grad_norm": 0.5296944975852966, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.084, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.016545295659772793, | |
| "grad_norm": 0.706341028213501, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.111, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.016661811826390913, | |
| "grad_norm": 0.5029020309448242, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.1032, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.01677832799300903, | |
| "grad_norm": 0.5768967270851135, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0746, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.016894844159627147, | |
| "grad_norm": 0.7191116213798523, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.109, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.017011360326245267, | |
| "grad_norm": 0.7465367913246155, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.113, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.017127876492863386, | |
| "grad_norm": 0.4132862687110901, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0611, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.0172443926594815, | |
| "grad_norm": 0.6758192777633667, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0872, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.01736090882609962, | |
| "grad_norm": 0.4488687813282013, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0652, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.01747742499271774, | |
| "grad_norm": 0.49048998951911926, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0512, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01747742499271774, | |
| "eval_loss": 0.17200881242752075, | |
| "eval_runtime": 1251.5657, | |
| "eval_samples_per_second": 11.55, | |
| "eval_steps_per_second": 2.888, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.01759394115933586, | |
| "grad_norm": 0.4155811369419098, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.2053, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.017710457325953975, | |
| "grad_norm": 0.5505126118659973, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.3425, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.017826973492572094, | |
| "grad_norm": 0.7541253566741943, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.5565, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.017943489659190213, | |
| "grad_norm": 0.7976937890052795, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.7044, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.018060005825808332, | |
| "grad_norm": 0.8429288268089294, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.7087, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.018176521992426448, | |
| "grad_norm": 0.8887481093406677, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.5215, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.018293038159044567, | |
| "grad_norm": 1.2595231533050537, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.4045, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.018409554325662686, | |
| "grad_norm": 0.6970676183700562, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.2036, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.018526070492280805, | |
| "grad_norm": 0.7819052338600159, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.2534, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.01864258665889892, | |
| "grad_norm": 0.7972256541252136, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.2844, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.01875910282551704, | |
| "grad_norm": 0.7078753113746643, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.1791, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.01887561899213516, | |
| "grad_norm": 0.7534687519073486, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.1762, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.018992135158753278, | |
| "grad_norm": 0.7150564193725586, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.2073, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.019108651325371394, | |
| "grad_norm": 1.4147645235061646, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.168, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.019225167491989513, | |
| "grad_norm": 1.569960355758667, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.1553, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.019341683658607632, | |
| "grad_norm": 1.1069717407226562, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0732, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.01945819982522575, | |
| "grad_norm": 0.6492745280265808, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.1296, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.019574715991843867, | |
| "grad_norm": 0.7327039241790771, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.1956, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.019691232158461986, | |
| "grad_norm": 0.6349549293518066, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.1908, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.019807748325080105, | |
| "grad_norm": 0.5512869358062744, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0881, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.019924264491698224, | |
| "grad_norm": 0.634693443775177, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.1207, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.02004078065831634, | |
| "grad_norm": 0.6891131401062012, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.1174, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.02015729682493446, | |
| "grad_norm": 0.5423218011856079, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0924, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.02027381299155258, | |
| "grad_norm": 0.5473216772079468, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0864, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.020390329158170697, | |
| "grad_norm": 0.4381336271762848, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0743, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.020506845324788813, | |
| "grad_norm": 0.6306963562965393, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.1231, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.020623361491406932, | |
| "grad_norm": 0.6433330178260803, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.1078, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.02073987765802505, | |
| "grad_norm": 0.5808589458465576, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.085, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.02085639382464317, | |
| "grad_norm": 0.41638869047164917, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0487, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.020972909991261286, | |
| "grad_norm": 0.4117186367511749, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0514, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.021089426157879405, | |
| "grad_norm": 0.5810941457748413, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1193, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.021205942324497525, | |
| "grad_norm": 0.43184494972229004, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0809, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.021322458491115644, | |
| "grad_norm": 0.45248666405677795, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0887, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.02143897465773376, | |
| "grad_norm": 0.6352959275245667, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0827, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.02155549082435188, | |
| "grad_norm": 0.46160057187080383, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.061, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.021672006990969998, | |
| "grad_norm": 0.8240150213241577, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0859, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.021788523157588117, | |
| "grad_norm": 0.5752792954444885, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.1188, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.021905039324206232, | |
| "grad_norm": 0.87974613904953, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0948, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.02202155549082435, | |
| "grad_norm": 0.4136914014816284, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0681, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.02213807165744247, | |
| "grad_norm": 0.6594626307487488, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0764, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.02225458782406059, | |
| "grad_norm": 0.5164365768432617, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.1033, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.022371103990678706, | |
| "grad_norm": 0.36651620268821716, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0428, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.022487620157296825, | |
| "grad_norm": 0.5214178562164307, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0851, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.022604136323914944, | |
| "grad_norm": 0.2698737680912018, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0382, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.022720652490533063, | |
| "grad_norm": 0.41610655188560486, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0816, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.02283716865715118, | |
| "grad_norm": 0.5368525981903076, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0744, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.022953684823769298, | |
| "grad_norm": 0.4773271083831787, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0537, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.023070200990387417, | |
| "grad_norm": 0.3635956346988678, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0239, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.023186717157005536, | |
| "grad_norm": 0.5151886940002441, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0531, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.023303233323623652, | |
| "grad_norm": 0.7246865630149841, | |
| "learning_rate": 0.0, | |
| "loss": 0.2167, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.023303233323623652, | |
| "eval_loss": 0.14146259427070618, | |
| "eval_runtime": 1250.2631, | |
| "eval_samples_per_second": 11.562, | |
| "eval_steps_per_second": 2.891, | |
| "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": 3.428097107699958e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |