| { | |
| "best_metric": 1.029740810394287, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.20539152759948653, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010269576379974327, | |
| "grad_norm": 5.585504531860352, | |
| "learning_rate": 1e-05, | |
| "loss": 3.8371, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0010269576379974327, | |
| "eval_loss": 3.9830703735351562, | |
| "eval_runtime": 124.0152, | |
| "eval_samples_per_second": 13.224, | |
| "eval_steps_per_second": 3.306, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0020539152759948653, | |
| "grad_norm": 5.609231948852539, | |
| "learning_rate": 2e-05, | |
| "loss": 3.7983, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0030808729139922978, | |
| "grad_norm": 5.031350612640381, | |
| "learning_rate": 3e-05, | |
| "loss": 3.2987, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.004107830551989731, | |
| "grad_norm": 5.907226085662842, | |
| "learning_rate": 4e-05, | |
| "loss": 3.7771, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.005134788189987163, | |
| "grad_norm": 5.4471893310546875, | |
| "learning_rate": 5e-05, | |
| "loss": 3.1447, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0061617458279845955, | |
| "grad_norm": 5.761320114135742, | |
| "learning_rate": 6e-05, | |
| "loss": 2.6588, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.007188703465982028, | |
| "grad_norm": 6.2230448722839355, | |
| "learning_rate": 7e-05, | |
| "loss": 2.3332, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.008215661103979461, | |
| "grad_norm": 4.650792598724365, | |
| "learning_rate": 8e-05, | |
| "loss": 1.8977, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.009242618741976894, | |
| "grad_norm": 3.659452199935913, | |
| "learning_rate": 9e-05, | |
| "loss": 1.5964, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.010269576379974325, | |
| "grad_norm": 3.362877130508423, | |
| "learning_rate": 0.0001, | |
| "loss": 1.2622, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.011296534017971758, | |
| "grad_norm": 2.6512365341186523, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 1.3291, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.012323491655969191, | |
| "grad_norm": 2.1288018226623535, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 1.0522, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.013350449293966624, | |
| "grad_norm": 2.0550765991210938, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 1.2976, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.014377406931964057, | |
| "grad_norm": 2.28726863861084, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 1.3289, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.01540436456996149, | |
| "grad_norm": 2.0314509868621826, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 1.3122, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.016431322207958923, | |
| "grad_norm": 1.7815243005752563, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 1.1654, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.017458279845956354, | |
| "grad_norm": 1.9978846311569214, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 1.2949, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.01848523748395379, | |
| "grad_norm": 2.303403615951538, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 1.4053, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.01951219512195122, | |
| "grad_norm": 1.6104648113250732, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 1.0159, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02053915275994865, | |
| "grad_norm": 1.691141963005066, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 1.1197, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.021566110397946085, | |
| "grad_norm": 1.792106032371521, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 1.1337, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.022593068035943516, | |
| "grad_norm": 1.6113625764846802, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.9475, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.02362002567394095, | |
| "grad_norm": 1.8670551776885986, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 1.1604, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.024646983311938382, | |
| "grad_norm": 1.916966199874878, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 1.3558, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.025673940949935817, | |
| "grad_norm": 1.7467409372329712, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 1.2813, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.026700898587933248, | |
| "grad_norm": 1.686872124671936, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 1.2217, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.02772785622593068, | |
| "grad_norm": 1.3099868297576904, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.8811, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.028754813863928114, | |
| "grad_norm": 1.6343042850494385, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 1.1632, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.029781771501925545, | |
| "grad_norm": 1.7230594158172607, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 1.2372, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03080872913992298, | |
| "grad_norm": 1.6117074489593506, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 1.2757, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03183568677792041, | |
| "grad_norm": 1.5568138360977173, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 1.0932, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.032862644415917845, | |
| "grad_norm": 1.5986229181289673, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 1.2666, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.03388960205391527, | |
| "grad_norm": 1.5515797138214111, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 1.2019, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.03491655969191271, | |
| "grad_norm": 1.5902036428451538, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 1.1958, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.03594351732991014, | |
| "grad_norm": 1.4724791049957275, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 1.2084, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.03697047496790758, | |
| "grad_norm": 1.457838773727417, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 1.0825, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.037997432605905004, | |
| "grad_norm": 1.4028418064117432, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 1.0587, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.03902439024390244, | |
| "grad_norm": 1.6538407802581787, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 1.3942, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.040051347881899874, | |
| "grad_norm": 1.5488269329071045, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.9359, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.0410783055198973, | |
| "grad_norm": 1.9670554399490356, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 1.2586, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.042105263157894736, | |
| "grad_norm": 1.5835074186325073, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 1.2913, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.04313222079589217, | |
| "grad_norm": 1.5879188776016235, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 1.3698, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.044159178433889605, | |
| "grad_norm": 1.410422444343567, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.9896, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.04518613607188703, | |
| "grad_norm": 1.7636178731918335, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 1.4927, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.04621309370988447, | |
| "grad_norm": 1.4318428039550781, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 1.1199, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0472400513478819, | |
| "grad_norm": 1.4867923259735107, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 1.3058, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.04826700898587933, | |
| "grad_norm": 1.2988286018371582, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 1.2124, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.049293966623876764, | |
| "grad_norm": 1.582822561264038, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.5932, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.0503209242618742, | |
| "grad_norm": 1.5368988513946533, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 1.2478, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.051347881899871634, | |
| "grad_norm": 2.2753803730010986, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 1.7581, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.051347881899871634, | |
| "eval_loss": 1.1154820919036865, | |
| "eval_runtime": 125.2539, | |
| "eval_samples_per_second": 13.093, | |
| "eval_steps_per_second": 3.273, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05237483953786906, | |
| "grad_norm": 1.506157636642456, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 1.114, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.053401797175866496, | |
| "grad_norm": 1.5577417612075806, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 1.2732, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.05442875481386393, | |
| "grad_norm": 1.572106957435608, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.9267, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.05545571245186136, | |
| "grad_norm": 1.1595114469528198, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.7669, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.05648267008985879, | |
| "grad_norm": 1.6699531078338623, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.9751, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.05750962772785623, | |
| "grad_norm": 1.687030553817749, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 1.1704, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.05853658536585366, | |
| "grad_norm": 1.3938051462173462, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 1.0738, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.05956354300385109, | |
| "grad_norm": 1.6238305568695068, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 1.0057, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.060590500641848524, | |
| "grad_norm": 1.4160147905349731, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 1.1402, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.06161745827984596, | |
| "grad_norm": 1.625758409500122, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 1.0202, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.0626444159178434, | |
| "grad_norm": 1.481742262840271, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 1.252, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.06367137355584082, | |
| "grad_norm": 1.5219374895095825, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.9907, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.06469833119383825, | |
| "grad_norm": 1.3596141338348389, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.9913, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.06572528883183569, | |
| "grad_norm": 1.3875155448913574, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 1.1392, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.06675224646983312, | |
| "grad_norm": 1.3491538763046265, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.9947, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.06777920410783055, | |
| "grad_norm": 1.1520646810531616, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.9159, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.06880616174582799, | |
| "grad_norm": 1.5157663822174072, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 1.0966, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.06983311938382541, | |
| "grad_norm": 1.7143036127090454, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 1.2112, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.07086007702182286, | |
| "grad_norm": 1.3679728507995605, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.8952, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.07188703465982028, | |
| "grad_norm": 1.4405745267868042, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 1.2661, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07291399229781771, | |
| "grad_norm": 1.1053047180175781, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.9232, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.07394094993581515, | |
| "grad_norm": 1.2901381254196167, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 1.0682, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.07496790757381258, | |
| "grad_norm": 1.3525822162628174, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 1.1854, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.07599486521181001, | |
| "grad_norm": 1.3107670545578003, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.8571, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.07702182284980745, | |
| "grad_norm": 1.4273250102996826, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 1.3406, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.07804878048780488, | |
| "grad_norm": 1.384560227394104, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 1.1467, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.0790757381258023, | |
| "grad_norm": 1.2310740947723389, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.9729, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.08010269576379975, | |
| "grad_norm": 1.2470556497573853, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.9723, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.08112965340179717, | |
| "grad_norm": 1.4023878574371338, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 1.0097, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.0821566110397946, | |
| "grad_norm": 1.3913087844848633, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 1.4104, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08318356867779204, | |
| "grad_norm": 1.4844576120376587, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 1.357, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.08421052631578947, | |
| "grad_norm": 1.5256000757217407, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 1.14, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.0852374839537869, | |
| "grad_norm": 1.420501708984375, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 1.082, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.08626444159178434, | |
| "grad_norm": 1.3145442008972168, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 1.2488, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.08729139922978177, | |
| "grad_norm": 1.3933221101760864, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.9962, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.08831835686777921, | |
| "grad_norm": 1.4686084985733032, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 1.2621, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.08934531450577664, | |
| "grad_norm": 1.2372997999191284, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.9163, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.09037227214377407, | |
| "grad_norm": 1.3515774011611938, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 1.098, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.09139922978177151, | |
| "grad_norm": 1.1676000356674194, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.9619, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.09242618741976893, | |
| "grad_norm": 1.5324835777282715, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 1.2791, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09345314505776636, | |
| "grad_norm": 1.2937912940979004, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.9691, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.0944801026957638, | |
| "grad_norm": 1.4573291540145874, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 1.2536, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.09550706033376123, | |
| "grad_norm": 1.2653285264968872, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 1.1946, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.09653401797175866, | |
| "grad_norm": 1.6669771671295166, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 1.0381, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.0975609756097561, | |
| "grad_norm": 1.4045584201812744, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 1.0461, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.09858793324775353, | |
| "grad_norm": 1.566693902015686, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 1.321, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.09961489088575096, | |
| "grad_norm": 1.5425057411193848, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 1.3089, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.1006418485237484, | |
| "grad_norm": 1.3378514051437378, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 1.1017, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.10166880616174583, | |
| "grad_norm": 2.1143946647644043, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 1.4048, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.10269576379974327, | |
| "grad_norm": 1.6918808221817017, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 1.5385, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.10269576379974327, | |
| "eval_loss": 1.0648125410079956, | |
| "eval_runtime": 125.2686, | |
| "eval_samples_per_second": 13.092, | |
| "eval_steps_per_second": 3.273, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1037227214377407, | |
| "grad_norm": 1.3108991384506226, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.9371, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.10474967907573812, | |
| "grad_norm": 1.371168851852417, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 1.0909, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.10577663671373556, | |
| "grad_norm": 1.2648215293884277, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.9844, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.10680359435173299, | |
| "grad_norm": 1.1569675207138062, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.8607, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.10783055198973042, | |
| "grad_norm": 1.4329310655593872, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0482, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.10885750962772786, | |
| "grad_norm": 1.3667007684707642, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 1.2122, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.10988446726572529, | |
| "grad_norm": 1.2392832040786743, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 1.0224, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.11091142490372272, | |
| "grad_norm": 1.276443600654602, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.943, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.11193838254172016, | |
| "grad_norm": 1.2456995248794556, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.8238, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.11296534017971759, | |
| "grad_norm": 1.1732423305511475, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.7917, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.11399229781771501, | |
| "grad_norm": 1.390889048576355, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 1.1468, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.11501925545571245, | |
| "grad_norm": 1.3246679306030273, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 1.0312, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.11604621309370988, | |
| "grad_norm": 1.3436546325683594, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.9952, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.11707317073170732, | |
| "grad_norm": 1.3831197023391724, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.9919, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.11810012836970475, | |
| "grad_norm": 1.3261679410934448, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 1.0498, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.11912708600770218, | |
| "grad_norm": 1.2425459623336792, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.8745, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.12015404364569962, | |
| "grad_norm": 1.4099633693695068, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.9697, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.12118100128369705, | |
| "grad_norm": 1.1804779767990112, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.7732, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.12220795892169448, | |
| "grad_norm": 1.2373385429382324, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 1.0115, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.12323491655969192, | |
| "grad_norm": 1.4665852785110474, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 1.0504, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.12426187419768935, | |
| "grad_norm": 1.3025273084640503, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 1.0018, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.1252888318356868, | |
| "grad_norm": 1.5236035585403442, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.9602, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.12631578947368421, | |
| "grad_norm": 1.4784255027770996, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 1.0923, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.12734274711168164, | |
| "grad_norm": 1.3002736568450928, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 1.1067, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.12836970474967907, | |
| "grad_norm": 1.1813057661056519, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 1.0069, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1293966623876765, | |
| "grad_norm": 1.4547677040100098, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 1.34, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.13042362002567395, | |
| "grad_norm": 1.2443901300430298, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.99, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.13145057766367138, | |
| "grad_norm": 1.306671380996704, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 1.0293, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.1324775353016688, | |
| "grad_norm": 1.114880084991455, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 1.0155, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.13350449293966624, | |
| "grad_norm": 1.3259719610214233, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.901, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.13453145057766366, | |
| "grad_norm": 1.3376728296279907, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 1.2342, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1355584082156611, | |
| "grad_norm": 1.4119170904159546, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 1.1953, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.13658536585365855, | |
| "grad_norm": 1.2630945444107056, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.9099, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.13761232349165597, | |
| "grad_norm": 1.5290549993515015, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 1.0176, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.1386392811296534, | |
| "grad_norm": 1.2973805665969849, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.8162, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.13966623876765083, | |
| "grad_norm": 1.369152307510376, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 1.1737, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.14069319640564826, | |
| "grad_norm": 1.1779505014419556, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.9875, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1417201540436457, | |
| "grad_norm": 1.4312304258346558, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 1.1641, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.14274711168164314, | |
| "grad_norm": 1.2555488348007202, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.9312, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.14377406931964057, | |
| "grad_norm": 1.1777236461639404, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 1.0722, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.144801026957638, | |
| "grad_norm": 1.284010410308838, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 1.1465, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.14582798459563542, | |
| "grad_norm": 1.186956524848938, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 1.1312, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.14685494223363285, | |
| "grad_norm": 4.312047958374023, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.9123, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1478818998716303, | |
| "grad_norm": 1.267554759979248, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 1.1334, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.14890885750962773, | |
| "grad_norm": 1.4156426191329956, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 1.2925, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.14993581514762516, | |
| "grad_norm": 1.2493308782577515, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 1.0263, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.1509627727856226, | |
| "grad_norm": 1.315033197402954, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 1.0537, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.15198973042362002, | |
| "grad_norm": 1.4786423444747925, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 1.3493, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.15301668806161745, | |
| "grad_norm": 1.4836056232452393, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 1.4355, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.1540436456996149, | |
| "grad_norm": 1.5168051719665527, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 1.4153, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1540436456996149, | |
| "eval_loss": 1.038415551185608, | |
| "eval_runtime": 125.2406, | |
| "eval_samples_per_second": 13.095, | |
| "eval_steps_per_second": 3.274, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.15507060333761233, | |
| "grad_norm": 1.1608384847640991, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.8995, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.15609756097560976, | |
| "grad_norm": 1.2924269437789917, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.9871, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.15712451861360718, | |
| "grad_norm": 1.3043228387832642, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.733, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.1581514762516046, | |
| "grad_norm": 1.1567844152450562, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.8186, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.15917843388960207, | |
| "grad_norm": 1.1795425415039062, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.8939, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.1602053915275995, | |
| "grad_norm": 1.1569355726242065, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.7749, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.16123234916559692, | |
| "grad_norm": 1.0397932529449463, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.8432, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.16225930680359435, | |
| "grad_norm": 1.162103533744812, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.8788, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.16328626444159178, | |
| "grad_norm": 1.310011386871338, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.7517, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.1643132220795892, | |
| "grad_norm": 1.0672956705093384, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.6905, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.16534017971758666, | |
| "grad_norm": 1.3960449695587158, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 1.0843, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.1663671373555841, | |
| "grad_norm": 1.1562626361846924, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.9394, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.16739409499358152, | |
| "grad_norm": 2.4240455627441406, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.9865, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.16842105263157894, | |
| "grad_norm": 1.2105358839035034, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.9786, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.16944801026957637, | |
| "grad_norm": 1.3006137609481812, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 1.0743, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.1704749679075738, | |
| "grad_norm": 1.2988253831863403, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 1.2425, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.17150192554557125, | |
| "grad_norm": 1.1558717489242554, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.824, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.17252888318356868, | |
| "grad_norm": 1.229203462600708, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 1.0019, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.1735558408215661, | |
| "grad_norm": 1.1899900436401367, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 1.0065, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.17458279845956354, | |
| "grad_norm": 1.2884047031402588, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.9634, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.17560975609756097, | |
| "grad_norm": 1.4624727964401245, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 1.1242, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.17663671373555842, | |
| "grad_norm": 1.1112769842147827, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.9212, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.17766367137355585, | |
| "grad_norm": 1.3890002965927124, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 1.1388, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.17869062901155328, | |
| "grad_norm": 1.3386279344558716, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 1.1646, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.1797175866495507, | |
| "grad_norm": 1.2495123147964478, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.9937, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.18074454428754813, | |
| "grad_norm": 1.3180009126663208, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 1.0133, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.18177150192554556, | |
| "grad_norm": 1.1055504083633423, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 1.0476, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.18279845956354301, | |
| "grad_norm": 1.3322672843933105, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 1.3564, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.18382541720154044, | |
| "grad_norm": 1.3621183633804321, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 1.1443, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.18485237483953787, | |
| "grad_norm": 1.285597801208496, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.9883, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.1858793324775353, | |
| "grad_norm": 1.2165967226028442, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 1.1198, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.18690629011553272, | |
| "grad_norm": 1.314223289489746, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.8705, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.18793324775353018, | |
| "grad_norm": 1.1345490217208862, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 1.0526, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.1889602053915276, | |
| "grad_norm": 1.3304117918014526, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.9337, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.18998716302952504, | |
| "grad_norm": 1.153505563735962, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 1.0127, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.19101412066752246, | |
| "grad_norm": 1.2609211206436157, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 1.1586, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.1920410783055199, | |
| "grad_norm": 1.2570704221725464, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.9933, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.19306803594351732, | |
| "grad_norm": 1.3671023845672607, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.9252, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.19409499358151477, | |
| "grad_norm": 1.1696784496307373, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.9925, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.1951219512195122, | |
| "grad_norm": 1.176924705505371, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.9838, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.19614890885750963, | |
| "grad_norm": 1.4577269554138184, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 1.2238, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.19717586649550706, | |
| "grad_norm": 1.4867146015167236, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 1.3498, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.19820282413350448, | |
| "grad_norm": 1.5544679164886475, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 1.3545, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.1992297817715019, | |
| "grad_norm": 1.1648284196853638, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.9968, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.20025673940949937, | |
| "grad_norm": 1.2332969903945923, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 1.0235, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.2012836970474968, | |
| "grad_norm": 1.2417412996292114, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 1.0495, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.20231065468549422, | |
| "grad_norm": 1.4818403720855713, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 1.3226, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.20333761232349165, | |
| "grad_norm": 1.5086746215820312, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 1.1384, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.20436456996148908, | |
| "grad_norm": 1.4683679342269897, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 1.5326, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.20539152759948653, | |
| "grad_norm": 1.513578176498413, | |
| "learning_rate": 0.0, | |
| "loss": 1.4335, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.20539152759948653, | |
| "eval_loss": 1.029740810394287, | |
| "eval_runtime": 125.2747, | |
| "eval_samples_per_second": 13.091, | |
| "eval_steps_per_second": 3.273, | |
| "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": 3.017027157491712e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |