| { | |
| "best_metric": 0.0027467531617730856, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 0.9888751545117429, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004944375772558714, | |
| "grad_norm": 0.30713972449302673, | |
| "learning_rate": 1e-05, | |
| "loss": 0.0624, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004944375772558714, | |
| "eval_loss": 0.6954474449157715, | |
| "eval_runtime": 49.4931, | |
| "eval_samples_per_second": 6.89, | |
| "eval_steps_per_second": 1.738, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.009888751545117428, | |
| "grad_norm": 0.5427398681640625, | |
| "learning_rate": 2e-05, | |
| "loss": 0.1095, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.014833127317676144, | |
| "grad_norm": 0.5647732019424438, | |
| "learning_rate": 3e-05, | |
| "loss": 0.1421, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.019777503090234856, | |
| "grad_norm": 0.5855129361152649, | |
| "learning_rate": 4e-05, | |
| "loss": 0.1552, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.024721878862793572, | |
| "grad_norm": 0.48615047335624695, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1517, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.029666254635352288, | |
| "grad_norm": 0.5483637452125549, | |
| "learning_rate": 6e-05, | |
| "loss": 0.1439, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.034610630407911, | |
| "grad_norm": 0.6646473407745361, | |
| "learning_rate": 7e-05, | |
| "loss": 0.1303, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03955500618046971, | |
| "grad_norm": 0.851876974105835, | |
| "learning_rate": 8e-05, | |
| "loss": 0.0926, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.04449938195302843, | |
| "grad_norm": 0.6939786076545715, | |
| "learning_rate": 9e-05, | |
| "loss": 0.0573, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.049443757725587144, | |
| "grad_norm": 0.5447797179222107, | |
| "learning_rate": 0.0001, | |
| "loss": 0.0319, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.054388133498145856, | |
| "grad_norm": 0.32662445306777954, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.017, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.059332509270704575, | |
| "grad_norm": 0.7598249316215515, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.0196, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.06427688504326329, | |
| "grad_norm": 0.33270764350891113, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.0082, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.069221260815822, | |
| "grad_norm": 0.3599153161048889, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.0052, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.07416563658838071, | |
| "grad_norm": 1.27303147315979, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.0028, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07911001236093942, | |
| "grad_norm": 0.4815302789211273, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.0045, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.08405438813349815, | |
| "grad_norm": 0.5560856461524963, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.0081, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.08899876390605686, | |
| "grad_norm": 0.22989942133426666, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.0013, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.09394313967861558, | |
| "grad_norm": 0.496055543422699, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.0029, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.09888751545117429, | |
| "grad_norm": 0.2765651345252991, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.002, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.103831891223733, | |
| "grad_norm": 0.492607057094574, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.0037, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.10877626699629171, | |
| "grad_norm": 0.10228199511766434, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.0015, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.11372064276885044, | |
| "grad_norm": 0.3049921989440918, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.0028, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.11866501854140915, | |
| "grad_norm": 0.28797903656959534, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.0016, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.12360939431396786, | |
| "grad_norm": 0.3941512107849121, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.0046, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.12855377008652658, | |
| "grad_norm": 0.4419980049133301, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.0026, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.1334981458590853, | |
| "grad_norm": 0.4615868628025055, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.0062, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.138442521631644, | |
| "grad_norm": 0.13651740550994873, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.0014, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.1433868974042027, | |
| "grad_norm": 0.3100791871547699, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.001, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.14833127317676142, | |
| "grad_norm": 0.20339711010456085, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.0012, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.15327564894932014, | |
| "grad_norm": 0.47062361240386963, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.0037, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.15822002472187885, | |
| "grad_norm": 1.2841547727584839, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.0064, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.1631644004944376, | |
| "grad_norm": 0.018028907477855682, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.0001, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.1681087762669963, | |
| "grad_norm": 0.4496955871582031, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.0113, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.173053152039555, | |
| "grad_norm": 0.6527530550956726, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.0099, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.17799752781211373, | |
| "grad_norm": 0.8808625936508179, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.0035, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.18294190358467244, | |
| "grad_norm": 0.17417174577713013, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0008, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.18788627935723115, | |
| "grad_norm": 6.353917121887207, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.0349, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.19283065512978986, | |
| "grad_norm": 4.7959208488464355, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.0128, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.19777503090234858, | |
| "grad_norm": 1.555270791053772, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.0126, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.2027194066749073, | |
| "grad_norm": 0.7290369868278503, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.0104, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.207663782447466, | |
| "grad_norm": 0.06998298317193985, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.0006, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.2126081582200247, | |
| "grad_norm": 0.6854199171066284, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.0094, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.21755253399258342, | |
| "grad_norm": 0.6097396612167358, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.0077, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.22249690976514216, | |
| "grad_norm": 0.6831669211387634, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.0045, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.22744128553770088, | |
| "grad_norm": 0.7666204571723938, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.008, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2323856613102596, | |
| "grad_norm": 0.4165506660938263, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.0045, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.2373300370828183, | |
| "grad_norm": 9.028611183166504, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.2058, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.242274412855377, | |
| "grad_norm": 2.077606201171875, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.0209, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.24721878862793573, | |
| "grad_norm": 5.175272464752197, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.0601, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.24721878862793573, | |
| "eval_loss": 0.05587746202945709, | |
| "eval_runtime": 50.2439, | |
| "eval_samples_per_second": 6.787, | |
| "eval_steps_per_second": 1.712, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.2521631644004944, | |
| "grad_norm": 16.072107315063477, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.3238, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.25710754017305315, | |
| "grad_norm": 1.7522956132888794, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.0625, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.2620519159456119, | |
| "grad_norm": 0.3604808449745178, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.0077, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.2669962917181706, | |
| "grad_norm": 0.15032720565795898, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.0012, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.2719406674907293, | |
| "grad_norm": 0.07196947932243347, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.0013, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.276885043263288, | |
| "grad_norm": 0.4468744695186615, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.0057, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.28182941903584674, | |
| "grad_norm": 0.268869012594223, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.0033, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.2867737948084054, | |
| "grad_norm": 0.4217556118965149, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0048, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.29171817058096416, | |
| "grad_norm": 0.17394770681858063, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.0046, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.29666254635352285, | |
| "grad_norm": 0.20989587903022766, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.0027, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.3016069221260816, | |
| "grad_norm": 0.13680729269981384, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.0025, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.3065512978986403, | |
| "grad_norm": 0.019447751343250275, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0007, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.311495673671199, | |
| "grad_norm": 0.01552569679915905, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.0005, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.3164400494437577, | |
| "grad_norm": 0.22481144964694977, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.0046, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.32138442521631644, | |
| "grad_norm": 0.05301257595419884, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.0011, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.3263288009888752, | |
| "grad_norm": 0.009488530457019806, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0003, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.33127317676143386, | |
| "grad_norm": 0.1371014565229416, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0018, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.3362175525339926, | |
| "grad_norm": 0.17769162356853485, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0028, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.3411619283065513, | |
| "grad_norm": 0.026774050667881966, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.0003, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.34610630407911, | |
| "grad_norm": 0.007345946971327066, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.0001, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.3510506798516687, | |
| "grad_norm": 0.07110948115587234, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.0006, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.35599505562422745, | |
| "grad_norm": 0.010747851803898811, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0002, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.36093943139678614, | |
| "grad_norm": 0.5883771777153015, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.0035, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3658838071693449, | |
| "grad_norm": 0.11903329938650131, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.0004, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.37082818294190356, | |
| "grad_norm": 0.012866495177149773, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.0001, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3757725587144623, | |
| "grad_norm": 0.008181729353964329, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0001, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.380716934487021, | |
| "grad_norm": 0.057840391993522644, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0002, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3856613102595797, | |
| "grad_norm": 0.01270771399140358, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0002, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.39060568603213847, | |
| "grad_norm": 0.0398685522377491, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0003, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.39555006180469715, | |
| "grad_norm": 0.35647302865982056, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.0058, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.4004944375772559, | |
| "grad_norm": 2.075082302093506, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0129, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.4054388133498146, | |
| "grad_norm": 0.00358505267649889, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0001, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.4103831891223733, | |
| "grad_norm": 1.2216726541519165, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0022, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.415327564894932, | |
| "grad_norm": 0.1473245769739151, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.001, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.42027194066749074, | |
| "grad_norm": 0.02910473383963108, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.0004, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.4252163164400494, | |
| "grad_norm": 0.05410793796181679, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0009, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.43016069221260816, | |
| "grad_norm": 0.39348503947257996, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0138, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.43510506798516685, | |
| "grad_norm": 0.2215208262205124, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0062, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.4400494437577256, | |
| "grad_norm": 0.037982840090990067, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.001, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.44499381953028433, | |
| "grad_norm": 0.11779870092868805, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.0018, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.449938195302843, | |
| "grad_norm": 0.36766117811203003, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.0042, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.45488257107540175, | |
| "grad_norm": 0.5219226479530334, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.0085, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.45982694684796044, | |
| "grad_norm": 0.3052183985710144, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.0128, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.4647713226205192, | |
| "grad_norm": 0.02772422693669796, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0007, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.46971569839307786, | |
| "grad_norm": 0.028806524351239204, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0006, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.4746600741656366, | |
| "grad_norm": 0.02134646289050579, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0005, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.4796044499381953, | |
| "grad_norm": 0.026467913761734962, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0006, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.484548825710754, | |
| "grad_norm": 0.15709063410758972, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0015, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.4894932014833127, | |
| "grad_norm": 0.023603875190019608, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0006, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.49443757725587145, | |
| "grad_norm": 0.018728632479906082, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0005, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.49443757725587145, | |
| "eval_loss": 0.0032831195276230574, | |
| "eval_runtime": 50.25, | |
| "eval_samples_per_second": 6.786, | |
| "eval_steps_per_second": 1.711, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.49938195302843014, | |
| "grad_norm": 4.372404098510742, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0238, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.5043263288009888, | |
| "grad_norm": 0.11278031766414642, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.0019, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.5092707045735476, | |
| "grad_norm": 0.03654538094997406, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.0006, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.5142150803461063, | |
| "grad_norm": 0.008726762607693672, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.0001, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.519159456118665, | |
| "grad_norm": 0.120099276304245, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0025, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.5241038318912238, | |
| "grad_norm": 0.18376444280147552, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.003, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.5290482076637825, | |
| "grad_norm": 0.04906797781586647, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0011, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.5339925834363412, | |
| "grad_norm": 0.10075504332780838, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.0031, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.5389369592088998, | |
| "grad_norm": 0.009199939668178558, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0001, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.5438813349814586, | |
| "grad_norm": 0.055573537945747375, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0005, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.5488257107540173, | |
| "grad_norm": 0.046735599637031555, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0005, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.553770086526576, | |
| "grad_norm": 0.033676985651254654, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0004, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.5587144622991347, | |
| "grad_norm": 0.03807822987437248, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0003, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5636588380716935, | |
| "grad_norm": 0.007940626703202724, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.0001, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5686032138442522, | |
| "grad_norm": 0.24582535028457642, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0017, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5735475896168108, | |
| "grad_norm": 0.007122722454369068, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.0001, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5784919653893696, | |
| "grad_norm": 0.09866178035736084, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0009, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5834363411619283, | |
| "grad_norm": 0.016884470358490944, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0002, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.588380716934487, | |
| "grad_norm": 0.01916952058672905, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0003, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.5933250927070457, | |
| "grad_norm": 0.017505362629890442, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0002, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5982694684796045, | |
| "grad_norm": 0.01098876167088747, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0001, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.6032138442521632, | |
| "grad_norm": 0.035289134830236435, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0002, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.6081582200247219, | |
| "grad_norm": 0.0005960287526249886, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.6131025957972805, | |
| "grad_norm": 0.0057175313122570515, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0001, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.6180469715698393, | |
| "grad_norm": 0.001557542011141777, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.622991347342398, | |
| "grad_norm": 0.1430407613515854, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0006, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.6279357231149567, | |
| "grad_norm": 0.3505101501941681, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0033, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.6328800988875154, | |
| "grad_norm": 0.0010281288996338844, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.6378244746600742, | |
| "grad_norm": 0.001765651861205697, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.0, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.6427688504326329, | |
| "grad_norm": 0.01193635817617178, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0001, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.6477132262051916, | |
| "grad_norm": 0.006312054581940174, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.6526576019777504, | |
| "grad_norm": 0.1974785029888153, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.0024, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.657601977750309, | |
| "grad_norm": 0.08454439789056778, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0003, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.6625463535228677, | |
| "grad_norm": 0.04529717564582825, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0002, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.6674907292954264, | |
| "grad_norm": 0.11105463653802872, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0003, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.6724351050679852, | |
| "grad_norm": 0.043228406459093094, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0004, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.6773794808405439, | |
| "grad_norm": 0.2780735492706299, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0018, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6823238566131026, | |
| "grad_norm": 0.010291093029081821, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0001, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6872682323856613, | |
| "grad_norm": 0.096773162484169, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0003, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.69221260815822, | |
| "grad_norm": 0.01383780688047409, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0001, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6971569839307787, | |
| "grad_norm": 0.07840044796466827, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0002, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.7021013597033374, | |
| "grad_norm": 0.0076470342464745045, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0001, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.7070457354758962, | |
| "grad_norm": 0.29595521092414856, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.002, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.7119901112484549, | |
| "grad_norm": 0.037160228937864304, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0002, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.7169344870210136, | |
| "grad_norm": 0.21545666456222534, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0018, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.7218788627935723, | |
| "grad_norm": 0.026764418929815292, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0002, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.7268232385661311, | |
| "grad_norm": 0.004044615663588047, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.0001, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.7317676143386898, | |
| "grad_norm": 0.15351343154907227, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0007, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.7367119901112484, | |
| "grad_norm": 1.8308340311050415, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.0054, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.7416563658838071, | |
| "grad_norm": 0.02782491408288479, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0002, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7416563658838071, | |
| "eval_loss": 0.0027467531617730856, | |
| "eval_runtime": 50.3157, | |
| "eval_samples_per_second": 6.777, | |
| "eval_steps_per_second": 1.709, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.7466007416563659, | |
| "grad_norm": 0.3695788085460663, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0077, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.7515451174289246, | |
| "grad_norm": 0.0908498466014862, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0021, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.7564894932014833, | |
| "grad_norm": 0.05082995444536209, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0001, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.761433868974042, | |
| "grad_norm": 0.013676106929779053, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0001, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.7663782447466008, | |
| "grad_norm": 0.003630921943113208, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.7713226205191595, | |
| "grad_norm": 0.07495836913585663, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0002, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.7762669962917181, | |
| "grad_norm": 0.002006974071264267, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.7812113720642769, | |
| "grad_norm": 0.0012374644866213202, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.7861557478368356, | |
| "grad_norm": 0.0207204706966877, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.7911001236093943, | |
| "grad_norm": 0.09979017078876495, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0052, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.796044499381953, | |
| "grad_norm": 0.01835060678422451, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0001, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.8009888751545118, | |
| "grad_norm": 0.0016144438413903117, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.8059332509270705, | |
| "grad_norm": 0.004404058679938316, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0001, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.8108776266996292, | |
| "grad_norm": 0.027366681024432182, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0004, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.8158220024721878, | |
| "grad_norm": 0.03986135497689247, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0003, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.8207663782447466, | |
| "grad_norm": 0.06387261301279068, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0006, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.8257107540173053, | |
| "grad_norm": 0.0023310771211981773, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.0001, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.830655129789864, | |
| "grad_norm": 0.012357814237475395, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0001, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.8355995055624228, | |
| "grad_norm": 0.06224483996629715, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0024, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.8405438813349815, | |
| "grad_norm": 0.005502707790583372, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.0001, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.8454882571075402, | |
| "grad_norm": 0.018754860386252403, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0001, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.8504326328800988, | |
| "grad_norm": 0.022459518164396286, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0001, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.8553770086526576, | |
| "grad_norm": 0.005487468559294939, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.8603213844252163, | |
| "grad_norm": 0.006107981316745281, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0001, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.865265760197775, | |
| "grad_norm": 0.038404908031225204, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0001, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.8702101359703337, | |
| "grad_norm": 0.0015994664281606674, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.8751545117428925, | |
| "grad_norm": 0.0016457941383123398, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.8800988875154512, | |
| "grad_norm": 0.0008344638044945896, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.8850432632880099, | |
| "grad_norm": 0.0071343593299388885, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.8899876390605687, | |
| "grad_norm": 1.7537447214126587, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0056, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.8949320148331273, | |
| "grad_norm": 0.011215045116841793, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0001, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.899876390605686, | |
| "grad_norm": 0.24245740473270416, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0008, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.9048207663782447, | |
| "grad_norm": 0.0015044215833768249, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.9097651421508035, | |
| "grad_norm": 0.01834225282073021, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0002, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.9147095179233622, | |
| "grad_norm": 0.005067096557468176, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0001, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.9196538936959209, | |
| "grad_norm": 0.014924520626664162, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0002, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.9245982694684796, | |
| "grad_norm": 0.0009934515692293644, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.9295426452410384, | |
| "grad_norm": 0.22189263999462128, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.0023, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.934487021013597, | |
| "grad_norm": 0.0038784232456237078, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0001, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.9394313967861557, | |
| "grad_norm": 0.09324667602777481, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0008, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.9443757725587144, | |
| "grad_norm": 0.55302894115448, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0019, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.9493201483312732, | |
| "grad_norm": 0.0053778099827468395, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0001, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.9542645241038319, | |
| "grad_norm": 0.01738799177110195, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0001, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.9592088998763906, | |
| "grad_norm": 0.016438327729701996, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0003, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.9641532756489494, | |
| "grad_norm": 0.0014637430431321263, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.969097651421508, | |
| "grad_norm": 0.0018220784841105342, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.9740420271940667, | |
| "grad_norm": 0.045078448951244354, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0004, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.9789864029666254, | |
| "grad_norm": 0.00537289772182703, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0001, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.9839307787391842, | |
| "grad_norm": 0.026542989537119865, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.0002, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.9888751545117429, | |
| "grad_norm": 0.21533222496509552, | |
| "learning_rate": 0.0, | |
| "loss": 0.0016, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.9888751545117429, | |
| "eval_loss": 0.0030253820586949587, | |
| "eval_runtime": 50.325, | |
| "eval_samples_per_second": 6.776, | |
| "eval_steps_per_second": 1.709, | |
| "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": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 6.034054314983424e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |