| { | |
| "best_metric": 10.344810485839844, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 1.7057569296375266, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.008528784648187633, | |
| "grad_norm": 0.04806673526763916, | |
| "learning_rate": 1e-05, | |
| "loss": 10.3811, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008528784648187633, | |
| "eval_loss": 10.380195617675781, | |
| "eval_runtime": 0.4738, | |
| "eval_samples_per_second": 417.912, | |
| "eval_steps_per_second": 105.533, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.017057569296375266, | |
| "grad_norm": 0.04450615867972374, | |
| "learning_rate": 2e-05, | |
| "loss": 10.3799, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0255863539445629, | |
| "grad_norm": 0.05155586078763008, | |
| "learning_rate": 3e-05, | |
| "loss": 10.3805, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.03411513859275053, | |
| "grad_norm": 0.051454830914735794, | |
| "learning_rate": 4e-05, | |
| "loss": 10.3808, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.042643923240938165, | |
| "grad_norm": 0.05457935854792595, | |
| "learning_rate": 5e-05, | |
| "loss": 10.3807, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0511727078891258, | |
| "grad_norm": 0.05271429941058159, | |
| "learning_rate": 6e-05, | |
| "loss": 10.3808, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.05970149253731343, | |
| "grad_norm": 0.052848007529973984, | |
| "learning_rate": 7e-05, | |
| "loss": 10.3801, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.06823027718550106, | |
| "grad_norm": 0.050696831196546555, | |
| "learning_rate": 8e-05, | |
| "loss": 10.3792, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0767590618336887, | |
| "grad_norm": 0.05586646497249603, | |
| "learning_rate": 9e-05, | |
| "loss": 10.3804, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.08528784648187633, | |
| "grad_norm": 0.05874478071928024, | |
| "learning_rate": 0.0001, | |
| "loss": 10.3787, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.09381663113006397, | |
| "grad_norm": 0.05769229680299759, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 10.3787, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.1023454157782516, | |
| "grad_norm": 0.05879262462258339, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 10.3794, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.11087420042643924, | |
| "grad_norm": 0.06456191837787628, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 10.3783, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.11940298507462686, | |
| "grad_norm": 0.06359651684761047, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 10.3781, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.1279317697228145, | |
| "grad_norm": 0.057515017688274384, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 10.3764, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.13646055437100213, | |
| "grad_norm": 0.07790736854076385, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 10.3775, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.14498933901918976, | |
| "grad_norm": 0.07829299569129944, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 10.3765, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1535181236673774, | |
| "grad_norm": 0.07686816900968552, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 10.3756, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.16204690831556504, | |
| "grad_norm": 0.09612306952476501, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 10.3737, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.17057569296375266, | |
| "grad_norm": 0.09043193608522415, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 10.3753, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.1791044776119403, | |
| "grad_norm": 0.09766452014446259, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 10.3741, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.18763326226012794, | |
| "grad_norm": 0.09835337847471237, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 10.3752, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.19616204690831557, | |
| "grad_norm": 0.11113500595092773, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 10.3718, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.2046908315565032, | |
| "grad_norm": 0.10947182774543762, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 10.3731, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.21321961620469082, | |
| "grad_norm": 0.13433454930782318, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 10.3729, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.22174840085287847, | |
| "grad_norm": 0.1330953687429428, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 10.3688, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.2302771855010661, | |
| "grad_norm": 0.13626112043857574, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 10.3697, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.23880597014925373, | |
| "grad_norm": 0.1350407749414444, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 10.3677, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.24733475479744135, | |
| "grad_norm": 0.14595931768417358, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 10.3679, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.255863539445629, | |
| "grad_norm": 0.12598378956317902, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 10.3686, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.26439232409381663, | |
| "grad_norm": 0.13980260491371155, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.3652, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.27292110874200426, | |
| "grad_norm": 0.14953792095184326, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 10.3656, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.2814498933901919, | |
| "grad_norm": 0.14920461177825928, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 10.3665, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.2899786780383795, | |
| "grad_norm": 0.1592891961336136, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 10.3624, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.29850746268656714, | |
| "grad_norm": 0.1480722576379776, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 10.3649, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.3070362473347548, | |
| "grad_norm": 0.15460079908370972, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 10.3623, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.31556503198294245, | |
| "grad_norm": 0.14019252359867096, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 10.3619, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.32409381663113007, | |
| "grad_norm": 0.14970314502716064, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 10.359, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.3326226012793177, | |
| "grad_norm": 0.14799943566322327, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 10.3582, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.3411513859275053, | |
| "grad_norm": 0.13339582085609436, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 10.3583, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.34968017057569295, | |
| "grad_norm": 0.13061460852622986, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 10.3571, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.3582089552238806, | |
| "grad_norm": 0.1157752275466919, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 10.3578, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.36673773987206826, | |
| "grad_norm": 0.11121196299791336, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 10.3565, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.3752665245202559, | |
| "grad_norm": 0.1145995706319809, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 10.3583, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.3837953091684435, | |
| "grad_norm": 0.11108949035406113, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 10.3553, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.39232409381663114, | |
| "grad_norm": 0.11794111877679825, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 10.3553, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.40085287846481876, | |
| "grad_norm": 0.10476233810186386, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 10.3557, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.4093816631130064, | |
| "grad_norm": 0.09984061866998672, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 10.3544, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.417910447761194, | |
| "grad_norm": 0.09178004413843155, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 10.3568, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.42643923240938164, | |
| "grad_norm": 0.0921761617064476, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 10.3546, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.42643923240938164, | |
| "eval_loss": 10.353571891784668, | |
| "eval_runtime": 0.4739, | |
| "eval_samples_per_second": 417.819, | |
| "eval_steps_per_second": 105.51, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.4349680170575693, | |
| "grad_norm": 0.09714477509260178, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 10.3513, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.44349680170575695, | |
| "grad_norm": 0.08725707978010178, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 10.3538, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.4520255863539446, | |
| "grad_norm": 0.09499584883451462, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 10.3533, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.4605543710021322, | |
| "grad_norm": 0.08286934345960617, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 10.3531, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.4690831556503198, | |
| "grad_norm": 0.07516016066074371, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 10.3521, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.47761194029850745, | |
| "grad_norm": 0.08789828419685364, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 10.3517, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.4861407249466951, | |
| "grad_norm": 0.0797482579946518, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 10.3498, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.4946695095948827, | |
| "grad_norm": 0.09739601612091064, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 10.3509, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.5031982942430704, | |
| "grad_norm": 0.07012306153774261, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 10.3515, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.511727078891258, | |
| "grad_norm": 0.048907216638326645, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 10.351, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.5202558635394456, | |
| "grad_norm": 0.04210042580962181, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 10.3499, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.5287846481876333, | |
| "grad_norm": 0.05919327959418297, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 10.349, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.5373134328358209, | |
| "grad_norm": 0.06116609275341034, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 10.3513, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.5458422174840085, | |
| "grad_norm": 0.056808412075042725, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 10.35, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.5543710021321961, | |
| "grad_norm": 0.04792425408959389, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 10.3506, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.5628997867803838, | |
| "grad_norm": 0.04250369220972061, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 10.3499, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.5714285714285714, | |
| "grad_norm": 0.05015909671783447, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 10.3506, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.579957356076759, | |
| "grad_norm": 0.05162850394845009, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 10.3478, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.5884861407249466, | |
| "grad_norm": 0.04289333149790764, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 10.3505, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.5970149253731343, | |
| "grad_norm": 0.03446891903877258, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 10.3517, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.605543710021322, | |
| "grad_norm": 0.03655017539858818, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 10.3499, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.6140724946695096, | |
| "grad_norm": 0.04696572571992874, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 10.3491, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.6226012793176973, | |
| "grad_norm": 0.0381445474922657, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 10.3496, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.6311300639658849, | |
| "grad_norm": 0.05801752954721451, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 10.3495, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.6396588486140725, | |
| "grad_norm": 0.04724515601992607, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 10.3488, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.6481876332622601, | |
| "grad_norm": 0.03297180309891701, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 10.3488, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.6567164179104478, | |
| "grad_norm": 0.0402832068502903, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 10.349, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.6652452025586354, | |
| "grad_norm": 0.052533362060785294, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 10.3477, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.673773987206823, | |
| "grad_norm": 0.026865648105740547, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 10.3487, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.6823027718550106, | |
| "grad_norm": 0.03491870313882828, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 10.3512, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.6908315565031983, | |
| "grad_norm": 0.056942131370306015, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 10.3485, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.6993603411513859, | |
| "grad_norm": 0.031359389424324036, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 10.3478, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.7078891257995735, | |
| "grad_norm": 0.033854518085718155, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 10.3486, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.7164179104477612, | |
| "grad_norm": 0.04069717973470688, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 10.3517, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.7249466950959488, | |
| "grad_norm": 0.034985728561878204, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 10.349, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.7334754797441365, | |
| "grad_norm": 0.055703889578580856, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 10.3499, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.7420042643923241, | |
| "grad_norm": 0.050922319293022156, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 10.3491, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.7505330490405118, | |
| "grad_norm": 0.04854780435562134, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 10.3488, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.7590618336886994, | |
| "grad_norm": 0.055957891047000885, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 10.3497, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.767590618336887, | |
| "grad_norm": 0.05815952271223068, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 10.3475, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.7761194029850746, | |
| "grad_norm": 0.056256912648677826, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 10.3498, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.7846481876332623, | |
| "grad_norm": 0.03536611050367355, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 10.3518, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.7931769722814499, | |
| "grad_norm": 0.05271325632929802, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 10.3496, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.8017057569296375, | |
| "grad_norm": 0.040671397000551224, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 10.3467, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.8102345415778252, | |
| "grad_norm": 0.04531000927090645, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 10.3475, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.8187633262260128, | |
| "grad_norm": 0.04373055696487427, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 10.3474, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.8272921108742004, | |
| "grad_norm": 0.03367888554930687, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 10.3483, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.835820895522388, | |
| "grad_norm": 0.03936287760734558, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 10.3485, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.8443496801705757, | |
| "grad_norm": 0.0355980359017849, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 10.3477, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.8528784648187633, | |
| "grad_norm": 0.03361448645591736, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 10.3496, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.8528784648187633, | |
| "eval_loss": 10.348418235778809, | |
| "eval_runtime": 0.477, | |
| "eval_samples_per_second": 415.097, | |
| "eval_steps_per_second": 104.822, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.8614072494669509, | |
| "grad_norm": 0.03691884130239487, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 10.3462, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.8699360341151386, | |
| "grad_norm": 0.03882606327533722, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 10.3486, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.8784648187633263, | |
| "grad_norm": 0.03755461797118187, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 10.3479, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.8869936034115139, | |
| "grad_norm": 0.04736875370144844, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 10.3471, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.8955223880597015, | |
| "grad_norm": 0.029074519872665405, | |
| "learning_rate": 5e-05, | |
| "loss": 10.3486, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.9040511727078892, | |
| "grad_norm": 0.031900398433208466, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 10.3472, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.9125799573560768, | |
| "grad_norm": 0.03517429158091545, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 10.3458, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.9211087420042644, | |
| "grad_norm": 0.03357727825641632, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 10.3479, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.929637526652452, | |
| "grad_norm": 0.03584626689553261, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 10.3478, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.9381663113006397, | |
| "grad_norm": 0.0471317358314991, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 10.3474, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.9466950959488273, | |
| "grad_norm": 0.038973815739154816, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 10.3483, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.9552238805970149, | |
| "grad_norm": 0.0343744233250618, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 10.3477, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.9637526652452025, | |
| "grad_norm": 0.037565600126981735, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 10.3485, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.9722814498933902, | |
| "grad_norm": 0.04688845947384834, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 10.3467, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.9808102345415778, | |
| "grad_norm": 0.05078093707561493, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 10.3465, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.9893390191897654, | |
| "grad_norm": 0.04903563857078552, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 10.3467, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.997867803837953, | |
| "grad_norm": 0.043443650007247925, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 10.3463, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.0063965884861408, | |
| "grad_norm": 0.06975448876619339, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 18.2617, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.0149253731343284, | |
| "grad_norm": 0.05583430454134941, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 10.1853, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.023454157782516, | |
| "grad_norm": 0.054079584777355194, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 10.4074, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.0319829424307037, | |
| "grad_norm": 0.05480353534221649, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 10.3191, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.0405117270788913, | |
| "grad_norm": 0.05557660385966301, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 10.2915, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.049040511727079, | |
| "grad_norm": 0.05558188259601593, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 10.326, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.0575692963752665, | |
| "grad_norm": 0.033676628023386, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 10.4113, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.0660980810234542, | |
| "grad_norm": 0.04203203693032265, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 10.3098, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.0746268656716418, | |
| "grad_norm": 0.040990713983774185, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 10.3694, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.0831556503198294, | |
| "grad_norm": 0.04310883209109306, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 10.3483, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.091684434968017, | |
| "grad_norm": 0.03962789475917816, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 10.3523, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.1002132196162047, | |
| "grad_norm": 0.04049231484532356, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 10.4013, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.1087420042643923, | |
| "grad_norm": 0.04194732382893562, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 10.2829, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.11727078891258, | |
| "grad_norm": 0.036313239485025406, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 10.3324, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.1257995735607675, | |
| "grad_norm": 0.046480074524879456, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 10.3625, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.1343283582089552, | |
| "grad_norm": 0.04361138865351677, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 10.3698, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.1428571428571428, | |
| "grad_norm": 0.04324062541127205, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 10.3045, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.1513859275053304, | |
| "grad_norm": 0.05079786852002144, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 10.3997, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.159914712153518, | |
| "grad_norm": 0.03967788442969322, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 10.3362, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.1684434968017057, | |
| "grad_norm": 0.04797113686800003, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 10.2458, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.1769722814498933, | |
| "grad_norm": 0.03926164284348488, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 10.4052, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.1855010660980811, | |
| "grad_norm": 0.046536751091480255, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 10.3523, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.1940298507462686, | |
| "grad_norm": 0.041127756237983704, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 10.3909, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.2025586353944564, | |
| "grad_norm": 0.05339064821600914, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 10.2972, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.2110874200426438, | |
| "grad_norm": 0.041502151638269424, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 10.3399, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.2196162046908317, | |
| "grad_norm": 0.04914531484246254, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 10.4175, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.2281449893390193, | |
| "grad_norm": 0.052515558898448944, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 10.2855, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.236673773987207, | |
| "grad_norm": 0.04797741025686264, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 10.3788, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.2452025586353945, | |
| "grad_norm": 0.04141698777675629, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 10.3915, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.2537313432835822, | |
| "grad_norm": 0.04757013916969299, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 10.4077, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.2622601279317698, | |
| "grad_norm": 0.050195809453725815, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 10.2124, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.2707889125799574, | |
| "grad_norm": 0.05028505623340607, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 10.3313, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.279317697228145, | |
| "grad_norm": 0.050132013857364655, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 10.398, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.279317697228145, | |
| "eval_loss": 10.345519065856934, | |
| "eval_runtime": 0.4732, | |
| "eval_samples_per_second": 418.456, | |
| "eval_steps_per_second": 105.671, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.2878464818763327, | |
| "grad_norm": 0.04571465775370598, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 10.3739, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.2963752665245203, | |
| "grad_norm": 0.049470044672489166, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 10.2484, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.304904051172708, | |
| "grad_norm": 0.05007629841566086, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 10.4077, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.3134328358208955, | |
| "grad_norm": 0.039351433515548706, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 10.3584, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.3219616204690832, | |
| "grad_norm": 0.03760872036218643, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 10.3071, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.3304904051172708, | |
| "grad_norm": 0.03820711746811867, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 10.3364, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.3390191897654584, | |
| "grad_norm": 0.045905131846666336, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 10.3887, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.347547974413646, | |
| "grad_norm": 0.052111707627773285, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 10.3202, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.3560767590618337, | |
| "grad_norm": 0.04230839014053345, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 10.3565, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.3646055437100213, | |
| "grad_norm": 0.049018844962120056, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 10.2842, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.373134328358209, | |
| "grad_norm": 0.0350567102432251, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 10.4259, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.3816631130063965, | |
| "grad_norm": 0.055679138749837875, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 10.2851, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.3901918976545842, | |
| "grad_norm": 0.040636684745550156, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 10.3985, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.3987206823027718, | |
| "grad_norm": 0.04427097365260124, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 10.317, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.4072494669509594, | |
| "grad_norm": 0.0511578805744648, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 10.3193, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.415778251599147, | |
| "grad_norm": 0.03863763436675072, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 10.3294, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.4243070362473347, | |
| "grad_norm": 0.043290864676237106, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 10.4039, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.4328358208955223, | |
| "grad_norm": 0.054091088473796844, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 10.326, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.44136460554371, | |
| "grad_norm": 0.044167011976242065, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 10.2472, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.4498933901918978, | |
| "grad_norm": 0.03642076253890991, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 10.4793, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.4584221748400852, | |
| "grad_norm": 0.046830929815769196, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 10.2462, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.466950959488273, | |
| "grad_norm": 0.050203170627355576, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 10.4537, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.4754797441364604, | |
| "grad_norm": 0.05073232203722, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 10.3044, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.4840085287846483, | |
| "grad_norm": 0.058260317891836166, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 10.313, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.4925373134328357, | |
| "grad_norm": 0.05089752748608589, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 10.4409, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.5010660980810235, | |
| "grad_norm": 0.04901741072535515, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 10.4278, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.509594882729211, | |
| "grad_norm": 0.057309091091156006, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 10.1737, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.5181236673773988, | |
| "grad_norm": 0.05338204279541969, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 10.4376, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.5266524520255862, | |
| "grad_norm": 0.04672335833311081, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 10.3005, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.535181236673774, | |
| "grad_norm": 0.040306366980075836, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 10.3444, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.5437100213219617, | |
| "grad_norm": 0.048869770020246506, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 10.2567, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.5522388059701493, | |
| "grad_norm": 0.04840242490172386, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 10.3477, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.560767590618337, | |
| "grad_norm": 0.054581016302108765, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 10.3663, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.5692963752665245, | |
| "grad_norm": 0.05537518858909607, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 10.3832, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 1.5778251599147122, | |
| "grad_norm": 0.04769334942102432, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 10.3319, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.5863539445628998, | |
| "grad_norm": 0.041567347943782806, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 10.3281, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 1.5948827292110874, | |
| "grad_norm": 0.05261693149805069, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 10.3977, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 1.603411513859275, | |
| "grad_norm": 0.048501595854759216, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 10.2834, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 1.6119402985074627, | |
| "grad_norm": 0.059001125395298004, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 10.3218, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 1.6204690831556503, | |
| "grad_norm": 0.04259403049945831, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 10.3916, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.628997867803838, | |
| "grad_norm": 0.05021576210856438, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 10.2634, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 1.6375266524520256, | |
| "grad_norm": 0.04164980351924896, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 10.4188, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 1.6460554371002132, | |
| "grad_norm": 0.05098918080329895, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 10.3068, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 1.6545842217484008, | |
| "grad_norm": 0.054227594286203384, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 10.3639, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 1.6631130063965884, | |
| "grad_norm": 0.041704799979925156, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 10.3768, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.671641791044776, | |
| "grad_norm": 0.04737214371562004, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 10.3402, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 1.680170575692964, | |
| "grad_norm": 0.044063325971364975, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 10.3622, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 1.6886993603411513, | |
| "grad_norm": 0.054453033953905106, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 10.2999, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 1.6972281449893392, | |
| "grad_norm": 0.037610042840242386, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 10.3255, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 1.7057569296375266, | |
| "grad_norm": 0.04368090257048607, | |
| "learning_rate": 0.0, | |
| "loss": 10.375, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.7057569296375266, | |
| "eval_loss": 10.344810485839844, | |
| "eval_runtime": 0.475, | |
| "eval_samples_per_second": 416.811, | |
| "eval_steps_per_second": 105.255, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "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": 22306646163456.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |