| { | |
| "best_metric": 0.3386111855506897, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.13348907058234608, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0006674453529117303, | |
| "grad_norm": 1.3115936517715454, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3883, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006674453529117303, | |
| "eval_loss": 0.5509489178657532, | |
| "eval_runtime": 341.2284, | |
| "eval_samples_per_second": 7.397, | |
| "eval_steps_per_second": 1.849, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0013348907058234607, | |
| "grad_norm": 1.6735014915466309, | |
| "learning_rate": 2e-05, | |
| "loss": 0.4738, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.002002336058735191, | |
| "grad_norm": 1.3301773071289062, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4661, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.0026697814116469214, | |
| "grad_norm": 1.0363891124725342, | |
| "learning_rate": 4e-05, | |
| "loss": 0.4858, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0033372267645586516, | |
| "grad_norm": 1.2841607332229614, | |
| "learning_rate": 5e-05, | |
| "loss": 0.422, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.004004672117470382, | |
| "grad_norm": 1.1053370237350464, | |
| "learning_rate": 6e-05, | |
| "loss": 0.4439, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.0046721174703821125, | |
| "grad_norm": 1.050881028175354, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4456, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.005339562823293843, | |
| "grad_norm": 0.9765268564224243, | |
| "learning_rate": 8e-05, | |
| "loss": 0.4314, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.006007008176205573, | |
| "grad_norm": 0.8982433080673218, | |
| "learning_rate": 9e-05, | |
| "loss": 0.44, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.006674453529117303, | |
| "grad_norm": 0.8392170071601868, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4002, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0073418988820290335, | |
| "grad_norm": 0.8162698745727539, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.4286, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.008009344234940765, | |
| "grad_norm": 0.8021808862686157, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.396, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.008676789587852495, | |
| "grad_norm": 0.8561233878135681, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.3681, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.009344234940764225, | |
| "grad_norm": 0.8698616027832031, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.3713, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.010011680293675955, | |
| "grad_norm": 0.9252683520317078, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.3912, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.010679125646587686, | |
| "grad_norm": 0.8583617210388184, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.3653, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.011346570999499416, | |
| "grad_norm": 1.0332088470458984, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.368, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.012014016352411146, | |
| "grad_norm": 0.8309275507926941, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.3778, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.012681461705322876, | |
| "grad_norm": 0.8020274639129639, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.3559, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.013348907058234607, | |
| "grad_norm": 0.7628377079963684, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.3719, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.014016352411146337, | |
| "grad_norm": 0.7645742297172546, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.3781, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.014683797764058067, | |
| "grad_norm": 0.8189457058906555, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.3673, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.015351243116969799, | |
| "grad_norm": 0.7485584020614624, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.3534, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.01601868846988153, | |
| "grad_norm": 0.8096043467521667, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.3719, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.01668613382279326, | |
| "grad_norm": 0.8690758347511292, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.3617, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.01735357917570499, | |
| "grad_norm": 0.9931303858757019, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.3474, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.01802102452861672, | |
| "grad_norm": 0.7992089986801147, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.325, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.01868846988152845, | |
| "grad_norm": 0.8595227003097534, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.3573, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.01935591523444018, | |
| "grad_norm": 0.8078036904335022, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.3238, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.02002336058735191, | |
| "grad_norm": 1.0003902912139893, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.3363, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.02069080594026364, | |
| "grad_norm": 1.2744858264923096, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3327, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.02135825129317537, | |
| "grad_norm": 0.8229513168334961, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.3599, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0220256966460871, | |
| "grad_norm": 0.8517017960548401, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.3159, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.02269314199899883, | |
| "grad_norm": 0.8195431232452393, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.347, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.023360587351910562, | |
| "grad_norm": 0.7758495807647705, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.3452, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.024028032704822292, | |
| "grad_norm": 0.9046258330345154, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.3283, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.024695478057734022, | |
| "grad_norm": 0.8832509517669678, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.3407, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.025362923410645753, | |
| "grad_norm": 0.8086565732955933, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.3328, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.026030368763557483, | |
| "grad_norm": 0.8168448209762573, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.3445, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.026697814116469213, | |
| "grad_norm": 0.8411976099014282, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.4082, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.027365259469380943, | |
| "grad_norm": 0.7408682107925415, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.3566, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.028032704822292673, | |
| "grad_norm": 0.7202665209770203, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.3621, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.028700150175204404, | |
| "grad_norm": 0.7591381669044495, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.3593, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.029367595528116134, | |
| "grad_norm": 0.8871503472328186, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.3219, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.030035040881027864, | |
| "grad_norm": 0.8089191317558289, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.3344, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.030702486233939598, | |
| "grad_norm": 0.7209922671318054, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.3429, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.03136993158685133, | |
| "grad_norm": 0.7959047555923462, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.3491, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.03203737693976306, | |
| "grad_norm": 0.7543495893478394, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.315, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.03270482229267479, | |
| "grad_norm": 0.7837464809417725, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.3248, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.03337226764558652, | |
| "grad_norm": 0.7841838002204895, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.2629, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03337226764558652, | |
| "eval_loss": 0.3772207200527191, | |
| "eval_runtime": 343.8739, | |
| "eval_samples_per_second": 7.34, | |
| "eval_steps_per_second": 1.835, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03403971299849825, | |
| "grad_norm": 0.6118820905685425, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.3743, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.03470715835140998, | |
| "grad_norm": 0.5983527898788452, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.3938, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.03537460370432171, | |
| "grad_norm": 0.5496025085449219, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.406, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.03604204905723344, | |
| "grad_norm": 0.48342323303222656, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.3577, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03670949441014517, | |
| "grad_norm": 0.5231099724769592, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.38, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.0373769397630569, | |
| "grad_norm": 0.6181307435035706, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.381, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03804438511596863, | |
| "grad_norm": 0.5407585501670837, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.3527, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03871183046888036, | |
| "grad_norm": 0.5559694766998291, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.3603, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03937927582179209, | |
| "grad_norm": 0.5667877197265625, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3589, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.04004672117470382, | |
| "grad_norm": 0.5622410178184509, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.3699, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04071416652761555, | |
| "grad_norm": 0.5705074071884155, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.3323, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.04138161188052728, | |
| "grad_norm": 0.627193808555603, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.3894, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.04204905723343901, | |
| "grad_norm": 0.6958135366439819, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.4073, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.04271650258635074, | |
| "grad_norm": 0.6556777358055115, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.3902, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.04338394793926247, | |
| "grad_norm": 0.6548662781715393, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.3973, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0440513932921742, | |
| "grad_norm": 0.560892641544342, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.3441, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.04471883864508593, | |
| "grad_norm": 0.6378101706504822, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.3869, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.04538628399799766, | |
| "grad_norm": 0.5806630253791809, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.3231, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.04605372935090939, | |
| "grad_norm": 0.6209865212440491, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.3884, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.046721174703821124, | |
| "grad_norm": 0.6875416040420532, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.3989, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.047388620056732854, | |
| "grad_norm": 0.6756691932678223, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.3941, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.048056065409644584, | |
| "grad_norm": 0.5739635229110718, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.3334, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.048723510762556314, | |
| "grad_norm": 0.6554338932037354, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.3746, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.049390956115468045, | |
| "grad_norm": 0.681486964225769, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.3759, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.050058401468379775, | |
| "grad_norm": 0.691131055355072, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.3621, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.050725846821291505, | |
| "grad_norm": 0.6469898223876953, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.3632, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.051393292174203235, | |
| "grad_norm": 0.6381687521934509, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.3235, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.052060737527114966, | |
| "grad_norm": 0.6917677521705627, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.3784, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.052728182880026696, | |
| "grad_norm": 0.7282185554504395, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.3829, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.053395628232938426, | |
| "grad_norm": 0.682506799697876, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.3377, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.054063073585850156, | |
| "grad_norm": 0.6474214792251587, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.337, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.054730518938761887, | |
| "grad_norm": 0.6790083646774292, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3625, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.05539796429167362, | |
| "grad_norm": 0.6722200512886047, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.3093, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.05606540964458535, | |
| "grad_norm": 0.7422890663146973, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.3259, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.05673285499749708, | |
| "grad_norm": 0.7811063528060913, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.2953, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.05740030035040881, | |
| "grad_norm": 0.7161905765533447, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.314, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.05806774570332054, | |
| "grad_norm": 0.7354428172111511, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.3034, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.05873519105623227, | |
| "grad_norm": 0.7286890149116516, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.3343, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.059402636409144, | |
| "grad_norm": 0.777157187461853, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.3367, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.06007008176205573, | |
| "grad_norm": 0.7120463252067566, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.3431, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.06073752711496746, | |
| "grad_norm": 1.0516895055770874, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.3649, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.061404972467879196, | |
| "grad_norm": 0.7355644106864929, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.3594, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.062072417820790926, | |
| "grad_norm": 0.7184754014015198, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3783, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.06273986317370266, | |
| "grad_norm": 0.6837098598480225, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.3575, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.06340730852661439, | |
| "grad_norm": 0.6682641506195068, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.3271, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.06407475387952612, | |
| "grad_norm": 0.7006938457489014, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3065, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.06474219923243785, | |
| "grad_norm": 0.6511181592941284, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.3133, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.06540964458534958, | |
| "grad_norm": 0.7467932105064392, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.3351, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.06607708993826131, | |
| "grad_norm": 0.684815526008606, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.2876, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.06674453529117304, | |
| "grad_norm": 0.800520122051239, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.292, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06674453529117304, | |
| "eval_loss": 0.36322954297065735, | |
| "eval_runtime": 344.1595, | |
| "eval_samples_per_second": 7.334, | |
| "eval_steps_per_second": 1.833, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.06741198064408477, | |
| "grad_norm": 0.4408794939517975, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.3187, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.0680794259969965, | |
| "grad_norm": 0.4651925265789032, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.3673, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.06874687134990823, | |
| "grad_norm": 0.5171566605567932, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.388, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.06941431670281996, | |
| "grad_norm": 0.5095441341400146, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.3783, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.07008176205573169, | |
| "grad_norm": 0.4939165413379669, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3746, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.07074920740864342, | |
| "grad_norm": 0.47254830598831177, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.3666, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.07141665276155515, | |
| "grad_norm": 0.5139265060424805, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.4048, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.07208409811446688, | |
| "grad_norm": 0.48584040999412537, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.37, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.07275154346737861, | |
| "grad_norm": 0.47279784083366394, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.3398, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.07341898882029034, | |
| "grad_norm": 0.5263993740081787, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.3838, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.07408643417320207, | |
| "grad_norm": 0.50913405418396, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.3555, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.0747538795261138, | |
| "grad_norm": 0.5122456550598145, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.3617, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.07542132487902553, | |
| "grad_norm": 0.5421679019927979, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.3563, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.07608877023193726, | |
| "grad_norm": 0.5710113644599915, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.3521, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.07675621558484899, | |
| "grad_norm": 0.6738095879554749, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.3588, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.07742366093776072, | |
| "grad_norm": 0.6016878485679626, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.3849, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.07809110629067245, | |
| "grad_norm": 0.5662630796432495, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.336, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.07875855164358418, | |
| "grad_norm": 0.5533269643783569, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.3504, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.07942599699649591, | |
| "grad_norm": 0.5861207246780396, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.3742, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.08009344234940764, | |
| "grad_norm": 0.6072835326194763, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.3469, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08076088770231937, | |
| "grad_norm": 0.5949785709381104, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.359, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.0814283330552311, | |
| "grad_norm": 0.5673805475234985, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.3714, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.08209577840814283, | |
| "grad_norm": 0.6179041266441345, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.3757, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.08276322376105456, | |
| "grad_norm": 0.5998288989067078, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.3688, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.0834306691139663, | |
| "grad_norm": 0.5819981098175049, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.3156, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.08409811446687802, | |
| "grad_norm": 0.6107437610626221, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.3381, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.08476555981978975, | |
| "grad_norm": 0.6253097653388977, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.3539, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.08543300517270148, | |
| "grad_norm": 0.5656948685646057, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.3233, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.08610045052561321, | |
| "grad_norm": 0.6172930598258972, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.3304, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.08676789587852494, | |
| "grad_norm": 0.6856399774551392, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.3715, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.08743534123143668, | |
| "grad_norm": 0.6139941811561584, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.2952, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.0881027865843484, | |
| "grad_norm": 0.7134049534797668, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.3904, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.08877023193726014, | |
| "grad_norm": 0.6316084861755371, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.2888, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.08943767729017187, | |
| "grad_norm": 0.6332045197486877, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.3344, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.0901051226430836, | |
| "grad_norm": 0.6992444396018982, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.3323, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.09077256799599533, | |
| "grad_norm": 0.6735237240791321, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.3158, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.09144001334890706, | |
| "grad_norm": 0.6418128609657288, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.2943, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.09210745870181879, | |
| "grad_norm": 0.6648815274238586, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.3328, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.09277490405473052, | |
| "grad_norm": 0.7809527516365051, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.3615, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.09344234940764225, | |
| "grad_norm": 0.7083166241645813, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.3637, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.09410979476055398, | |
| "grad_norm": 0.7833213210105896, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.3493, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.09477724011346571, | |
| "grad_norm": 0.6797055602073669, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3815, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.09544468546637744, | |
| "grad_norm": 0.6998314261436462, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3814, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.09611213081928917, | |
| "grad_norm": 0.5843608379364014, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.345, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.0967795761722009, | |
| "grad_norm": 0.6214026212692261, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.2986, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.09744702152511263, | |
| "grad_norm": 0.6351425051689148, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.3294, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.09811446687802436, | |
| "grad_norm": 0.6896306276321411, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.3331, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.09878191223093609, | |
| "grad_norm": 0.6924558877944946, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.3495, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.09944935758384782, | |
| "grad_norm": 0.6954474449157715, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.2826, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.10011680293675955, | |
| "grad_norm": 0.7600224018096924, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.2769, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10011680293675955, | |
| "eval_loss": 0.34324362874031067, | |
| "eval_runtime": 344.312, | |
| "eval_samples_per_second": 7.331, | |
| "eval_steps_per_second": 1.833, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.10078424828967128, | |
| "grad_norm": 0.8608183860778809, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.3091, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.10145169364258301, | |
| "grad_norm": 0.42263585329055786, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.358, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.10211913899549474, | |
| "grad_norm": 0.4742768108844757, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.4032, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.10278658434840647, | |
| "grad_norm": 0.5449939966201782, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.4052, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.1034540297013182, | |
| "grad_norm": 0.5281736850738525, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.3445, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.10412147505422993, | |
| "grad_norm": 0.45331403613090515, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.3432, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.10478892040714166, | |
| "grad_norm": 0.5306539535522461, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.395, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.10545636576005339, | |
| "grad_norm": 0.5132590532302856, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.3891, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.10612381111296512, | |
| "grad_norm": 0.5419913530349731, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.3931, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.10679125646587685, | |
| "grad_norm": 0.4792773723602295, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.3441, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.10745870181878858, | |
| "grad_norm": 0.48551034927368164, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.3566, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.10812614717170031, | |
| "grad_norm": 0.5094127058982849, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.3794, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.10879359252461204, | |
| "grad_norm": 0.511148989200592, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.3726, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.10946103787752377, | |
| "grad_norm": 0.47986695170402527, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.3284, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.1101284832304355, | |
| "grad_norm": 0.5611073970794678, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.3388, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.11079592858334723, | |
| "grad_norm": 0.4874250292778015, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.3511, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.11146337393625896, | |
| "grad_norm": 0.5071490406990051, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.3208, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.1121308192891707, | |
| "grad_norm": 0.5272539854049683, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.3499, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.11279826464208242, | |
| "grad_norm": 0.5543425679206848, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.3633, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.11346570999499415, | |
| "grad_norm": 0.5473048090934753, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.3683, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.11413315534790588, | |
| "grad_norm": 0.5861783623695374, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.3467, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.11480060070081761, | |
| "grad_norm": 0.5844394564628601, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.3234, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.11546804605372935, | |
| "grad_norm": 0.6197385191917419, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.3568, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.11613549140664108, | |
| "grad_norm": 0.6510120630264282, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.356, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.1168029367595528, | |
| "grad_norm": 0.5716838240623474, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.3402, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.11747038211246454, | |
| "grad_norm": 0.5871708393096924, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.3399, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.11813782746537627, | |
| "grad_norm": 0.6544736623764038, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.3745, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.118805272818288, | |
| "grad_norm": 0.695075273513794, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.3232, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.11947271817119973, | |
| "grad_norm": 0.6274528503417969, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.3146, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.12014016352411146, | |
| "grad_norm": 0.6290251612663269, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.3027, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12080760887702319, | |
| "grad_norm": 0.7067568302154541, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.3186, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.12147505422993492, | |
| "grad_norm": 0.6486525535583496, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.3225, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.12214249958284665, | |
| "grad_norm": 0.6556540727615356, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.2597, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.12280994493575839, | |
| "grad_norm": 0.6495266556739807, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.3142, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.12347739028867012, | |
| "grad_norm": 0.6407210826873779, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3371, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.12414483564158185, | |
| "grad_norm": 0.6632188558578491, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.3023, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.12481228099449358, | |
| "grad_norm": 0.6833558082580566, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.3353, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.1254797263474053, | |
| "grad_norm": 0.6944228410720825, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.287, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.12614717170031703, | |
| "grad_norm": 0.704785943031311, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3246, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.12681461705322877, | |
| "grad_norm": 0.6699495911598206, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.3084, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.1274820624061405, | |
| "grad_norm": 0.6259564757347107, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.2979, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.12814950775905223, | |
| "grad_norm": 0.6411551833152771, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.3363, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.12881695311196395, | |
| "grad_norm": 0.6074274182319641, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3065, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.1294843984648757, | |
| "grad_norm": 0.6387959122657776, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.332, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.1301518438177874, | |
| "grad_norm": 0.6455732583999634, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.331, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.13081928917069915, | |
| "grad_norm": 0.6572890877723694, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.3082, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.13148673452361087, | |
| "grad_norm": 0.7414137721061707, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3649, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.13215417987652262, | |
| "grad_norm": 0.7173470854759216, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.3285, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.13282162522943433, | |
| "grad_norm": 0.7355753183364868, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.3157, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.13348907058234608, | |
| "grad_norm": 0.7989836931228638, | |
| "learning_rate": 0.0, | |
| "loss": 0.2944, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.13348907058234608, | |
| "eval_loss": 0.3386111855506897, | |
| "eval_runtime": 344.2109, | |
| "eval_samples_per_second": 7.333, | |
| "eval_steps_per_second": 1.833, | |
| "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": 5.255343145494774e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |