| { | |
| "best_metric": 10.339560508728027, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 2.867383512544803, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.014336917562724014, | |
| "grad_norm": 0.08282607793807983, | |
| "learning_rate": 1e-05, | |
| "loss": 10.3782, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.014336917562724014, | |
| "eval_loss": 10.371087074279785, | |
| "eval_runtime": 0.2627, | |
| "eval_samples_per_second": 449.138, | |
| "eval_steps_per_second": 114.188, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02867383512544803, | |
| "grad_norm": 0.0957234650850296, | |
| "learning_rate": 2e-05, | |
| "loss": 10.379, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.043010752688172046, | |
| "grad_norm": 0.10738931596279144, | |
| "learning_rate": 3e-05, | |
| "loss": 10.3775, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05734767025089606, | |
| "grad_norm": 0.09554025530815125, | |
| "learning_rate": 4e-05, | |
| "loss": 10.3735, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.07168458781362007, | |
| "grad_norm": 0.08328013122081757, | |
| "learning_rate": 5e-05, | |
| "loss": 10.3751, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.08602150537634409, | |
| "grad_norm": 0.09642637521028519, | |
| "learning_rate": 6e-05, | |
| "loss": 10.3778, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.1003584229390681, | |
| "grad_norm": 0.09173403680324554, | |
| "learning_rate": 7e-05, | |
| "loss": 10.3731, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.11469534050179211, | |
| "grad_norm": 0.13879632949829102, | |
| "learning_rate": 8e-05, | |
| "loss": 10.3711, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.12903225806451613, | |
| "grad_norm": 0.10093166679143906, | |
| "learning_rate": 9e-05, | |
| "loss": 10.37, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.14336917562724014, | |
| "grad_norm": 0.12951916456222534, | |
| "learning_rate": 0.0001, | |
| "loss": 10.3716, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.15770609318996415, | |
| "grad_norm": 0.11753611266613007, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 10.3818, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.17204301075268819, | |
| "grad_norm": 0.11250785738229752, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 10.3699, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.1863799283154122, | |
| "grad_norm": 0.13701201975345612, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 10.3641, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.2007168458781362, | |
| "grad_norm": 0.13668373227119446, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 10.3775, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.21505376344086022, | |
| "grad_norm": 0.13729792833328247, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 10.3554, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.22939068100358423, | |
| "grad_norm": 0.1807527244091034, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 10.3553, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.24372759856630824, | |
| "grad_norm": 0.2271457463502884, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 10.3534, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.25806451612903225, | |
| "grad_norm": 0.1049678847193718, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 10.3753, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.2724014336917563, | |
| "grad_norm": 0.15400785207748413, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 10.3708, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.2867383512544803, | |
| "grad_norm": 0.17220549285411835, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 10.3722, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.3010752688172043, | |
| "grad_norm": 0.11970847100019455, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 10.3667, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.3154121863799283, | |
| "grad_norm": 0.1458829641342163, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 10.3668, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.32974910394265233, | |
| "grad_norm": 0.1512281894683838, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 10.3686, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.34408602150537637, | |
| "grad_norm": 0.17232315242290497, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 10.3667, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.35842293906810035, | |
| "grad_norm": 0.20148640871047974, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 10.371, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.3727598566308244, | |
| "grad_norm": 0.2048002928495407, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 10.3693, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.3870967741935484, | |
| "grad_norm": 0.20468057692050934, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 10.3645, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.4014336917562724, | |
| "grad_norm": 0.1436544507741928, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 10.3537, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.4157706093189964, | |
| "grad_norm": 0.22871632874011993, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 10.3525, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.43010752688172044, | |
| "grad_norm": 0.29437798261642456, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 10.3614, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.2697261571884155, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 10.3437, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.45878136200716846, | |
| "grad_norm": 0.246367409825325, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 10.3493, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.4731182795698925, | |
| "grad_norm": 0.2847760021686554, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 10.3421, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.4874551971326165, | |
| "grad_norm": 0.4077546000480652, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 10.3345, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.5017921146953405, | |
| "grad_norm": 0.2409476637840271, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 10.3692, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.5161290322580645, | |
| "grad_norm": 0.20892679691314697, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 10.3658, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.5304659498207885, | |
| "grad_norm": 0.2638150751590729, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 10.3582, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.5448028673835126, | |
| "grad_norm": 0.16109246015548706, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 10.3589, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5591397849462365, | |
| "grad_norm": 0.23983745276927948, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 10.3541, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.5734767025089605, | |
| "grad_norm": 0.20890073478221893, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 10.3587, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.5878136200716846, | |
| "grad_norm": 0.24057435989379883, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 10.3518, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.6021505376344086, | |
| "grad_norm": 0.1949206441640854, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 10.3515, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.6164874551971327, | |
| "grad_norm": 0.21265587210655212, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 10.3506, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.6308243727598566, | |
| "grad_norm": 0.2160329818725586, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 10.3557, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.6451612903225806, | |
| "grad_norm": 0.202266126871109, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 10.3502, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.6594982078853047, | |
| "grad_norm": 0.27038994431495667, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 10.3537, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6738351254480287, | |
| "grad_norm": 0.19615745544433594, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 10.347, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.6881720430107527, | |
| "grad_norm": 0.28301727771759033, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 10.3514, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.7025089605734767, | |
| "grad_norm": 0.37925800681114197, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 10.3419, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.7168458781362007, | |
| "grad_norm": 0.3209330141544342, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 10.349, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7168458781362007, | |
| "eval_loss": 10.347315788269043, | |
| "eval_runtime": 0.2704, | |
| "eval_samples_per_second": 436.347, | |
| "eval_steps_per_second": 110.936, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.7311827956989247, | |
| "grad_norm": 0.3832900822162628, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 10.3353, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.7455197132616488, | |
| "grad_norm": 0.20099468529224396, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 10.3604, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.7598566308243727, | |
| "grad_norm": 0.24598540365695953, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 10.3491, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.7741935483870968, | |
| "grad_norm": 0.15631051361560822, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 10.3555, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.7885304659498208, | |
| "grad_norm": 0.14090418815612793, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 10.3504, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.8028673835125448, | |
| "grad_norm": 0.13437707722187042, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 10.3459, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.8172043010752689, | |
| "grad_norm": 0.1333155781030655, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 10.3524, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.8315412186379928, | |
| "grad_norm": 0.1498744636774063, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 10.3492, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.8458781362007168, | |
| "grad_norm": 0.17131388187408447, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 10.3469, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.8602150537634409, | |
| "grad_norm": 0.17683598399162292, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 10.3416, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.8745519713261649, | |
| "grad_norm": 0.21371674537658691, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 10.3518, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 0.15684473514556885, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 10.3461, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.9032258064516129, | |
| "grad_norm": 0.21612265706062317, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 10.3446, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.9175627240143369, | |
| "grad_norm": 0.14588145911693573, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 10.3342, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.931899641577061, | |
| "grad_norm": 0.25889652967453003, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 10.3497, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.946236559139785, | |
| "grad_norm": 0.23751391470432281, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 10.3378, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.9605734767025089, | |
| "grad_norm": 0.22873426973819733, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 10.3358, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.974910394265233, | |
| "grad_norm": 0.41565361618995667, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 10.3327, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.989247311827957, | |
| "grad_norm": 0.1489182412624359, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 10.3516, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 1.003584229390681, | |
| "grad_norm": 0.23786625266075134, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 11.3016, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 1.017921146953405, | |
| "grad_norm": 0.28726816177368164, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 12.7302, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 1.032258064516129, | |
| "grad_norm": 0.19457994401454926, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 9.471, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 1.0465949820788532, | |
| "grad_norm": 0.17783616483211517, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 9.9642, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 1.060931899641577, | |
| "grad_norm": 0.2229432314634323, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 11.005, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 1.075268817204301, | |
| "grad_norm": 0.1576322615146637, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 10.4776, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 1.0896057347670252, | |
| "grad_norm": 0.1837427020072937, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 10.7237, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 1.103942652329749, | |
| "grad_norm": 0.14871269464492798, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 10.2532, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 1.118279569892473, | |
| "grad_norm": 0.19259674847126007, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 10.0265, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 1.1326164874551972, | |
| "grad_norm": 0.2265091836452484, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 10.3781, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 1.146953405017921, | |
| "grad_norm": 0.19976967573165894, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 10.4453, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.1612903225806452, | |
| "grad_norm": 0.16976790130138397, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 10.3426, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 1.1756272401433692, | |
| "grad_norm": 0.1536082774400711, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 9.6291, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 1.189964157706093, | |
| "grad_norm": 0.17330212891101837, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 11.1494, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 1.2043010752688172, | |
| "grad_norm": 0.26698726415634155, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 10.0345, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 1.2186379928315412, | |
| "grad_norm": 0.22625815868377686, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 10.2258, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 1.2329749103942653, | |
| "grad_norm": 0.2806665897369385, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 10.6632, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 1.2473118279569892, | |
| "grad_norm": 0.2715948820114136, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 8.4377, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 1.2616487455197132, | |
| "grad_norm": 0.22190995514392853, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 11.7135, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 1.2759856630824373, | |
| "grad_norm": 0.1984441578388214, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 10.4027, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 1.2903225806451613, | |
| "grad_norm": 0.2430347055196762, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 10.8187, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 1.3046594982078852, | |
| "grad_norm": 0.13875645399093628, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 9.5375, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 1.3189964157706093, | |
| "grad_norm": 0.16168849170207977, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 10.3123, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.13042473793029785, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 10.8341, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.3476702508960574, | |
| "grad_norm": 0.14306622743606567, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 10.2414, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.3620071684587813, | |
| "grad_norm": 0.1657639443874359, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 10.4378, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.3763440860215055, | |
| "grad_norm": 0.18026864528656006, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 9.7768, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.3906810035842294, | |
| "grad_norm": 0.20865190029144287, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 10.9124, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.4050179211469533, | |
| "grad_norm": 0.1652052104473114, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 10.0559, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.4193548387096775, | |
| "grad_norm": 0.14710059762001038, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 10.4244, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.4336917562724014, | |
| "grad_norm": 0.20172981917858124, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 10.6797, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.4336917562724014, | |
| "eval_loss": 10.341728210449219, | |
| "eval_runtime": 0.2645, | |
| "eval_samples_per_second": 446.051, | |
| "eval_steps_per_second": 113.403, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.4480286738351253, | |
| "grad_norm": 0.30038508772850037, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 10.3015, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.4623655913978495, | |
| "grad_norm": 0.18767149746418, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 10.3658, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.4767025089605734, | |
| "grad_norm": 0.2205027937889099, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 10.7814, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.4910394265232976, | |
| "grad_norm": 0.2686139941215515, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 8.3392, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.5053763440860215, | |
| "grad_norm": 0.22885169088840485, | |
| "learning_rate": 5e-05, | |
| "loss": 11.8265, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.5197132616487457, | |
| "grad_norm": 0.2417667657136917, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 10.4914, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.5340501792114696, | |
| "grad_norm": 0.18876677751541138, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 10.5046, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.5483870967741935, | |
| "grad_norm": 0.15323854982852936, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 9.4612, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.5627240143369177, | |
| "grad_norm": 0.20265346765518188, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 10.8386, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.5770609318996416, | |
| "grad_norm": 0.13983668386936188, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 10.775, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.5913978494623655, | |
| "grad_norm": 0.11668044328689575, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 10.2212, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.6057347670250897, | |
| "grad_norm": 0.14640139043331146, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 10.6441, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.6200716845878136, | |
| "grad_norm": 0.1578124761581421, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 9.6487, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.6344086021505375, | |
| "grad_norm": 0.17161884903907776, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 11.3263, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.6487455197132617, | |
| "grad_norm": 0.1484452784061432, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 9.0846, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.6630824372759858, | |
| "grad_norm": 0.1627979278564453, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 10.82, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.6774193548387095, | |
| "grad_norm": 0.19332915544509888, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 10.6226, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.6917562724014337, | |
| "grad_norm": 0.2519383132457733, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 10.1521, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.7060931899641578, | |
| "grad_norm": 0.15560027956962585, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 10.4671, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.7204301075268817, | |
| "grad_norm": 0.2606619894504547, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 10.092, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.7347670250896057, | |
| "grad_norm": 0.2495698481798172, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 9.5014, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.7491039426523298, | |
| "grad_norm": 0.2609870433807373, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 11.2544, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.7634408602150538, | |
| "grad_norm": 0.2345278114080429, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 10.3839, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.16896145045757294, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 10.2853, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.7921146953405018, | |
| "grad_norm": 0.15526142716407776, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 10.4349, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.8064516129032258, | |
| "grad_norm": 0.13907863199710846, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 9.7537, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.8207885304659497, | |
| "grad_norm": 0.1379098892211914, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 11.6596, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.8351254480286738, | |
| "grad_norm": 0.1726650893688202, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 8.8195, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.849462365591398, | |
| "grad_norm": 0.19816069304943085, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 9.7752, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.863799283154122, | |
| "grad_norm": 0.2384025752544403, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 11.4723, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.8781362007168458, | |
| "grad_norm": 0.19121915102005005, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 10.116, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.89247311827957, | |
| "grad_norm": 0.202835351228714, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 10.3309, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.906810035842294, | |
| "grad_norm": 0.15531320869922638, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 11.5329, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.9211469534050178, | |
| "grad_norm": 0.17083214223384857, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 9.4032, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.935483870967742, | |
| "grad_norm": 0.23794014751911163, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 10.791, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.949820788530466, | |
| "grad_norm": 0.18013781309127808, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 9.6148, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.9641577060931898, | |
| "grad_norm": 0.23243537545204163, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 11.0864, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.978494623655914, | |
| "grad_norm": 0.3368525803089142, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 11.3936, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.9928315412186381, | |
| "grad_norm": 0.16337698698043823, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 10.2335, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 2.007168458781362, | |
| "grad_norm": 0.1126953512430191, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 10.3546, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.021505376344086, | |
| "grad_norm": 0.29826295375823975, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 11.8035, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 2.03584229390681, | |
| "grad_norm": 0.20435789227485657, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 10.1002, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 2.050179211469534, | |
| "grad_norm": 0.2144787609577179, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 10.2434, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 2.064516129032258, | |
| "grad_norm": 0.09925316274166107, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 10.3356, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 2.078853046594982, | |
| "grad_norm": 0.13362076878547668, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 10.249, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 2.0931899641577063, | |
| "grad_norm": 0.1269536316394806, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 10.9884, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 2.10752688172043, | |
| "grad_norm": 0.15948514640331268, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 10.475, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 2.121863799283154, | |
| "grad_norm": 0.17508427798748016, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 9.9104, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 2.1362007168458783, | |
| "grad_norm": 0.1521162986755371, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 10.2029, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 2.150537634408602, | |
| "grad_norm": 0.21580275893211365, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 10.1022, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.150537634408602, | |
| "eval_loss": 10.339735984802246, | |
| "eval_runtime": 0.2628, | |
| "eval_samples_per_second": 449.029, | |
| "eval_steps_per_second": 114.16, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 2.164874551971326, | |
| "grad_norm": 0.1470649093389511, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 10.353, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 2.1792114695340503, | |
| "grad_norm": 0.15960943698883057, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 10.8421, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 2.193548387096774, | |
| "grad_norm": 0.15838678181171417, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 10.8225, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 2.207885304659498, | |
| "grad_norm": 0.24349571764469147, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 9.4784, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 2.2222222222222223, | |
| "grad_norm": 0.18625831604003906, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 9.8724, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 2.236559139784946, | |
| "grad_norm": 0.22831754386425018, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 11.1881, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 2.25089605734767, | |
| "grad_norm": 0.17270688712596893, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 7.1167, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 2.2652329749103943, | |
| "grad_norm": 0.23757074773311615, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 13.8299, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 2.279569892473118, | |
| "grad_norm": 0.21444980800151825, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 9.8176, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 2.293906810035842, | |
| "grad_norm": 0.1668054312467575, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 9.9919, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.3082437275985663, | |
| "grad_norm": 0.17417311668395996, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 10.0724, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 2.3225806451612905, | |
| "grad_norm": 0.13801367580890656, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 10.5747, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 2.336917562724014, | |
| "grad_norm": 0.17021505534648895, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 11.2412, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 2.3512544802867383, | |
| "grad_norm": 0.16260457038879395, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 8.9377, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 2.3655913978494625, | |
| "grad_norm": 0.24675720930099487, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 10.7951, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 2.379928315412186, | |
| "grad_norm": 0.23776426911354065, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 11.2508, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 2.3942652329749103, | |
| "grad_norm": 0.19111722707748413, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 9.24, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 2.4086021505376345, | |
| "grad_norm": 0.20960883796215057, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 10.989, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 2.4229390681003586, | |
| "grad_norm": 0.07739323377609253, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 10.2063, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 2.4372759856630823, | |
| "grad_norm": 0.25707653164863586, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 10.3645, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 2.4516129032258065, | |
| "grad_norm": 0.20341019332408905, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 10.6096, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 2.4659498207885306, | |
| "grad_norm": 0.22398626804351807, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 10.1366, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 2.4802867383512543, | |
| "grad_norm": 0.3793470561504364, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 9.6642, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 2.4946236559139785, | |
| "grad_norm": 0.1528487354516983, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 10.5233, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 2.5089605734767026, | |
| "grad_norm": 0.2509865164756775, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 10.4963, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 2.5232974910394264, | |
| "grad_norm": 0.20558923482894897, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 10.7993, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 2.5376344086021505, | |
| "grad_norm": 0.16977569460868835, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 10.2529, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 2.5519713261648747, | |
| "grad_norm": 0.18404771387577057, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 10.9373, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 2.5663082437275984, | |
| "grad_norm": 0.1456303596496582, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 9.747, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 2.5806451612903225, | |
| "grad_norm": 0.16044515371322632, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 11.1105, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.5949820788530467, | |
| "grad_norm": 0.10143168270587921, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 8.5955, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 2.6093189964157704, | |
| "grad_norm": 0.22769451141357422, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 11.6662, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 2.6236559139784945, | |
| "grad_norm": 0.1898529827594757, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 9.2088, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 2.6379928315412187, | |
| "grad_norm": 0.13910135626792908, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 10.9813, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.652329749103943, | |
| "grad_norm": 0.20146462321281433, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 10.0841, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.6666666666666665, | |
| "grad_norm": 0.14200545847415924, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 10.5609, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.6810035842293907, | |
| "grad_norm": 0.16796408593654633, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 10.3375, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.695340501792115, | |
| "grad_norm": 0.22269295156002045, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 10.5795, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.709677419354839, | |
| "grad_norm": 0.17484870553016663, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 9.7547, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.7240143369175627, | |
| "grad_norm": 0.2725757360458374, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 11.6649, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.738351254480287, | |
| "grad_norm": 0.21425926685333252, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 8.5501, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.752688172043011, | |
| "grad_norm": 0.18418702483177185, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 11.2186, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.7670250896057347, | |
| "grad_norm": 0.1718948483467102, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 10.2777, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.781362007168459, | |
| "grad_norm": 0.1467687338590622, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 9.606, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.795698924731183, | |
| "grad_norm": 0.16910839080810547, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 10.9749, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.8100358422939067, | |
| "grad_norm": 0.14424005150794983, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 10.3218, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.824372759856631, | |
| "grad_norm": 0.15643349289894104, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 11.1777, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.838709677419355, | |
| "grad_norm": 0.164554625749588, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 9.7567, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.8530465949820787, | |
| "grad_norm": 0.17041726410388947, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 10.8188, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.867383512544803, | |
| "grad_norm": 0.15139539539813995, | |
| "learning_rate": 0.0, | |
| "loss": 9.8542, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.867383512544803, | |
| "eval_loss": 10.339560508728027, | |
| "eval_runtime": 0.2658, | |
| "eval_samples_per_second": 443.941, | |
| "eval_steps_per_second": 112.866, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 200, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 22310132121600.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |