| { | |
| "best_metric": 10.287935256958008, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.8565310492505354, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004282655246252677, | |
| "grad_norm": 1.0140916109085083, | |
| "learning_rate": 1e-05, | |
| "loss": 41.5163, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.004282655246252677, | |
| "eval_loss": 10.378774642944336, | |
| "eval_runtime": 0.833, | |
| "eval_samples_per_second": 472.969, | |
| "eval_steps_per_second": 118.842, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008565310492505354, | |
| "grad_norm": 0.9732224345207214, | |
| "learning_rate": 2e-05, | |
| "loss": 41.5317, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.01284796573875803, | |
| "grad_norm": 0.9495458006858826, | |
| "learning_rate": 3e-05, | |
| "loss": 41.529, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.017130620985010708, | |
| "grad_norm": 1.0050394535064697, | |
| "learning_rate": 4e-05, | |
| "loss": 41.527, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.021413276231263382, | |
| "grad_norm": 0.8522498607635498, | |
| "learning_rate": 5e-05, | |
| "loss": 41.5381, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.02569593147751606, | |
| "grad_norm": 0.7451391220092773, | |
| "learning_rate": 6e-05, | |
| "loss": 41.5266, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.029978586723768737, | |
| "grad_norm": 0.9513540267944336, | |
| "learning_rate": 7e-05, | |
| "loss": 41.5287, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.034261241970021415, | |
| "grad_norm": 0.809018611907959, | |
| "learning_rate": 8e-05, | |
| "loss": 41.5078, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03854389721627409, | |
| "grad_norm": 0.9694377183914185, | |
| "learning_rate": 9e-05, | |
| "loss": 41.5079, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.042826552462526764, | |
| "grad_norm": 0.9219012260437012, | |
| "learning_rate": 0.0001, | |
| "loss": 41.5011, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.047109207708779445, | |
| "grad_norm": 0.9172611832618713, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 41.4916, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05139186295503212, | |
| "grad_norm": 0.9734646677970886, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 41.4979, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.055674518201284794, | |
| "grad_norm": 1.1014163494110107, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 41.4956, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.059957173447537475, | |
| "grad_norm": 1.0327798128128052, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 41.4725, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06423982869379015, | |
| "grad_norm": 1.2098652124404907, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 41.4752, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.06852248394004283, | |
| "grad_norm": 1.0268454551696777, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 41.4617, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.0728051391862955, | |
| "grad_norm": 1.2437031269073486, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 41.4602, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07708779443254818, | |
| "grad_norm": 1.2205047607421875, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 41.4568, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08137044967880086, | |
| "grad_norm": 1.4317915439605713, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 41.4491, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08565310492505353, | |
| "grad_norm": 1.5389832258224487, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 41.4467, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.08993576017130621, | |
| "grad_norm": 1.5515321493148804, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 41.4331, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.09421841541755889, | |
| "grad_norm": 1.739243507385254, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 41.4354, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.09850107066381156, | |
| "grad_norm": 1.4590630531311035, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 41.4432, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.10278372591006424, | |
| "grad_norm": 2.056377649307251, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 41.3832, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.10706638115631692, | |
| "grad_norm": 1.8177523612976074, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 41.393, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11134903640256959, | |
| "grad_norm": 1.6252126693725586, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 41.3884, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.11563169164882227, | |
| "grad_norm": 1.8270189762115479, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 41.3723, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.11991434689507495, | |
| "grad_norm": 1.8442226648330688, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 41.3694, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12419700214132762, | |
| "grad_norm": 1.9399042129516602, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 41.358, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.1284796573875803, | |
| "grad_norm": 1.9133156538009644, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 41.3635, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13276231263383298, | |
| "grad_norm": 1.9715404510498047, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 41.314, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.13704496788008566, | |
| "grad_norm": 2.019277572631836, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 41.3179, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.14132762312633834, | |
| "grad_norm": 1.889991044998169, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 41.3112, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.145610278372591, | |
| "grad_norm": 1.585730791091919, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 41.3056, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.14989293361884368, | |
| "grad_norm": 1.5292245149612427, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 41.2858, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.15417558886509636, | |
| "grad_norm": 1.7086025476455688, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 41.2667, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.15845824411134904, | |
| "grad_norm": 1.6152453422546387, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 41.2465, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16274089935760172, | |
| "grad_norm": 1.4000519514083862, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 41.2462, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1670235546038544, | |
| "grad_norm": 1.553257703781128, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 41.2412, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.17130620985010706, | |
| "grad_norm": 1.296061396598816, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 41.2366, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.17558886509635974, | |
| "grad_norm": 1.3130500316619873, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 41.239, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.17987152034261242, | |
| "grad_norm": 1.3968032598495483, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 41.202, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1841541755888651, | |
| "grad_norm": 1.1028584241867065, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 41.2066, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.18843683083511778, | |
| "grad_norm": 1.223429560661316, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 41.2302, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.19271948608137046, | |
| "grad_norm": 1.0133888721466064, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 41.2274, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.19700214132762311, | |
| "grad_norm": 1.038590908050537, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 41.1885, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.2012847965738758, | |
| "grad_norm": 1.2082979679107666, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 41.1656, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.20556745182012848, | |
| "grad_norm": 1.0701316595077515, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 41.1831, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.20985010706638116, | |
| "grad_norm": 0.8915174007415771, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 41.1721, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.21413276231263384, | |
| "grad_norm": 0.9999569058418274, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 41.1658, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.21413276231263384, | |
| "eval_loss": 10.300396919250488, | |
| "eval_runtime": 0.8194, | |
| "eval_samples_per_second": 480.82, | |
| "eval_steps_per_second": 120.815, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.21841541755888652, | |
| "grad_norm": 1.213214635848999, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 41.2533, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.22269807280513917, | |
| "grad_norm": 0.897641122341156, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 41.2424, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.22698072805139186, | |
| "grad_norm": 0.9352962970733643, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 41.2243, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.23126338329764454, | |
| "grad_norm": 0.7871758341789246, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 41.2395, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.23554603854389722, | |
| "grad_norm": 0.8345593810081482, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 41.2352, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.2398286937901499, | |
| "grad_norm": 0.7352098822593689, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 41.2216, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.24411134903640258, | |
| "grad_norm": 0.7078768610954285, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 41.2238, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.24839400428265523, | |
| "grad_norm": 0.6202782988548279, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 41.2145, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.25267665952890794, | |
| "grad_norm": 0.6747980713844299, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 41.2165, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2569593147751606, | |
| "grad_norm": 0.6640037298202515, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 41.2129, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.26124197002141325, | |
| "grad_norm": 0.5419409871101379, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 41.1924, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.26552462526766596, | |
| "grad_norm": 0.5575965046882629, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 41.1859, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.2698072805139186, | |
| "grad_norm": 0.3973611891269684, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 41.2037, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.2740899357601713, | |
| "grad_norm": 0.49591419100761414, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 41.1752, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.278372591006424, | |
| "grad_norm": 0.5320420861244202, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 41.1887, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.2826552462526767, | |
| "grad_norm": 0.40315917134284973, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 41.1731, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.28693790149892934, | |
| "grad_norm": 0.4538324475288391, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 41.1811, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.291220556745182, | |
| "grad_norm": 0.3954373598098755, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 41.1942, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2955032119914347, | |
| "grad_norm": 0.5449280738830566, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 41.1376, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.29978586723768735, | |
| "grad_norm": 0.37435728311538696, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 41.2013, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.30406852248394006, | |
| "grad_norm": 0.415919691324234, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 41.1889, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3083511777301927, | |
| "grad_norm": 0.3450535237789154, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 41.1531, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.31263383297644537, | |
| "grad_norm": 0.44574135541915894, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 41.192, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.3169164882226981, | |
| "grad_norm": 0.3757171034812927, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 41.1906, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.32119914346895073, | |
| "grad_norm": 0.4182075262069702, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 41.1905, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.32548179871520344, | |
| "grad_norm": 0.42849573493003845, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 41.1425, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.3297644539614561, | |
| "grad_norm": 0.4211777150630951, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 41.1518, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3340471092077088, | |
| "grad_norm": 0.3975631594657898, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 41.1944, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.33832976445396146, | |
| "grad_norm": 0.3846378028392792, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 41.1686, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3426124197002141, | |
| "grad_norm": 0.37172695994377136, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 41.1602, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.3468950749464668, | |
| "grad_norm": 0.43355971574783325, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 41.1777, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.3511777301927195, | |
| "grad_norm": 0.4843752086162567, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 41.1754, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.3554603854389722, | |
| "grad_norm": 0.44365033507347107, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 41.1627, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.35974304068522484, | |
| "grad_norm": 0.4922822415828705, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 41.1995, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.3640256959314775, | |
| "grad_norm": 0.5284846425056458, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 41.1786, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3683083511777302, | |
| "grad_norm": 0.6352524161338806, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 41.1377, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.37259100642398285, | |
| "grad_norm": 0.4986525774002075, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 41.1724, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.37687366167023556, | |
| "grad_norm": 0.5726274847984314, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 41.145, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3811563169164882, | |
| "grad_norm": 0.4305708110332489, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 41.1494, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3854389721627409, | |
| "grad_norm": 0.47623881697654724, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 41.1446, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3897216274089936, | |
| "grad_norm": 0.517105221748352, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 41.1379, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.39400428265524623, | |
| "grad_norm": 0.7964802384376526, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 41.1429, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.39828693790149894, | |
| "grad_norm": 0.45318299531936646, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 41.1773, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.4025695931477516, | |
| "grad_norm": 0.48306235671043396, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 41.1437, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.4068522483940043, | |
| "grad_norm": 0.6395437717437744, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 41.1413, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.41113490364025695, | |
| "grad_norm": 0.5044965744018555, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 41.1423, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.41541755888650966, | |
| "grad_norm": 0.5140703916549683, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 41.1529, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.4197002141327623, | |
| "grad_norm": 0.7176982164382935, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 41.1448, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.42398286937901497, | |
| "grad_norm": 0.6121039986610413, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 41.154, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4282655246252677, | |
| "grad_norm": 1.0461997985839844, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 41.1598, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4282655246252677, | |
| "eval_loss": 10.292905807495117, | |
| "eval_runtime": 0.8308, | |
| "eval_samples_per_second": 474.222, | |
| "eval_steps_per_second": 119.157, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.43254817987152033, | |
| "grad_norm": 1.1328588724136353, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 41.2356, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.43683083511777304, | |
| "grad_norm": 0.8626485466957092, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 41.2255, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4411134903640257, | |
| "grad_norm": 0.5959879755973816, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 41.1977, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.44539614561027835, | |
| "grad_norm": 0.6696794033050537, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 41.2135, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.44967880085653106, | |
| "grad_norm": 0.7611036896705627, | |
| "learning_rate": 5e-05, | |
| "loss": 41.1939, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.4539614561027837, | |
| "grad_norm": 0.8138741254806519, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 41.2055, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4582441113490364, | |
| "grad_norm": 0.5165987610816956, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 41.1803, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.4625267665952891, | |
| "grad_norm": 0.6895624995231628, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 41.2141, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.4668094218415418, | |
| "grad_norm": 0.5425256490707397, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 41.1917, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.47109207708779444, | |
| "grad_norm": 0.4485233426094055, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 41.1746, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4753747323340471, | |
| "grad_norm": 0.5313943028450012, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 41.1849, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.4796573875802998, | |
| "grad_norm": 0.6793656945228577, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 41.1816, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.48394004282655245, | |
| "grad_norm": 0.5230658054351807, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 41.1465, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.48822269807280516, | |
| "grad_norm": 0.39127117395401, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 41.1744, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.4925053533190578, | |
| "grad_norm": 0.45096760988235474, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 41.1803, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.49678800856531047, | |
| "grad_norm": 0.3252319097518921, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 41.1872, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5010706638115632, | |
| "grad_norm": 0.43201541900634766, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 41.1486, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5053533190578159, | |
| "grad_norm": 0.42343848943710327, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 41.1491, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5096359743040685, | |
| "grad_norm": 0.48265865445137024, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 41.1537, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.5139186295503212, | |
| "grad_norm": 0.3337669372558594, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 41.1761, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.5182012847965739, | |
| "grad_norm": 0.5385298728942871, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 41.1872, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5224839400428265, | |
| "grad_norm": 0.43514499068260193, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 41.1601, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5267665952890792, | |
| "grad_norm": 0.47093290090560913, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 41.1639, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5310492505353319, | |
| "grad_norm": 0.4684036076068878, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 41.1672, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5353319057815846, | |
| "grad_norm": 0.3665253221988678, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 41.1986, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5396145610278372, | |
| "grad_norm": 0.47622647881507874, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 41.1763, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5438972162740899, | |
| "grad_norm": 0.44692617654800415, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 41.1306, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5481798715203426, | |
| "grad_norm": 0.40188875794410706, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 41.1763, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5524625267665952, | |
| "grad_norm": 0.4522937536239624, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 41.1487, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.556745182012848, | |
| "grad_norm": 0.455946147441864, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 41.1591, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.5610278372591007, | |
| "grad_norm": 0.38201573491096497, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 41.1784, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.5653104925053534, | |
| "grad_norm": 0.5712035298347473, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 41.1288, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.569593147751606, | |
| "grad_norm": 0.4802549183368683, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 41.1453, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5738758029978587, | |
| "grad_norm": 0.5156493782997131, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 41.1488, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5781584582441114, | |
| "grad_norm": 0.44495895504951477, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 41.1394, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.582441113490364, | |
| "grad_norm": 0.46417051553726196, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 41.126, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.5867237687366167, | |
| "grad_norm": 0.3869642913341522, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 41.1446, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.5910064239828694, | |
| "grad_norm": 0.46840083599090576, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 41.1703, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5952890792291221, | |
| "grad_norm": 0.43437740206718445, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 41.1154, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5995717344753747, | |
| "grad_norm": 0.43356087803840637, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 41.1531, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6038543897216274, | |
| "grad_norm": 0.49236592650413513, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 41.1226, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6081370449678801, | |
| "grad_norm": 0.515826940536499, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 41.1367, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.6124197002141327, | |
| "grad_norm": 0.5549665093421936, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 41.1751, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.6167023554603854, | |
| "grad_norm": 0.5821396708488464, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 41.1535, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.6209850107066381, | |
| "grad_norm": 0.9271972179412842, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 41.1525, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6252676659528907, | |
| "grad_norm": 0.603287398815155, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 41.167, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6295503211991434, | |
| "grad_norm": 0.7227380871772766, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 41.1586, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6338329764453962, | |
| "grad_norm": 0.7638959884643555, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 41.115, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6381156316916489, | |
| "grad_norm": 0.9707202315330505, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 41.1299, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6423982869379015, | |
| "grad_norm": 1.1904702186584473, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 41.1096, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6423982869379015, | |
| "eval_loss": 10.288755416870117, | |
| "eval_runtime": 0.8147, | |
| "eval_samples_per_second": 483.6, | |
| "eval_steps_per_second": 121.514, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6466809421841542, | |
| "grad_norm": 0.8743388056755066, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 41.2343, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6509635974304069, | |
| "grad_norm": 0.8372085690498352, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 41.2028, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6552462526766595, | |
| "grad_norm": 0.6851708889007568, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 41.2184, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.6595289079229122, | |
| "grad_norm": 0.6139897704124451, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 41.2032, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6638115631691649, | |
| "grad_norm": 0.5757542252540588, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 41.1716, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6680942184154176, | |
| "grad_norm": 0.5816591382026672, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 41.1707, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6723768736616702, | |
| "grad_norm": 0.6130126118659973, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 41.1754, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.6766595289079229, | |
| "grad_norm": 0.6503409743309021, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 41.1595, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.6809421841541756, | |
| "grad_norm": 0.5550835728645325, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 41.1744, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.6852248394004282, | |
| "grad_norm": 0.5488256812095642, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 41.1501, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.6895074946466809, | |
| "grad_norm": 0.42393171787261963, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 41.1699, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.6937901498929336, | |
| "grad_norm": 0.4661500155925751, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 41.1465, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.6980728051391863, | |
| "grad_norm": 0.5130885243415833, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 41.1654, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.702355460385439, | |
| "grad_norm": 0.4037836790084839, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 41.1324, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.7066381156316917, | |
| "grad_norm": 0.501358687877655, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 41.1913, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.7109207708779444, | |
| "grad_norm": 0.6229903101921082, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 41.1398, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.715203426124197, | |
| "grad_norm": 0.5036523938179016, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 41.1777, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.7194860813704497, | |
| "grad_norm": 0.4165601134300232, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 41.1167, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.7237687366167024, | |
| "grad_norm": 0.5183447003364563, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 41.1806, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.728051391862955, | |
| "grad_norm": 0.37226128578186035, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 41.1631, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7323340471092077, | |
| "grad_norm": 0.43647095561027527, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 41.1716, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7366167023554604, | |
| "grad_norm": 0.37477391958236694, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 41.1419, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7408993576017131, | |
| "grad_norm": 0.38343745470046997, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 41.1524, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7451820128479657, | |
| "grad_norm": 0.457410603761673, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 41.1392, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7494646680942184, | |
| "grad_norm": 0.4119358956813812, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 41.159, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.7537473233404711, | |
| "grad_norm": 0.4237579107284546, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 41.1824, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7580299785867237, | |
| "grad_norm": 0.5307998061180115, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 41.1349, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7623126338329764, | |
| "grad_norm": 0.39958441257476807, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 41.1412, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7665952890792291, | |
| "grad_norm": 0.36851799488067627, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 41.1155, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7708779443254818, | |
| "grad_norm": 0.4594135880470276, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 41.1634, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.7751605995717344, | |
| "grad_norm": 0.4165166914463043, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 41.1465, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.7794432548179872, | |
| "grad_norm": 0.47898855805397034, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 41.177, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.7837259100642399, | |
| "grad_norm": 0.3837435245513916, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 41.1068, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.7880085653104925, | |
| "grad_norm": 0.5167264342308044, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 41.1565, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.7922912205567452, | |
| "grad_norm": 0.413090318441391, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 41.1915, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.7965738758029979, | |
| "grad_norm": 0.5370827913284302, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 41.1455, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8008565310492506, | |
| "grad_norm": 0.39179834723472595, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 41.1149, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8051391862955032, | |
| "grad_norm": 0.6464431881904602, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 41.1741, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8094218415417559, | |
| "grad_norm": 0.40582799911499023, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 41.1555, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.8137044967880086, | |
| "grad_norm": 0.5137993693351746, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 41.1078, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8179871520342612, | |
| "grad_norm": 0.531497597694397, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 41.1521, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.8222698072805139, | |
| "grad_norm": 0.5072067379951477, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 41.084, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.8265524625267666, | |
| "grad_norm": 0.7040443420410156, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 41.0847, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.8308351177730193, | |
| "grad_norm": 0.5859010815620422, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 41.1341, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8351177730192719, | |
| "grad_norm": 0.6137392520904541, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 41.1505, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.8394004282655246, | |
| "grad_norm": 0.5699211359024048, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 41.1112, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8436830835117773, | |
| "grad_norm": 0.768570065498352, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 41.1204, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8479657387580299, | |
| "grad_norm": 0.7753411531448364, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 41.1279, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8522483940042827, | |
| "grad_norm": 0.8194027543067932, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 41.0847, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8565310492505354, | |
| "grad_norm": 0.9929121732711792, | |
| "learning_rate": 0.0, | |
| "loss": 41.1802, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8565310492505354, | |
| "eval_loss": 10.287935256958008, | |
| "eval_runtime": 0.8236, | |
| "eval_samples_per_second": 478.394, | |
| "eval_steps_per_second": 120.206, | |
| "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": 43874775465984.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |