| { | |
| "best_metric": 0.08209174126386642, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-150", | |
| "epoch": 2.168021680216802, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01084010840108401, | |
| "grad_norm": 1.999291181564331, | |
| "learning_rate": 1e-05, | |
| "loss": 0.771, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.01084010840108401, | |
| "eval_loss": 1.4562032222747803, | |
| "eval_runtime": 3.7676, | |
| "eval_samples_per_second": 41.406, | |
| "eval_steps_per_second": 10.351, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.02168021680216802, | |
| "grad_norm": 1.9173811674118042, | |
| "learning_rate": 2e-05, | |
| "loss": 0.924, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.032520325203252036, | |
| "grad_norm": 2.1829094886779785, | |
| "learning_rate": 3e-05, | |
| "loss": 0.7751, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.04336043360433604, | |
| "grad_norm": 2.005078077316284, | |
| "learning_rate": 4e-05, | |
| "loss": 0.8226, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.05420054200542006, | |
| "grad_norm": 2.3045454025268555, | |
| "learning_rate": 5e-05, | |
| "loss": 1.0235, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.06504065040650407, | |
| "grad_norm": 1.9272059202194214, | |
| "learning_rate": 6e-05, | |
| "loss": 0.8711, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.07588075880758807, | |
| "grad_norm": 1.5134131908416748, | |
| "learning_rate": 7e-05, | |
| "loss": 0.5777, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.08672086720867209, | |
| "grad_norm": 1.6031486988067627, | |
| "learning_rate": 8e-05, | |
| "loss": 0.6303, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0975609756097561, | |
| "grad_norm": 1.7916780710220337, | |
| "learning_rate": 9e-05, | |
| "loss": 0.5908, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.10840108401084012, | |
| "grad_norm": 1.5881563425064087, | |
| "learning_rate": 0.0001, | |
| "loss": 0.5093, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.11924119241192412, | |
| "grad_norm": 2.0340209007263184, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.5138, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.13008130081300814, | |
| "grad_norm": 2.537306547164917, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.3805, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.14092140921409213, | |
| "grad_norm": 1.6741464138031006, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.4951, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.15176151761517614, | |
| "grad_norm": 2.495614767074585, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.2785, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.16260162601626016, | |
| "grad_norm": 1.6360888481140137, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.3944, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.17344173441734417, | |
| "grad_norm": 1.269214391708374, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.1705, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.1842818428184282, | |
| "grad_norm": 1.8224081993103027, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.2173, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.1951219512195122, | |
| "grad_norm": 1.7426091432571411, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.2531, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.20596205962059622, | |
| "grad_norm": 1.7515729665756226, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.2463, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.21680216802168023, | |
| "grad_norm": 1.3095314502716064, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.2323, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.22764227642276422, | |
| "grad_norm": 1.9506779909133911, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.4159, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.23848238482384823, | |
| "grad_norm": 2.067443370819092, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.342, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.24932249322493225, | |
| "grad_norm": 2.9920454025268555, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.5486, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.2601626016260163, | |
| "grad_norm": 1.9352222681045532, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.1548, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.27100271002710025, | |
| "grad_norm": 2.535074234008789, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.1409, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.28184281842818426, | |
| "grad_norm": 0.8707655072212219, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.0885, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.2926829268292683, | |
| "grad_norm": 0.7993746399879456, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.1716, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.3035230352303523, | |
| "grad_norm": 0.9028083086013794, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.1121, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.3143631436314363, | |
| "grad_norm": 1.4698336124420166, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.2176, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.3252032520325203, | |
| "grad_norm": 0.9743578433990479, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.175, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.33604336043360433, | |
| "grad_norm": 0.6200997233390808, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.173, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.34688346883468835, | |
| "grad_norm": 0.7226480841636658, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.1423, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.35772357723577236, | |
| "grad_norm": 0.6630494594573975, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.1464, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.3685636856368564, | |
| "grad_norm": 0.8004043698310852, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.1884, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.3794037940379404, | |
| "grad_norm": 0.7413557767868042, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.1865, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.3902439024390244, | |
| "grad_norm": 0.7330390810966492, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.0886, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.4010840108401084, | |
| "grad_norm": 0.6222772598266602, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.0897, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.41192411924119243, | |
| "grad_norm": 0.7401354908943176, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.108, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.42276422764227645, | |
| "grad_norm": 0.3237884044647217, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.032, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.43360433604336046, | |
| "grad_norm": 0.9113665223121643, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.1656, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.4444444444444444, | |
| "grad_norm": 0.5773696899414062, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.0497, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.45528455284552843, | |
| "grad_norm": 0.4682125151157379, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.0352, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.46612466124661245, | |
| "grad_norm": 0.8373630046844482, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.083, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.47696476964769646, | |
| "grad_norm": 1.2706354856491089, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.1535, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.4878048780487805, | |
| "grad_norm": 1.339192271232605, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.1267, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.4986449864498645, | |
| "grad_norm": 2.027174711227417, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.2265, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.5094850948509485, | |
| "grad_norm": 2.6765296459198, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.2349, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.5203252032520326, | |
| "grad_norm": 1.0258209705352783, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.1282, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.5311653116531165, | |
| "grad_norm": 2.309070348739624, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.1447, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.5420054200542005, | |
| "grad_norm": 0.6303039193153381, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.1437, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5420054200542005, | |
| "eval_loss": 0.12350469827651978, | |
| "eval_runtime": 3.6979, | |
| "eval_samples_per_second": 42.186, | |
| "eval_steps_per_second": 10.546, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5528455284552846, | |
| "grad_norm": 0.5761507153511047, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.1187, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.5636856368563685, | |
| "grad_norm": 0.6159778237342834, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.0727, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.5745257452574526, | |
| "grad_norm": 0.8006864786148071, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.0819, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.5853658536585366, | |
| "grad_norm": 0.7325972318649292, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.1168, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.5962059620596206, | |
| "grad_norm": 0.6329816579818726, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.1117, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.6070460704607046, | |
| "grad_norm": 0.8424084186553955, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.093, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.6178861788617886, | |
| "grad_norm": 0.6663784980773926, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.1303, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.6287262872628726, | |
| "grad_norm": 0.523448646068573, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.0914, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.6395663956639567, | |
| "grad_norm": 0.7643095254898071, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.1144, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.6504065040650406, | |
| "grad_norm": 0.4949480891227722, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.0521, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.6612466124661247, | |
| "grad_norm": 0.6844000220298767, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.1302, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.6720867208672087, | |
| "grad_norm": 0.6087970733642578, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.0696, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.6829268292682927, | |
| "grad_norm": 0.6328505277633667, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.1631, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.6937669376693767, | |
| "grad_norm": 0.7826113104820251, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.174, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.7046070460704607, | |
| "grad_norm": 0.7086710333824158, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.1047, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.7154471544715447, | |
| "grad_norm": 0.5617452263832092, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.0978, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.7262872628726287, | |
| "grad_norm": 0.5778073668479919, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.0718, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.7371273712737128, | |
| "grad_norm": 0.836530327796936, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.0869, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.7479674796747967, | |
| "grad_norm": 0.9944505095481873, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.1184, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.7588075880758808, | |
| "grad_norm": 0.4123961627483368, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.1155, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.7696476964769647, | |
| "grad_norm": 0.4990311563014984, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.0938, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.7804878048780488, | |
| "grad_norm": 0.28021708130836487, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.0481, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.7913279132791328, | |
| "grad_norm": 0.4944506585597992, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.146, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.8021680216802168, | |
| "grad_norm": 0.7031269669532776, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.1715, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.8130081300813008, | |
| "grad_norm": 0.4086264371871948, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.1055, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.8238482384823849, | |
| "grad_norm": 0.33718255162239075, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.0451, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.8346883468834688, | |
| "grad_norm": 0.4637962877750397, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.0806, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.8455284552845529, | |
| "grad_norm": 0.5173426866531372, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.0526, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.8563685636856369, | |
| "grad_norm": 0.44306081533432007, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.0655, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.8672086720867209, | |
| "grad_norm": 0.7521052360534668, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.1433, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.8780487804878049, | |
| "grad_norm": 0.5706959962844849, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.0591, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.8888888888888888, | |
| "grad_norm": 0.46802663803100586, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.0467, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.8997289972899729, | |
| "grad_norm": 0.711159348487854, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.0826, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.9105691056910569, | |
| "grad_norm": 0.813417375087738, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.1174, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.9214092140921409, | |
| "grad_norm": 0.677147626876831, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.139, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.9322493224932249, | |
| "grad_norm": 0.41531622409820557, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.0284, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.943089430894309, | |
| "grad_norm": 0.3974713683128357, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.0475, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.9539295392953929, | |
| "grad_norm": 0.6932405829429626, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.0966, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.964769647696477, | |
| "grad_norm": 0.386869877576828, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.0464, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.975609756097561, | |
| "grad_norm": 0.8761130571365356, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.1191, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.986449864498645, | |
| "grad_norm": 0.5070018768310547, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.029, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.997289972899729, | |
| "grad_norm": 1.7458750009536743, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.1672, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 1.008130081300813, | |
| "grad_norm": 1.0867496728897095, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.1955, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 1.018970189701897, | |
| "grad_norm": 0.27728036046028137, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.0539, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 1.029810298102981, | |
| "grad_norm": 0.23874682188034058, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.0637, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 1.040650406504065, | |
| "grad_norm": 0.15150757133960724, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.0186, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 1.051490514905149, | |
| "grad_norm": 0.35452330112457275, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.0713, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 1.062330623306233, | |
| "grad_norm": 0.549517035484314, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.0866, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 1.0731707317073171, | |
| "grad_norm": 0.44955000281333923, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.0723, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 1.084010840108401, | |
| "grad_norm": 0.3139077425003052, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.0249, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.084010840108401, | |
| "eval_loss": 0.09879136830568314, | |
| "eval_runtime": 3.6916, | |
| "eval_samples_per_second": 42.258, | |
| "eval_steps_per_second": 10.565, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.094850948509485, | |
| "grad_norm": 0.4146367907524109, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.0374, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 1.1056910569105691, | |
| "grad_norm": 0.36523959040641785, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.035, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 1.1165311653116532, | |
| "grad_norm": 0.3721497356891632, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.0389, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 1.127371273712737, | |
| "grad_norm": 0.2500871419906616, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.0333, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 1.1382113821138211, | |
| "grad_norm": 0.5672609210014343, | |
| "learning_rate": 5e-05, | |
| "loss": 0.0445, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 1.1490514905149052, | |
| "grad_norm": 0.5433923006057739, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.0574, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 1.1598915989159893, | |
| "grad_norm": 0.43868547677993774, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.0306, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 1.170731707317073, | |
| "grad_norm": 0.43001192808151245, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.0389, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 1.1815718157181572, | |
| "grad_norm": 0.18207843601703644, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.0123, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 1.1924119241192412, | |
| "grad_norm": 0.4550378620624542, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.0243, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.203252032520325, | |
| "grad_norm": 0.27300575375556946, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.0098, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 1.2140921409214092, | |
| "grad_norm": 0.17592142522335052, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.0104, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 1.2249322493224932, | |
| "grad_norm": 0.48420414328575134, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.0202, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 1.2357723577235773, | |
| "grad_norm": 1.371536135673523, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.1445, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 1.2466124661246614, | |
| "grad_norm": 0.7464447617530823, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.0311, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.2574525745257452, | |
| "grad_norm": 1.5276039838790894, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.1417, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 1.2682926829268293, | |
| "grad_norm": 0.26058825850486755, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.0589, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 1.2791327913279134, | |
| "grad_norm": 0.3222558796405792, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.0245, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 1.2899728997289972, | |
| "grad_norm": 0.44456252455711365, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.0413, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 1.3008130081300813, | |
| "grad_norm": 0.27726078033447266, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.0523, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.3116531165311653, | |
| "grad_norm": 0.5933356285095215, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.0652, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 1.3224932249322494, | |
| "grad_norm": 0.6122859120368958, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.0435, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 1.3333333333333333, | |
| "grad_norm": 0.8560364842414856, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.0874, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 1.3441734417344173, | |
| "grad_norm": 0.20532087981700897, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.0102, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 1.3550135501355014, | |
| "grad_norm": 0.7897826433181763, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.0957, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.3658536585365852, | |
| "grad_norm": 0.41313865780830383, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.0335, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 1.3766937669376693, | |
| "grad_norm": 0.2979642152786255, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.0426, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 1.3875338753387534, | |
| "grad_norm": 0.3756512701511383, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.0628, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 1.3983739837398375, | |
| "grad_norm": 0.5913124084472656, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.057, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 1.4092140921409215, | |
| "grad_norm": 0.5157326459884644, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.0371, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.4200542005420054, | |
| "grad_norm": 0.39981499314308167, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.0232, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 1.4308943089430894, | |
| "grad_norm": 0.2877366244792938, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.015, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 1.4417344173441735, | |
| "grad_norm": 0.16340596973896027, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.0079, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 1.4525745257452574, | |
| "grad_norm": 0.2619611322879791, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.0171, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 1.4634146341463414, | |
| "grad_norm": 1.1646555662155151, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.0712, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.4742547425474255, | |
| "grad_norm": 0.5431463122367859, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.0385, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 1.4850948509485096, | |
| "grad_norm": 0.19880428910255432, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.0091, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 1.4959349593495934, | |
| "grad_norm": 0.32186517119407654, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.0166, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 1.5067750677506775, | |
| "grad_norm": 0.30488982796669006, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.0523, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 1.5176151761517616, | |
| "grad_norm": 0.48191314935684204, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.0501, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.5284552845528454, | |
| "grad_norm": 0.45830032229423523, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.0759, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 1.5392953929539295, | |
| "grad_norm": 0.2845216393470764, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.0503, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 1.5501355013550135, | |
| "grad_norm": 0.3950357437133789, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.041, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 1.5609756097560976, | |
| "grad_norm": 0.27710777521133423, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.0405, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 1.5718157181571817, | |
| "grad_norm": 0.14560440182685852, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.0065, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.5826558265582655, | |
| "grad_norm": 0.6007800102233887, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.0633, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 1.5934959349593496, | |
| "grad_norm": 0.3015676736831665, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.04, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 1.6043360433604335, | |
| "grad_norm": 0.39085838198661804, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.0315, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 1.6151761517615175, | |
| "grad_norm": 0.47567927837371826, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.068, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 1.6260162601626016, | |
| "grad_norm": 0.25586554408073425, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.0149, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.6260162601626016, | |
| "eval_loss": 0.08209174126386642, | |
| "eval_runtime": 3.6946, | |
| "eval_samples_per_second": 42.224, | |
| "eval_steps_per_second": 10.556, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.6368563685636857, | |
| "grad_norm": 0.3803505003452301, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.0318, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 1.6476964769647697, | |
| "grad_norm": 0.3315448462963104, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.0313, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 1.6585365853658538, | |
| "grad_norm": 0.35780826210975647, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.0289, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 1.6693766937669376, | |
| "grad_norm": 0.2160133570432663, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.0174, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 1.6802168021680217, | |
| "grad_norm": 0.3088828921318054, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.0355, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.6910569105691056, | |
| "grad_norm": 0.19804991781711578, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.0135, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 1.7018970189701896, | |
| "grad_norm": 0.23907789587974548, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.0066, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 1.7127371273712737, | |
| "grad_norm": 0.670276939868927, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.0446, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 1.7235772357723578, | |
| "grad_norm": 0.13110598921775818, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.0066, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 1.7344173441734418, | |
| "grad_norm": 0.2818697392940521, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.0108, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.7452574525745257, | |
| "grad_norm": 1.065281867980957, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.0302, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 1.7560975609756098, | |
| "grad_norm": 0.3171882927417755, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.0679, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 1.7669376693766936, | |
| "grad_norm": 0.35744714736938477, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.0406, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 1.7777777777777777, | |
| "grad_norm": 0.3069745600223541, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.0203, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 1.7886178861788617, | |
| "grad_norm": 0.27485328912734985, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.0396, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.7994579945799458, | |
| "grad_norm": 0.3660500645637512, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.0399, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 1.8102981029810299, | |
| "grad_norm": 0.388435035943985, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.06, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 1.821138211382114, | |
| "grad_norm": 0.24849200248718262, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.0239, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 1.8319783197831978, | |
| "grad_norm": 0.36250999569892883, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.0331, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 1.8428184281842819, | |
| "grad_norm": 0.6042048931121826, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.1239, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.8536585365853657, | |
| "grad_norm": 0.35673898458480835, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.0597, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 1.8644986449864498, | |
| "grad_norm": 0.40320390462875366, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.0567, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 1.8753387533875339, | |
| "grad_norm": 0.4188055992126465, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.0477, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 1.886178861788618, | |
| "grad_norm": 0.4100712239742279, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.0193, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 1.897018970189702, | |
| "grad_norm": 0.3308526575565338, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.0419, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.907859078590786, | |
| "grad_norm": 0.3961979150772095, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.0537, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 1.91869918699187, | |
| "grad_norm": 0.6286945939064026, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.0783, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 1.9295392953929538, | |
| "grad_norm": 0.2913045287132263, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.0297, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 1.9403794037940378, | |
| "grad_norm": 0.3832351267337799, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.0231, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 1.951219512195122, | |
| "grad_norm": 0.4545169174671173, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.0508, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.962059620596206, | |
| "grad_norm": 0.49141600728034973, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.0508, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 1.97289972899729, | |
| "grad_norm": 0.5272740721702576, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.0669, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 1.9837398373983741, | |
| "grad_norm": 0.6121299266815186, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.0802, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 1.994579945799458, | |
| "grad_norm": 0.5990219116210938, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.0176, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 2.005420054200542, | |
| "grad_norm": 0.25653988122940063, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.0329, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 2.016260162601626, | |
| "grad_norm": 0.19759732484817505, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.0258, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 2.02710027100271, | |
| "grad_norm": 0.2972887456417084, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.0279, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 2.037940379403794, | |
| "grad_norm": 0.22970226407051086, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.024, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 2.048780487804878, | |
| "grad_norm": 0.40008115768432617, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.0673, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 2.059620596205962, | |
| "grad_norm": 0.26496702432632446, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.0308, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 2.0704607046070462, | |
| "grad_norm": 0.23216529190540314, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.0187, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 2.08130081300813, | |
| "grad_norm": 0.284787118434906, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.0324, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 2.092140921409214, | |
| "grad_norm": 0.44172969460487366, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.0713, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 2.102981029810298, | |
| "grad_norm": 0.23959121108055115, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.0526, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 2.113821138211382, | |
| "grad_norm": 0.3692215085029602, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.0328, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 2.124661246612466, | |
| "grad_norm": 0.2914968729019165, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.0285, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 2.13550135501355, | |
| "grad_norm": 0.23906463384628296, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.0128, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 2.1463414634146343, | |
| "grad_norm": 0.10579202324151993, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.0058, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 2.1571815718157183, | |
| "grad_norm": 0.2909102141857147, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.02, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 2.168021680216802, | |
| "grad_norm": 0.11848969012498856, | |
| "learning_rate": 0.0, | |
| "loss": 0.0057, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 2.168021680216802, | |
| "eval_loss": 0.08219181001186371, | |
| "eval_runtime": 3.7029, | |
| "eval_samples_per_second": 42.13, | |
| "eval_steps_per_second": 10.532, | |
| "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": 1 | |
| } | |
| }, | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.268515586048e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |