| { | |
| "best_metric": 0.5245915055274963, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.11245431543435479, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.000562271577171774, | |
| "grad_norm": 0.9277647137641907, | |
| "learning_rate": 1e-05, | |
| "loss": 0.9523, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.000562271577171774, | |
| "eval_loss": 1.107572078704834, | |
| "eval_runtime": 60.3006, | |
| "eval_samples_per_second": 49.668, | |
| "eval_steps_per_second": 12.421, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.001124543154343548, | |
| "grad_norm": 1.1947993040084839, | |
| "learning_rate": 2e-05, | |
| "loss": 0.9678, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0016868147315153219, | |
| "grad_norm": 1.1136486530303955, | |
| "learning_rate": 3e-05, | |
| "loss": 0.9377, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.002249086308687096, | |
| "grad_norm": 1.3076503276824951, | |
| "learning_rate": 4e-05, | |
| "loss": 1.055, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.0028113578858588698, | |
| "grad_norm": 1.0048154592514038, | |
| "learning_rate": 5e-05, | |
| "loss": 0.9123, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.0033736294630306437, | |
| "grad_norm": 1.0582215785980225, | |
| "learning_rate": 6e-05, | |
| "loss": 0.9838, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.003935901040202418, | |
| "grad_norm": 1.1351639032363892, | |
| "learning_rate": 7e-05, | |
| "loss": 0.9163, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.004498172617374192, | |
| "grad_norm": 1.0295337438583374, | |
| "learning_rate": 8e-05, | |
| "loss": 0.9018, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.005060444194545966, | |
| "grad_norm": 0.9168885946273804, | |
| "learning_rate": 9e-05, | |
| "loss": 0.8344, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.0056227157717177395, | |
| "grad_norm": 1.0283502340316772, | |
| "learning_rate": 0.0001, | |
| "loss": 0.8446, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.006184987348889514, | |
| "grad_norm": 1.1670539379119873, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.787, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.0067472589260612875, | |
| "grad_norm": 0.9883449673652649, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.7885, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.007309530503233062, | |
| "grad_norm": 1.1391271352767944, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.8175, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.007871802080404836, | |
| "grad_norm": 0.9470503330230713, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.7658, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.00843407365757661, | |
| "grad_norm": 0.8227817416191101, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.6894, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.008996345234748383, | |
| "grad_norm": 0.8031889200210571, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.753, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.009558616811920157, | |
| "grad_norm": 0.8018306493759155, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.7226, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.010120888389091932, | |
| "grad_norm": 0.867193341255188, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.6762, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.010683159966263706, | |
| "grad_norm": 0.7935461401939392, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.7003, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.011245431543435479, | |
| "grad_norm": 0.722403883934021, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.5762, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.011807703120607253, | |
| "grad_norm": 0.7102622389793396, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.7118, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.012369974697779028, | |
| "grad_norm": 0.8430309295654297, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.8028, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.012932246274950801, | |
| "grad_norm": 0.7064953446388245, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.6563, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.013494517852122575, | |
| "grad_norm": 0.7702734470367432, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.5956, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.014056789429294348, | |
| "grad_norm": 0.8048498034477234, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.6192, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.014619061006466124, | |
| "grad_norm": 0.769148588180542, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.6339, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.015181332583637897, | |
| "grad_norm": 0.744289755821228, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.5355, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.015743604160809672, | |
| "grad_norm": 0.7577502131462097, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.6171, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.016305875737981444, | |
| "grad_norm": 0.824184238910675, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.6907, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.01686814731515322, | |
| "grad_norm": 0.8315814733505249, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.5212, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.01743041889232499, | |
| "grad_norm": 0.8605427742004395, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.5392, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.017992690469496767, | |
| "grad_norm": 0.7918468117713928, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.4832, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.018554962046668542, | |
| "grad_norm": 0.8169512152671814, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.5747, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.019117233623840314, | |
| "grad_norm": 0.8559743165969849, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.5924, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.01967950520101209, | |
| "grad_norm": 0.801286518573761, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.5012, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.020241776778183864, | |
| "grad_norm": 0.8353800177574158, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.4738, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.020804048355355636, | |
| "grad_norm": 0.7663769125938416, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.5555, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.02136631993252741, | |
| "grad_norm": 0.7065837383270264, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.4453, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.021928591509699186, | |
| "grad_norm": 0.8448126912117004, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.4219, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.022490863086870958, | |
| "grad_norm": 0.8258911967277527, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.4561, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.023053134664042733, | |
| "grad_norm": 0.8381558656692505, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.5233, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.023615406241214505, | |
| "grad_norm": 0.9373324513435364, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.5464, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.02417767781838628, | |
| "grad_norm": 0.9225761890411377, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.5119, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.024739949395558056, | |
| "grad_norm": 0.9429646730422974, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.4632, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.025302220972729828, | |
| "grad_norm": 0.9394980072975159, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.5444, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.025864492549901603, | |
| "grad_norm": 1.0874422788619995, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.6728, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.026426764127073378, | |
| "grad_norm": 1.0107288360595703, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.5739, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.02698903570424515, | |
| "grad_norm": 1.1551225185394287, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.4969, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.027551307281416925, | |
| "grad_norm": 1.3663979768753052, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.7977, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.028113578858588697, | |
| "grad_norm": 2.2082934379577637, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.7978, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.028113578858588697, | |
| "eval_loss": 0.6958192586898804, | |
| "eval_runtime": 60.4229, | |
| "eval_samples_per_second": 49.567, | |
| "eval_steps_per_second": 12.396, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.028675850435760472, | |
| "grad_norm": 1.617163062095642, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.8293, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.029238122012932247, | |
| "grad_norm": 1.5228155851364136, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.834, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.02980039359010402, | |
| "grad_norm": 1.3696954250335693, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.8516, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.030362665167275794, | |
| "grad_norm": 0.9437366724014282, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.7302, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.03092493674444757, | |
| "grad_norm": 0.7396307587623596, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.6234, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.031487208321619345, | |
| "grad_norm": 0.6352867484092712, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.6283, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.03204947989879112, | |
| "grad_norm": 0.6413976550102234, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.7315, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.03261175147596289, | |
| "grad_norm": 0.6398018598556519, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.5883, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.03317402305313467, | |
| "grad_norm": 0.6652855277061462, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.6167, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.03373629463030644, | |
| "grad_norm": 0.6662874817848206, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.6626, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.03429856620747821, | |
| "grad_norm": 0.6972583532333374, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.6922, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.03486083778464998, | |
| "grad_norm": 0.5826914310455322, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.5906, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.03542310936182176, | |
| "grad_norm": 0.6414178013801575, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.5979, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.03598538093899353, | |
| "grad_norm": 0.614965558052063, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.6806, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.036547652516165305, | |
| "grad_norm": 0.6080766320228577, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.6056, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.037109924093337084, | |
| "grad_norm": 0.599414587020874, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.5893, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.037672195670508855, | |
| "grad_norm": 0.6107190847396851, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.6567, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.03823446724768063, | |
| "grad_norm": 0.6296833157539368, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.6634, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.038796738824852406, | |
| "grad_norm": 0.6974261999130249, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.5943, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.03935901040202418, | |
| "grad_norm": 0.650062084197998, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.5836, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.03992128197919595, | |
| "grad_norm": 0.6339553594589233, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.5484, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.04048355355636773, | |
| "grad_norm": 0.6875646710395813, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.6858, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.0410458251335395, | |
| "grad_norm": 0.64603191614151, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.623, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.04160809671071127, | |
| "grad_norm": 0.6082363128662109, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.5215, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.04217036828788305, | |
| "grad_norm": 0.6290409564971924, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.4936, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.04273263986505482, | |
| "grad_norm": 0.6085454821586609, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.5462, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.043294911442226594, | |
| "grad_norm": 0.7018570303916931, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.6391, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.04385718301939837, | |
| "grad_norm": 0.5906082391738892, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.4946, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.044419454596570145, | |
| "grad_norm": 0.645088791847229, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.4762, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.044981726173741916, | |
| "grad_norm": 0.6799085140228271, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.543, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.04554399775091369, | |
| "grad_norm": 0.6760143041610718, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.5326, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.04610626932808547, | |
| "grad_norm": 0.6451261639595032, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.4561, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.04666854090525724, | |
| "grad_norm": 0.6407421231269836, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.4899, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.04723081248242901, | |
| "grad_norm": 0.659145712852478, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.4942, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.04779308405960079, | |
| "grad_norm": 0.6394976377487183, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.4354, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.04835535563677256, | |
| "grad_norm": 0.6170784831047058, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.455, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.04891762721394433, | |
| "grad_norm": 0.6625480651855469, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.458, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.04947989879111611, | |
| "grad_norm": 0.6765840649604797, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.3961, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.05004217036828788, | |
| "grad_norm": 0.6819695234298706, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.5176, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.050604441945459655, | |
| "grad_norm": 0.7992807626724243, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.4787, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.051166713522631434, | |
| "grad_norm": 0.6881734132766724, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.4361, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.051728985099803206, | |
| "grad_norm": 0.7603946328163147, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.4724, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.05229125667697498, | |
| "grad_norm": 0.8886284828186035, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.5389, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.052853528254146756, | |
| "grad_norm": 0.7603895664215088, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.5042, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.05341579983131853, | |
| "grad_norm": 0.8904877305030823, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.6202, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.0539780714084903, | |
| "grad_norm": 0.726782500743866, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.5166, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.05454034298566207, | |
| "grad_norm": 0.8900876045227051, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.6052, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.05510261456283385, | |
| "grad_norm": 0.919487714767456, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.5764, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.05566488614000562, | |
| "grad_norm": 1.2879389524459839, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.6897, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.056227157717177394, | |
| "grad_norm": 1.9414801597595215, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.4912, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.056227157717177394, | |
| "eval_loss": 0.5721317529678345, | |
| "eval_runtime": 60.446, | |
| "eval_samples_per_second": 49.548, | |
| "eval_steps_per_second": 12.391, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05678942929434917, | |
| "grad_norm": 0.7900543212890625, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.7565, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.057351700871520944, | |
| "grad_norm": 0.7188913226127625, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.6344, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.057913972448692716, | |
| "grad_norm": 0.7000232338905334, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.7186, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.058476244025864495, | |
| "grad_norm": 0.7585048675537109, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.742, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.059038515603036266, | |
| "grad_norm": 0.6727244257926941, | |
| "learning_rate": 5e-05, | |
| "loss": 0.6335, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.05960078718020804, | |
| "grad_norm": 0.635672926902771, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.7268, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.06016305875737982, | |
| "grad_norm": 0.5601180791854858, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.6556, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.06072533033455159, | |
| "grad_norm": 0.5831847786903381, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.5855, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.06128760191172336, | |
| "grad_norm": 0.5406240224838257, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.663, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.06184987348889514, | |
| "grad_norm": 0.5173934102058411, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.5705, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.06241214506606691, | |
| "grad_norm": 0.5559889674186707, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.591, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.06297441664323869, | |
| "grad_norm": 0.5619823336601257, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.5909, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.06353668822041046, | |
| "grad_norm": 0.5657660365104675, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.6236, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.06409895979758223, | |
| "grad_norm": 0.6045258641242981, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.5462, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.064661231374754, | |
| "grad_norm": 0.583362340927124, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.5561, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.06522350295192578, | |
| "grad_norm": 0.6070014834403992, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.6116, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.06578577452909755, | |
| "grad_norm": 0.5519648194313049, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.591, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.06634804610626933, | |
| "grad_norm": 0.5768078565597534, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.5516, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.0669103176834411, | |
| "grad_norm": 0.5689049959182739, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.5451, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.06747258926061288, | |
| "grad_norm": 0.611607551574707, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.5779, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.06803486083778465, | |
| "grad_norm": 0.6133226752281189, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.5086, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.06859713241495642, | |
| "grad_norm": 0.6262627243995667, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.606, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.0691594039921282, | |
| "grad_norm": 0.5980281829833984, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.5636, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.06972167556929997, | |
| "grad_norm": 0.6119388341903687, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.5313, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.07028394714647175, | |
| "grad_norm": 0.5762614607810974, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.4927, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.07084621872364352, | |
| "grad_norm": 0.656383752822876, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.632, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.0714084903008153, | |
| "grad_norm": 0.5831140279769897, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.4951, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.07197076187798707, | |
| "grad_norm": 0.6357362270355225, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.5096, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.07253303345515884, | |
| "grad_norm": 0.6704001426696777, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.6236, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.07309530503233061, | |
| "grad_norm": 0.5634742975234985, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.4083, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0736575766095024, | |
| "grad_norm": 0.6350815892219543, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.4295, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.07421984818667417, | |
| "grad_norm": 0.6566004157066345, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.4454, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.07478211976384594, | |
| "grad_norm": 0.6152843236923218, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.4091, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.07534439134101771, | |
| "grad_norm": 0.7407167553901672, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.5535, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.07590666291818948, | |
| "grad_norm": 0.6638815402984619, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.4944, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.07646893449536125, | |
| "grad_norm": 0.6325174570083618, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.399, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.07703120607253304, | |
| "grad_norm": 0.683302640914917, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.4391, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.07759347764970481, | |
| "grad_norm": 0.6366162300109863, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.3555, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.07815574922687658, | |
| "grad_norm": 0.67439866065979, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.4154, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.07871802080404836, | |
| "grad_norm": 0.6465699076652527, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.378, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.07928029238122013, | |
| "grad_norm": 0.6464277505874634, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.3991, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.0798425639583919, | |
| "grad_norm": 0.6784489154815674, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3894, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.08040483553556367, | |
| "grad_norm": 0.7543945908546448, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.4289, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.08096710711273546, | |
| "grad_norm": 0.728791356086731, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.4247, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.08152937868990723, | |
| "grad_norm": 0.8233468532562256, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.5633, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.082091650267079, | |
| "grad_norm": 0.7600351572036743, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.4912, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.08265392184425077, | |
| "grad_norm": 0.823025107383728, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.5105, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.08321619342142254, | |
| "grad_norm": 0.9203780889511108, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.5627, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.08377846499859432, | |
| "grad_norm": 1.0546104907989502, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.644, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.0843407365757661, | |
| "grad_norm": 1.2494205236434937, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.3982, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.0843407365757661, | |
| "eval_loss": 0.5304158329963684, | |
| "eval_runtime": 60.4588, | |
| "eval_samples_per_second": 49.538, | |
| "eval_steps_per_second": 12.389, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.08490300815293787, | |
| "grad_norm": 0.4984513819217682, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.6096, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.08546527973010964, | |
| "grad_norm": 0.5627269744873047, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.6625, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.08602755130728142, | |
| "grad_norm": 0.5592852234840393, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.6395, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.08658982288445319, | |
| "grad_norm": 0.5454599857330322, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.6092, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.08715209446162496, | |
| "grad_norm": 0.5715036392211914, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.6648, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.08771436603879675, | |
| "grad_norm": 0.5295091271400452, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.5753, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.08827663761596852, | |
| "grad_norm": 0.5699715614318848, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.6065, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.08883890919314029, | |
| "grad_norm": 0.5259392261505127, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.5584, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.08940118077031206, | |
| "grad_norm": 0.6719173192977905, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.613, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.08996345234748383, | |
| "grad_norm": 0.5674479007720947, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.5934, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.0905257239246556, | |
| "grad_norm": 0.6157873868942261, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.6343, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.09108799550182738, | |
| "grad_norm": 0.6046155691146851, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.6198, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.09165026707899916, | |
| "grad_norm": 0.5647667646408081, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.6186, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.09221253865617093, | |
| "grad_norm": 0.5314237475395203, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.4828, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.0927748102333427, | |
| "grad_norm": 0.5461083650588989, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.5188, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.09333708181051448, | |
| "grad_norm": 0.5884983539581299, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.6037, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.09389935338768625, | |
| "grad_norm": 0.5966055393218994, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.5606, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.09446162496485802, | |
| "grad_norm": 0.5838494300842285, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.5659, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.0950238965420298, | |
| "grad_norm": 0.5209010243415833, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.4508, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.09558616811920158, | |
| "grad_norm": 0.5918200612068176, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.539, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.09614843969637335, | |
| "grad_norm": 0.6243396997451782, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.608, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.09671071127354512, | |
| "grad_norm": 0.6295937299728394, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.6011, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.0972729828507169, | |
| "grad_norm": 0.6207840442657471, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.5537, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.09783525442788867, | |
| "grad_norm": 0.5702746510505676, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.4806, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.09839752600506045, | |
| "grad_norm": 0.6419647932052612, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.5246, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.09895979758223222, | |
| "grad_norm": 0.5579126477241516, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.4836, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.099522069159404, | |
| "grad_norm": 0.666792094707489, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.562, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.10008434073657577, | |
| "grad_norm": 0.6097604036331177, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.4448, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.10064661231374754, | |
| "grad_norm": 0.6671866774559021, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.534, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.10120888389091931, | |
| "grad_norm": 0.6225870847702026, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.4926, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.10177115546809108, | |
| "grad_norm": 0.6297417283058167, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.4599, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.10233342704526287, | |
| "grad_norm": 0.5996748208999634, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.4007, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.10289569862243464, | |
| "grad_norm": 0.7020567059516907, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.467, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.10345797019960641, | |
| "grad_norm": 0.6249246001243591, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.4417, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.10402024177677818, | |
| "grad_norm": 0.6193715333938599, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.4229, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.10458251335394995, | |
| "grad_norm": 0.6512302756309509, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.4499, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.10514478493112173, | |
| "grad_norm": 0.6990061402320862, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.4468, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.10570705650829351, | |
| "grad_norm": 0.7123408317565918, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.4648, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.10626932808546528, | |
| "grad_norm": 0.6349745392799377, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.3529, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.10683159966263706, | |
| "grad_norm": 0.682070791721344, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.4053, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.10739387123980883, | |
| "grad_norm": 0.7470072507858276, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.4741, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.1079561428169806, | |
| "grad_norm": 0.7900189757347107, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.4458, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.10851841439415237, | |
| "grad_norm": 0.7109037041664124, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3639, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.10908068597132414, | |
| "grad_norm": 0.8317274451255798, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.4855, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.10964295754849593, | |
| "grad_norm": 0.80501788854599, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.5061, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.1102052291256677, | |
| "grad_norm": 0.7943066954612732, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.4833, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.11076750070283947, | |
| "grad_norm": 0.8239003419876099, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.5212, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.11132977228001124, | |
| "grad_norm": 1.0709213018417358, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.6827, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.11189204385718302, | |
| "grad_norm": 0.9007803797721863, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.5218, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.11245431543435479, | |
| "grad_norm": 1.3899116516113281, | |
| "learning_rate": 0.0, | |
| "loss": 0.5689, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.11245431543435479, | |
| "eval_loss": 0.5245915055274963, | |
| "eval_runtime": 60.4684, | |
| "eval_samples_per_second": 49.53, | |
| "eval_steps_per_second": 12.387, | |
| "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": 4.26629570494464e+16, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |