| { | |
| "best_metric": 0.35021281242370605, | |
| "best_model_checkpoint": "miner_id_24/checkpoint-200", | |
| "epoch": 0.6872852233676976, | |
| "eval_steps": 50, | |
| "global_step": 200, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003436426116838488, | |
| "grad_norm": 0.5733091831207275, | |
| "learning_rate": 1e-05, | |
| "loss": 0.6782, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003436426116838488, | |
| "eval_loss": 0.6093477010726929, | |
| "eval_runtime": 38.0159, | |
| "eval_samples_per_second": 12.889, | |
| "eval_steps_per_second": 3.235, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.006872852233676976, | |
| "grad_norm": 0.4493546783924103, | |
| "learning_rate": 2e-05, | |
| "loss": 0.5529, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.010309278350515464, | |
| "grad_norm": 0.44986408948898315, | |
| "learning_rate": 3e-05, | |
| "loss": 0.5017, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.013745704467353952, | |
| "grad_norm": 0.4682570993900299, | |
| "learning_rate": 4e-05, | |
| "loss": 0.5541, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01718213058419244, | |
| "grad_norm": 0.48950836062431335, | |
| "learning_rate": 5e-05, | |
| "loss": 0.5314, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.020618556701030927, | |
| "grad_norm": 0.33151769638061523, | |
| "learning_rate": 6e-05, | |
| "loss": 0.4097, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.024054982817869417, | |
| "grad_norm": 0.3640710115432739, | |
| "learning_rate": 7e-05, | |
| "loss": 0.4086, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.027491408934707903, | |
| "grad_norm": 0.4267600178718567, | |
| "learning_rate": 8e-05, | |
| "loss": 0.413, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.030927835051546393, | |
| "grad_norm": 0.4396994113922119, | |
| "learning_rate": 9e-05, | |
| "loss": 0.3891, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.03436426116838488, | |
| "grad_norm": 0.3950016498565674, | |
| "learning_rate": 0.0001, | |
| "loss": 0.4027, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.037800687285223365, | |
| "grad_norm": 0.2967038154602051, | |
| "learning_rate": 9.999316524962345e-05, | |
| "loss": 0.402, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.041237113402061855, | |
| "grad_norm": 0.30045998096466064, | |
| "learning_rate": 9.997266286704631e-05, | |
| "loss": 0.3756, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.044673539518900345, | |
| "grad_norm": 0.3312864899635315, | |
| "learning_rate": 9.993849845741524e-05, | |
| "loss": 0.4172, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.048109965635738834, | |
| "grad_norm": 0.3025706708431244, | |
| "learning_rate": 9.989068136093873e-05, | |
| "loss": 0.3579, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05154639175257732, | |
| "grad_norm": 0.3310057818889618, | |
| "learning_rate": 9.98292246503335e-05, | |
| "loss": 0.3654, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.054982817869415807, | |
| "grad_norm": 0.30491307377815247, | |
| "learning_rate": 9.975414512725057e-05, | |
| "loss": 0.3619, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.058419243986254296, | |
| "grad_norm": 0.3243735134601593, | |
| "learning_rate": 9.966546331768191e-05, | |
| "loss": 0.4163, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.061855670103092786, | |
| "grad_norm": 0.3227108120918274, | |
| "learning_rate": 9.956320346634876e-05, | |
| "loss": 0.3129, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06529209621993128, | |
| "grad_norm": 0.32361820340156555, | |
| "learning_rate": 9.944739353007344e-05, | |
| "loss": 0.3911, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.06872852233676977, | |
| "grad_norm": 0.30928871035575867, | |
| "learning_rate": 9.931806517013612e-05, | |
| "loss": 0.3446, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07216494845360824, | |
| "grad_norm": 0.30245766043663025, | |
| "learning_rate": 9.917525374361912e-05, | |
| "loss": 0.3309, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.07560137457044673, | |
| "grad_norm": 0.30670011043548584, | |
| "learning_rate": 9.901899829374047e-05, | |
| "loss": 0.3215, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.07903780068728522, | |
| "grad_norm": 0.31684255599975586, | |
| "learning_rate": 9.884934153917997e-05, | |
| "loss": 0.329, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08247422680412371, | |
| "grad_norm": 0.32323160767555237, | |
| "learning_rate": 9.86663298624003e-05, | |
| "loss": 0.3681, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.0859106529209622, | |
| "grad_norm": 0.34258484840393066, | |
| "learning_rate": 9.847001329696653e-05, | |
| "loss": 0.3539, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.08934707903780069, | |
| "grad_norm": 0.330561101436615, | |
| "learning_rate": 9.826044551386744e-05, | |
| "loss": 0.3564, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09278350515463918, | |
| "grad_norm": 0.3758370578289032, | |
| "learning_rate": 9.803768380684242e-05, | |
| "loss": 0.3739, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.09621993127147767, | |
| "grad_norm": 0.3608083724975586, | |
| "learning_rate": 9.780178907671789e-05, | |
| "loss": 0.4126, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.09965635738831616, | |
| "grad_norm": 0.35548463463783264, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.3937, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.10309278350515463, | |
| "grad_norm": 0.32006946206092834, | |
| "learning_rate": 9.729086208503174e-05, | |
| "loss": 0.3032, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.10652920962199312, | |
| "grad_norm": 0.3427230715751648, | |
| "learning_rate": 9.701596950580806e-05, | |
| "loss": 0.3538, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.10996563573883161, | |
| "grad_norm": 0.36741241812705994, | |
| "learning_rate": 9.672822322997305e-05, | |
| "loss": 0.338, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.1134020618556701, | |
| "grad_norm": 0.38069799542427063, | |
| "learning_rate": 9.642770192448536e-05, | |
| "loss": 0.3347, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.11683848797250859, | |
| "grad_norm": 0.4192986488342285, | |
| "learning_rate": 9.611448774886924e-05, | |
| "loss": 0.4048, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.12027491408934708, | |
| "grad_norm": 0.39656129479408264, | |
| "learning_rate": 9.578866633275288e-05, | |
| "loss": 0.3682, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.12371134020618557, | |
| "grad_norm": 0.3693166673183441, | |
| "learning_rate": 9.545032675245813e-05, | |
| "loss": 0.3379, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.12714776632302405, | |
| "grad_norm": 0.3976968228816986, | |
| "learning_rate": 9.509956150664796e-05, | |
| "loss": 0.343, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.13058419243986255, | |
| "grad_norm": 0.39774540066719055, | |
| "learning_rate": 9.473646649103818e-05, | |
| "loss": 0.3586, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.13402061855670103, | |
| "grad_norm": 0.44844967126846313, | |
| "learning_rate": 9.43611409721806e-05, | |
| "loss": 0.4116, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.13745704467353953, | |
| "grad_norm": 0.37088271975517273, | |
| "learning_rate": 9.397368756032445e-05, | |
| "loss": 0.3276, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.140893470790378, | |
| "grad_norm": 0.4003569781780243, | |
| "learning_rate": 9.357421218136386e-05, | |
| "loss": 0.371, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.14432989690721648, | |
| "grad_norm": 0.39975181221961975, | |
| "learning_rate": 9.316282404787871e-05, | |
| "loss": 0.3567, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.14776632302405499, | |
| "grad_norm": 0.39595574140548706, | |
| "learning_rate": 9.273963562927695e-05, | |
| "loss": 0.3431, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.15120274914089346, | |
| "grad_norm": 0.42598557472229004, | |
| "learning_rate": 9.230476262104677e-05, | |
| "loss": 0.3845, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.15463917525773196, | |
| "grad_norm": 0.39888542890548706, | |
| "learning_rate": 9.185832391312644e-05, | |
| "loss": 0.3375, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.15807560137457044, | |
| "grad_norm": 0.41041457653045654, | |
| "learning_rate": 9.140044155740101e-05, | |
| "loss": 0.347, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.16151202749140894, | |
| "grad_norm": 0.41749244928359985, | |
| "learning_rate": 9.093124073433463e-05, | |
| "loss": 0.3002, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.16494845360824742, | |
| "grad_norm": 0.5060552358627319, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.3924, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.16838487972508592, | |
| "grad_norm": 0.5001901984214783, | |
| "learning_rate": 8.995939984474624e-05, | |
| "loss": 0.3882, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.1718213058419244, | |
| "grad_norm": 0.5671999454498291, | |
| "learning_rate": 8.945702546981969e-05, | |
| "loss": 0.3857, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.1718213058419244, | |
| "eval_loss": 0.38847917318344116, | |
| "eval_runtime": 38.9339, | |
| "eval_samples_per_second": 12.585, | |
| "eval_steps_per_second": 3.159, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.17525773195876287, | |
| "grad_norm": 0.32187727093696594, | |
| "learning_rate": 8.894386393810563e-05, | |
| "loss": 0.5475, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.17869415807560138, | |
| "grad_norm": 0.2936631143093109, | |
| "learning_rate": 8.842005554284296e-05, | |
| "loss": 0.495, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.18213058419243985, | |
| "grad_norm": 0.2527267634868622, | |
| "learning_rate": 8.788574348801675e-05, | |
| "loss": 0.4335, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.18556701030927836, | |
| "grad_norm": 0.22823600471019745, | |
| "learning_rate": 8.73410738492077e-05, | |
| "loss": 0.3888, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.18900343642611683, | |
| "grad_norm": 0.239506334066391, | |
| "learning_rate": 8.678619553365659e-05, | |
| "loss": 0.4601, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.19243986254295534, | |
| "grad_norm": 0.2324398010969162, | |
| "learning_rate": 8.622126023955446e-05, | |
| "loss": 0.3477, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.1958762886597938, | |
| "grad_norm": 0.22715219855308533, | |
| "learning_rate": 8.564642241456986e-05, | |
| "loss": 0.3189, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.19931271477663232, | |
| "grad_norm": 0.25551649928092957, | |
| "learning_rate": 8.506183921362443e-05, | |
| "loss": 0.41, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2027491408934708, | |
| "grad_norm": 0.25390133261680603, | |
| "learning_rate": 8.44676704559283e-05, | |
| "loss": 0.3671, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.20618556701030927, | |
| "grad_norm": 0.25685325264930725, | |
| "learning_rate": 8.386407858128706e-05, | |
| "loss": 0.3913, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.20962199312714777, | |
| "grad_norm": 0.2510710060596466, | |
| "learning_rate": 8.32512286056924e-05, | |
| "loss": 0.3536, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.21305841924398625, | |
| "grad_norm": 0.2768595814704895, | |
| "learning_rate": 8.262928807620843e-05, | |
| "loss": 0.3931, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.21649484536082475, | |
| "grad_norm": 0.2498641163110733, | |
| "learning_rate": 8.199842702516583e-05, | |
| "loss": 0.2963, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.21993127147766323, | |
| "grad_norm": 0.269641250371933, | |
| "learning_rate": 8.135881792367686e-05, | |
| "loss": 0.3781, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.22336769759450173, | |
| "grad_norm": 0.27110132575035095, | |
| "learning_rate": 8.07106356344834e-05, | |
| "loss": 0.3668, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.2268041237113402, | |
| "grad_norm": 0.2595538794994354, | |
| "learning_rate": 8.005405736415126e-05, | |
| "loss": 0.3224, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.23024054982817868, | |
| "grad_norm": 0.2633829712867737, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.3319, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.23367697594501718, | |
| "grad_norm": 0.31550440192222595, | |
| "learning_rate": 7.871643313414718e-05, | |
| "loss": 0.383, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.23711340206185566, | |
| "grad_norm": 0.27217811346054077, | |
| "learning_rate": 7.803575286758364e-05, | |
| "loss": 0.3109, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.24054982817869416, | |
| "grad_norm": 0.2758517265319824, | |
| "learning_rate": 7.734740790612136e-05, | |
| "loss": 0.3148, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.24398625429553264, | |
| "grad_norm": 0.26816055178642273, | |
| "learning_rate": 7.66515864363997e-05, | |
| "loss": 0.3261, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.24742268041237114, | |
| "grad_norm": 0.28605079650878906, | |
| "learning_rate": 7.594847868906076e-05, | |
| "loss": 0.3367, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.2508591065292096, | |
| "grad_norm": 0.3040841221809387, | |
| "learning_rate": 7.52382768867422e-05, | |
| "loss": 0.3138, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.2542955326460481, | |
| "grad_norm": 0.32193273305892944, | |
| "learning_rate": 7.452117519152542e-05, | |
| "loss": 0.3349, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.25773195876288657, | |
| "grad_norm": 0.3038463294506073, | |
| "learning_rate": 7.379736965185368e-05, | |
| "loss": 0.3179, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2611683848797251, | |
| "grad_norm": 0.36102867126464844, | |
| "learning_rate": 7.30670581489344e-05, | |
| "loss": 0.3629, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2646048109965636, | |
| "grad_norm": 0.31307026743888855, | |
| "learning_rate": 7.233044034264034e-05, | |
| "loss": 0.3498, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.26804123711340205, | |
| "grad_norm": 0.30272242426872253, | |
| "learning_rate": 7.158771761692464e-05, | |
| "loss": 0.3238, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.27147766323024053, | |
| "grad_norm": 0.3155359625816345, | |
| "learning_rate": 7.083909302476453e-05, | |
| "loss": 0.3326, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.27491408934707906, | |
| "grad_norm": 0.33321860432624817, | |
| "learning_rate": 7.008477123264848e-05, | |
| "loss": 0.3419, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.27835051546391754, | |
| "grad_norm": 0.34092503786087036, | |
| "learning_rate": 6.932495846462261e-05, | |
| "loss": 0.3541, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.281786941580756, | |
| "grad_norm": 0.34512829780578613, | |
| "learning_rate": 6.855986244591104e-05, | |
| "loss": 0.3774, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2852233676975945, | |
| "grad_norm": 0.347236692905426, | |
| "learning_rate": 6.778969234612584e-05, | |
| "loss": 0.3677, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.28865979381443296, | |
| "grad_norm": 0.32511982321739197, | |
| "learning_rate": 6.701465872208216e-05, | |
| "loss": 0.3219, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.2920962199312715, | |
| "grad_norm": 0.3431488573551178, | |
| "learning_rate": 6.623497346023418e-05, | |
| "loss": 0.3346, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.29553264604810997, | |
| "grad_norm": 0.3518548309803009, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.3584, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.29896907216494845, | |
| "grad_norm": 0.36024248600006104, | |
| "learning_rate": 6.466250186922325e-05, | |
| "loss": 0.3661, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3024054982817869, | |
| "grad_norm": 0.3647770285606384, | |
| "learning_rate": 6.387014543809223e-05, | |
| "loss": 0.3545, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.30584192439862545, | |
| "grad_norm": 0.36317867040634155, | |
| "learning_rate": 6.307399704769099e-05, | |
| "loss": 0.3932, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.30927835051546393, | |
| "grad_norm": 0.3579953610897064, | |
| "learning_rate": 6.227427435703997e-05, | |
| "loss": 0.3292, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3127147766323024, | |
| "grad_norm": 0.39782488346099854, | |
| "learning_rate": 6.147119600233758e-05, | |
| "loss": 0.3742, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.3161512027491409, | |
| "grad_norm": 0.3665030300617218, | |
| "learning_rate": 6.066498153718735e-05, | |
| "loss": 0.3307, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.31958762886597936, | |
| "grad_norm": 0.40009766817092896, | |
| "learning_rate": 5.985585137257401e-05, | |
| "loss": 0.3847, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.3230240549828179, | |
| "grad_norm": 0.3672364354133606, | |
| "learning_rate": 5.90440267166055e-05, | |
| "loss": 0.2994, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.32646048109965636, | |
| "grad_norm": 0.3939048945903778, | |
| "learning_rate": 5.8229729514036705e-05, | |
| "loss": 0.3579, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.32989690721649484, | |
| "grad_norm": 0.4292741119861603, | |
| "learning_rate": 5.74131823855921e-05, | |
| "loss": 0.3323, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.42444631457328796, | |
| "learning_rate": 5.6594608567103456e-05, | |
| "loss": 0.3379, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.33676975945017185, | |
| "grad_norm": 0.4796450436115265, | |
| "learning_rate": 5.577423184847932e-05, | |
| "loss": 0.3793, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.3402061855670103, | |
| "grad_norm": 0.45188021659851074, | |
| "learning_rate": 5.495227651252315e-05, | |
| "loss": 0.362, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.3436426116838488, | |
| "grad_norm": 0.4613451659679413, | |
| "learning_rate": 5.4128967273616625e-05, | |
| "loss": 0.3426, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3436426116838488, | |
| "eval_loss": 0.3695892095565796, | |
| "eval_runtime": 38.9229, | |
| "eval_samples_per_second": 12.589, | |
| "eval_steps_per_second": 3.16, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.3470790378006873, | |
| "grad_norm": 0.24407999217510223, | |
| "learning_rate": 5.330452921628497e-05, | |
| "loss": 0.4208, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.35051546391752575, | |
| "grad_norm": 0.2361270934343338, | |
| "learning_rate": 5.247918773366112e-05, | |
| "loss": 0.4136, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.3539518900343643, | |
| "grad_norm": 0.24302305281162262, | |
| "learning_rate": 5.165316846586541e-05, | |
| "loss": 0.4044, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.35738831615120276, | |
| "grad_norm": 0.22112108767032623, | |
| "learning_rate": 5.0826697238317935e-05, | |
| "loss": 0.3555, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.36082474226804123, | |
| "grad_norm": 0.2189406156539917, | |
| "learning_rate": 5e-05, | |
| "loss": 0.3777, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.3642611683848797, | |
| "grad_norm": 0.2311684489250183, | |
| "learning_rate": 4.917330276168208e-05, | |
| "loss": 0.4014, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.36769759450171824, | |
| "grad_norm": 0.2109665721654892, | |
| "learning_rate": 4.834683153413459e-05, | |
| "loss": 0.3139, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3711340206185567, | |
| "grad_norm": 0.2351345419883728, | |
| "learning_rate": 4.7520812266338885e-05, | |
| "loss": 0.4292, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3745704467353952, | |
| "grad_norm": 0.24277913570404053, | |
| "learning_rate": 4.669547078371504e-05, | |
| "loss": 0.3879, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.37800687285223367, | |
| "grad_norm": 0.25869399309158325, | |
| "learning_rate": 4.5871032726383386e-05, | |
| "loss": 0.3807, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.38144329896907214, | |
| "grad_norm": 0.2533753216266632, | |
| "learning_rate": 4.504772348747687e-05, | |
| "loss": 0.3833, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.3848797250859107, | |
| "grad_norm": 0.23109927773475647, | |
| "learning_rate": 4.4225768151520694e-05, | |
| "loss": 0.3085, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.38831615120274915, | |
| "grad_norm": 0.2492460310459137, | |
| "learning_rate": 4.3405391432896555e-05, | |
| "loss": 0.34, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.3917525773195876, | |
| "grad_norm": 0.25698450207710266, | |
| "learning_rate": 4.2586817614407895e-05, | |
| "loss": 0.3636, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.3951890034364261, | |
| "grad_norm": 0.2784944772720337, | |
| "learning_rate": 4.17702704859633e-05, | |
| "loss": 0.393, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.39862542955326463, | |
| "grad_norm": 0.2814372181892395, | |
| "learning_rate": 4.095597328339452e-05, | |
| "loss": 0.317, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4020618556701031, | |
| "grad_norm": 0.23726582527160645, | |
| "learning_rate": 4.0144148627425993e-05, | |
| "loss": 0.2839, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4054982817869416, | |
| "grad_norm": 0.2546195089817047, | |
| "learning_rate": 3.933501846281267e-05, | |
| "loss": 0.3287, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.40893470790378006, | |
| "grad_norm": 0.2537376880645752, | |
| "learning_rate": 3.852880399766243e-05, | |
| "loss": 0.3108, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.41237113402061853, | |
| "grad_norm": 0.2786731719970703, | |
| "learning_rate": 3.772572564296005e-05, | |
| "loss": 0.3214, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.41580756013745707, | |
| "grad_norm": 0.2733478546142578, | |
| "learning_rate": 3.6926002952309016e-05, | |
| "loss": 0.3153, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.41924398625429554, | |
| "grad_norm": 0.2947427034378052, | |
| "learning_rate": 3.612985456190778e-05, | |
| "loss": 0.3336, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.422680412371134, | |
| "grad_norm": 0.26663312315940857, | |
| "learning_rate": 3.533749813077677e-05, | |
| "loss": 0.3148, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.4261168384879725, | |
| "grad_norm": 0.2836472690105438, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.3274, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.42955326460481097, | |
| "grad_norm": 0.2884660065174103, | |
| "learning_rate": 3.3765026539765834e-05, | |
| "loss": 0.3378, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.4329896907216495, | |
| "grad_norm": 0.3011506199836731, | |
| "learning_rate": 3.298534127791785e-05, | |
| "loss": 0.3511, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.436426116838488, | |
| "grad_norm": 0.2971234619617462, | |
| "learning_rate": 3.221030765387417e-05, | |
| "loss": 0.3377, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.43986254295532645, | |
| "grad_norm": 0.28751340508461, | |
| "learning_rate": 3.144013755408895e-05, | |
| "loss": 0.3262, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.44329896907216493, | |
| "grad_norm": 0.32300904393196106, | |
| "learning_rate": 3.0675041535377405e-05, | |
| "loss": 0.3357, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.44673539518900346, | |
| "grad_norm": 0.28600189089775085, | |
| "learning_rate": 2.991522876735154e-05, | |
| "loss": 0.2971, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.45017182130584193, | |
| "grad_norm": 0.3008926212787628, | |
| "learning_rate": 2.916090697523549e-05, | |
| "loss": 0.328, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.4536082474226804, | |
| "grad_norm": 0.28959494829177856, | |
| "learning_rate": 2.8412282383075363e-05, | |
| "loss": 0.3009, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.4570446735395189, | |
| "grad_norm": 0.30336546897888184, | |
| "learning_rate": 2.766955965735968e-05, | |
| "loss": 0.3039, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.46048109965635736, | |
| "grad_norm": 0.3182055950164795, | |
| "learning_rate": 2.693294185106562e-05, | |
| "loss": 0.3407, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.4639175257731959, | |
| "grad_norm": 0.33289095759391785, | |
| "learning_rate": 2.6202630348146324e-05, | |
| "loss": 0.3472, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.46735395189003437, | |
| "grad_norm": 0.33430078625679016, | |
| "learning_rate": 2.547882480847461e-05, | |
| "loss": 0.322, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.47079037800687284, | |
| "grad_norm": 0.32794615626335144, | |
| "learning_rate": 2.476172311325783e-05, | |
| "loss": 0.3087, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4742268041237113, | |
| "grad_norm": 0.35182681679725647, | |
| "learning_rate": 2.405152131093926e-05, | |
| "loss": 0.3488, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.47766323024054985, | |
| "grad_norm": 0.3695390522480011, | |
| "learning_rate": 2.3348413563600325e-05, | |
| "loss": 0.345, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.48109965635738833, | |
| "grad_norm": 0.35909849405288696, | |
| "learning_rate": 2.2652592093878666e-05, | |
| "loss": 0.3391, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.4845360824742268, | |
| "grad_norm": 0.35425567626953125, | |
| "learning_rate": 2.196424713241637e-05, | |
| "loss": 0.3528, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.4879725085910653, | |
| "grad_norm": 0.36728376150131226, | |
| "learning_rate": 2.128356686585282e-05, | |
| "loss": 0.3394, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.49140893470790376, | |
| "grad_norm": 0.3439100980758667, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.3116, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.4948453608247423, | |
| "grad_norm": 0.3803477883338928, | |
| "learning_rate": 1.9945942635848748e-05, | |
| "loss": 0.3494, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.49828178694158076, | |
| "grad_norm": 0.36791571974754333, | |
| "learning_rate": 1.928936436551661e-05, | |
| "loss": 0.3469, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5017182130584192, | |
| "grad_norm": 0.372040718793869, | |
| "learning_rate": 1.8641182076323148e-05, | |
| "loss": 0.3198, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5051546391752577, | |
| "grad_norm": 0.3909667730331421, | |
| "learning_rate": 1.800157297483417e-05, | |
| "loss": 0.3511, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5085910652920962, | |
| "grad_norm": 0.4031165838241577, | |
| "learning_rate": 1.7370711923791567e-05, | |
| "loss": 0.3235, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.5120274914089347, | |
| "grad_norm": 0.39335718750953674, | |
| "learning_rate": 1.6748771394307585e-05, | |
| "loss": 0.3248, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.5154639175257731, | |
| "grad_norm": 0.4456535279750824, | |
| "learning_rate": 1.6135921418712956e-05, | |
| "loss": 0.3335, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5154639175257731, | |
| "eval_loss": 0.35342445969581604, | |
| "eval_runtime": 38.6164, | |
| "eval_samples_per_second": 12.689, | |
| "eval_steps_per_second": 3.185, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5189003436426117, | |
| "grad_norm": 0.19197984039783478, | |
| "learning_rate": 1.553232954407171e-05, | |
| "loss": 0.422, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.5223367697594502, | |
| "grad_norm": 0.21861949563026428, | |
| "learning_rate": 1.4938160786375572e-05, | |
| "loss": 0.509, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.5257731958762887, | |
| "grad_norm": 0.191892609000206, | |
| "learning_rate": 1.435357758543015e-05, | |
| "loss": 0.3431, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.5292096219931272, | |
| "grad_norm": 0.21896794438362122, | |
| "learning_rate": 1.3778739760445552e-05, | |
| "loss": 0.4461, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.5326460481099656, | |
| "grad_norm": 0.2197161316871643, | |
| "learning_rate": 1.3213804466343421e-05, | |
| "loss": 0.3671, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5360824742268041, | |
| "grad_norm": 0.20610028505325317, | |
| "learning_rate": 1.2658926150792322e-05, | |
| "loss": 0.3367, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.5395189003436426, | |
| "grad_norm": 0.2370217889547348, | |
| "learning_rate": 1.2114256511983274e-05, | |
| "loss": 0.4019, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.5429553264604811, | |
| "grad_norm": 0.22654864192008972, | |
| "learning_rate": 1.157994445715706e-05, | |
| "loss": 0.3634, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.5463917525773195, | |
| "grad_norm": 0.24236921966075897, | |
| "learning_rate": 1.1056136061894384e-05, | |
| "loss": 0.3944, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.5498281786941581, | |
| "grad_norm": 0.23310494422912598, | |
| "learning_rate": 1.0542974530180327e-05, | |
| "loss": 0.3492, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5532646048109966, | |
| "grad_norm": 0.2678241431713104, | |
| "learning_rate": 1.0040600155253765e-05, | |
| "loss": 0.3862, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.5567010309278351, | |
| "grad_norm": 0.2302623689174652, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.3071, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5601374570446735, | |
| "grad_norm": 0.22068384289741516, | |
| "learning_rate": 9.068759265665384e-06, | |
| "loss": 0.2984, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.563573883161512, | |
| "grad_norm": 0.21957820653915405, | |
| "learning_rate": 8.599558442598998e-06, | |
| "loss": 0.3295, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.5670103092783505, | |
| "grad_norm": 0.2356969565153122, | |
| "learning_rate": 8.141676086873572e-06, | |
| "loss": 0.3221, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.570446735395189, | |
| "grad_norm": 0.23873502016067505, | |
| "learning_rate": 7.695237378953223e-06, | |
| "loss": 0.3225, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.5738831615120275, | |
| "grad_norm": 0.2684608995914459, | |
| "learning_rate": 7.260364370723044e-06, | |
| "loss": 0.4114, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.5773195876288659, | |
| "grad_norm": 0.25298547744750977, | |
| "learning_rate": 6.837175952121306e-06, | |
| "loss": 0.3003, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.5807560137457045, | |
| "grad_norm": 0.2470017522573471, | |
| "learning_rate": 6.425787818636131e-06, | |
| "loss": 0.3279, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.584192439862543, | |
| "grad_norm": 0.2631200850009918, | |
| "learning_rate": 6.026312439675552e-06, | |
| "loss": 0.3561, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.5876288659793815, | |
| "grad_norm": 0.27026766538619995, | |
| "learning_rate": 5.6388590278194096e-06, | |
| "loss": 0.3044, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.5910652920962199, | |
| "grad_norm": 0.2582063376903534, | |
| "learning_rate": 5.263533508961827e-06, | |
| "loss": 0.3315, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.5945017182130584, | |
| "grad_norm": 0.27334722876548767, | |
| "learning_rate": 4.900438493352055e-06, | |
| "loss": 0.3588, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.5979381443298969, | |
| "grad_norm": 0.2814502716064453, | |
| "learning_rate": 4.549673247541875e-06, | |
| "loss": 0.3332, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.6013745704467354, | |
| "grad_norm": 0.257357120513916, | |
| "learning_rate": 4.2113336672471245e-06, | |
| "loss": 0.2865, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6048109965635738, | |
| "grad_norm": 0.29060375690460205, | |
| "learning_rate": 3.885512251130763e-06, | |
| "loss": 0.3319, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6082474226804123, | |
| "grad_norm": 0.28878647089004517, | |
| "learning_rate": 3.5722980755146517e-06, | |
| "loss": 0.3641, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.6116838487972509, | |
| "grad_norm": 0.2936228811740875, | |
| "learning_rate": 3.271776770026963e-06, | |
| "loss": 0.3436, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.6151202749140894, | |
| "grad_norm": 0.3015434741973877, | |
| "learning_rate": 2.9840304941919415e-06, | |
| "loss": 0.3722, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.6185567010309279, | |
| "grad_norm": 0.2890324294567108, | |
| "learning_rate": 2.7091379149682685e-06, | |
| "loss": 0.3297, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6219931271477663, | |
| "grad_norm": 0.3039141297340393, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.3549, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.6254295532646048, | |
| "grad_norm": 0.3125923275947571, | |
| "learning_rate": 2.1982109232821178e-06, | |
| "loss": 0.3352, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.6288659793814433, | |
| "grad_norm": 0.2943466603755951, | |
| "learning_rate": 1.962316193157593e-06, | |
| "loss": 0.302, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.6323024054982818, | |
| "grad_norm": 0.33996695280075073, | |
| "learning_rate": 1.7395544861325718e-06, | |
| "loss": 0.3672, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.6357388316151202, | |
| "grad_norm": 0.32115960121154785, | |
| "learning_rate": 1.5299867030334814e-06, | |
| "loss": 0.3209, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.6391752577319587, | |
| "grad_norm": 0.3367302715778351, | |
| "learning_rate": 1.333670137599713e-06, | |
| "loss": 0.3409, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.6426116838487973, | |
| "grad_norm": 0.33328408002853394, | |
| "learning_rate": 1.1506584608200367e-06, | |
| "loss": 0.3786, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.6460481099656358, | |
| "grad_norm": 0.3016452491283417, | |
| "learning_rate": 9.810017062595322e-07, | |
| "loss": 0.2946, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.6494845360824743, | |
| "grad_norm": 0.3220769762992859, | |
| "learning_rate": 8.247462563808817e-07, | |
| "loss": 0.309, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.6529209621993127, | |
| "grad_norm": 0.33597269654273987, | |
| "learning_rate": 6.819348298638839e-07, | |
| "loss": 0.3244, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6563573883161512, | |
| "grad_norm": 0.32418060302734375, | |
| "learning_rate": 5.526064699265753e-07, | |
| "loss": 0.3025, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.6597938144329897, | |
| "grad_norm": 0.35037267208099365, | |
| "learning_rate": 4.367965336512403e-07, | |
| "loss": 0.3441, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.6632302405498282, | |
| "grad_norm": 0.37282636761665344, | |
| "learning_rate": 3.3453668231809286e-07, | |
| "loss": 0.3251, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.34526291489601135, | |
| "learning_rate": 2.458548727494292e-07, | |
| "loss": 0.3168, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.6701030927835051, | |
| "grad_norm": 0.3517741858959198, | |
| "learning_rate": 1.7077534966650766e-07, | |
| "loss": 0.2909, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.6735395189003437, | |
| "grad_norm": 0.3964909613132477, | |
| "learning_rate": 1.0931863906127327e-07, | |
| "loss": 0.3364, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.6769759450171822, | |
| "grad_norm": 0.3922007381916046, | |
| "learning_rate": 6.150154258476315e-08, | |
| "loss": 0.3208, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.6804123711340206, | |
| "grad_norm": 0.4229685366153717, | |
| "learning_rate": 2.7337132953697554e-08, | |
| "loss": 0.3542, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.6838487972508591, | |
| "grad_norm": 0.436335027217865, | |
| "learning_rate": 6.834750376549792e-09, | |
| "loss": 0.383, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.6872852233676976, | |
| "grad_norm": 0.5090466737747192, | |
| "learning_rate": 0.0, | |
| "loss": 0.357, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.6872852233676976, | |
| "eval_loss": 0.35021281242370605, | |
| "eval_runtime": 38.3433, | |
| "eval_samples_per_second": 12.779, | |
| "eval_steps_per_second": 3.208, | |
| "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": 3.062282564854088e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |