| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 0, | |
| "global_step": 277, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0036101083032490976, | |
| "grad_norm": 0.8931787610054016, | |
| "learning_rate": 1e-05, | |
| "loss": 2.3455, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.007220216606498195, | |
| "grad_norm": 0.882213294506073, | |
| "learning_rate": 9.96389891696751e-06, | |
| "loss": 2.3626, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.010830324909747292, | |
| "grad_norm": 0.8512553572654724, | |
| "learning_rate": 9.92779783393502e-06, | |
| "loss": 2.3614, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01444043321299639, | |
| "grad_norm": 0.8682896494865417, | |
| "learning_rate": 9.891696750902527e-06, | |
| "loss": 2.3861, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.018050541516245487, | |
| "grad_norm": 0.8518972396850586, | |
| "learning_rate": 9.855595667870036e-06, | |
| "loss": 2.4183, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.021660649819494584, | |
| "grad_norm": 0.8164824843406677, | |
| "learning_rate": 9.819494584837546e-06, | |
| "loss": 2.354, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.02527075812274368, | |
| "grad_norm": 0.775917112827301, | |
| "learning_rate": 9.783393501805055e-06, | |
| "loss": 2.3267, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.02888086642599278, | |
| "grad_norm": 0.735588550567627, | |
| "learning_rate": 9.747292418772564e-06, | |
| "loss": 2.303, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.032490974729241874, | |
| "grad_norm": 0.7198289632797241, | |
| "learning_rate": 9.711191335740074e-06, | |
| "loss": 2.3494, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.036101083032490974, | |
| "grad_norm": 0.6802542209625244, | |
| "learning_rate": 9.675090252707581e-06, | |
| "loss": 2.2845, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.039711191335740074, | |
| "grad_norm": 0.6937660574913025, | |
| "learning_rate": 9.63898916967509e-06, | |
| "loss": 2.2922, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.04332129963898917, | |
| "grad_norm": 0.6162929534912109, | |
| "learning_rate": 9.6028880866426e-06, | |
| "loss": 2.2021, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.04693140794223827, | |
| "grad_norm": 0.60593181848526, | |
| "learning_rate": 9.56678700361011e-06, | |
| "loss": 2.2494, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.05054151624548736, | |
| "grad_norm": 0.6108725070953369, | |
| "learning_rate": 9.530685920577619e-06, | |
| "loss": 2.2608, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.05415162454873646, | |
| "grad_norm": 0.5801817774772644, | |
| "learning_rate": 9.494584837545126e-06, | |
| "loss": 2.225, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.05776173285198556, | |
| "grad_norm": 0.5604182481765747, | |
| "learning_rate": 9.458483754512636e-06, | |
| "loss": 2.1908, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.061371841155234655, | |
| "grad_norm": 0.579058051109314, | |
| "learning_rate": 9.422382671480145e-06, | |
| "loss": 2.2386, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.06498194945848375, | |
| "grad_norm": 0.550123393535614, | |
| "learning_rate": 9.386281588447654e-06, | |
| "loss": 2.2308, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.06859205776173286, | |
| "grad_norm": 0.5217147469520569, | |
| "learning_rate": 9.350180505415164e-06, | |
| "loss": 2.1533, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.07220216606498195, | |
| "grad_norm": 0.5268492102622986, | |
| "learning_rate": 9.314079422382673e-06, | |
| "loss": 2.1248, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.07581227436823104, | |
| "grad_norm": 0.4969961941242218, | |
| "learning_rate": 9.27797833935018e-06, | |
| "loss": 2.1072, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.07942238267148015, | |
| "grad_norm": 0.48244956135749817, | |
| "learning_rate": 9.24187725631769e-06, | |
| "loss": 2.0037, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.08303249097472924, | |
| "grad_norm": 0.516872227191925, | |
| "learning_rate": 9.2057761732852e-06, | |
| "loss": 2.0801, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.08664259927797834, | |
| "grad_norm": 0.5113118886947632, | |
| "learning_rate": 9.169675090252709e-06, | |
| "loss": 2.1388, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.09025270758122744, | |
| "grad_norm": 0.5139282941818237, | |
| "learning_rate": 9.133574007220218e-06, | |
| "loss": 2.1148, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.09386281588447654, | |
| "grad_norm": 0.5160923004150391, | |
| "learning_rate": 9.097472924187727e-06, | |
| "loss": 2.1511, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.09747292418772563, | |
| "grad_norm": 0.4702168107032776, | |
| "learning_rate": 9.061371841155235e-06, | |
| "loss": 2.0701, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.10108303249097472, | |
| "grad_norm": 0.5095406770706177, | |
| "learning_rate": 9.025270758122744e-06, | |
| "loss": 2.1188, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.10469314079422383, | |
| "grad_norm": 0.4923248589038849, | |
| "learning_rate": 8.989169675090254e-06, | |
| "loss": 2.1105, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.10830324909747292, | |
| "grad_norm": 0.49832674860954285, | |
| "learning_rate": 8.953068592057763e-06, | |
| "loss": 2.1197, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.11191335740072202, | |
| "grad_norm": 0.46398353576660156, | |
| "learning_rate": 8.916967509025272e-06, | |
| "loss": 2.0533, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.11552346570397112, | |
| "grad_norm": 0.43949949741363525, | |
| "learning_rate": 8.88086642599278e-06, | |
| "loss": 2.0422, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.11913357400722022, | |
| "grad_norm": 0.4486154615879059, | |
| "learning_rate": 8.84476534296029e-06, | |
| "loss": 1.9696, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.12274368231046931, | |
| "grad_norm": 0.4580928385257721, | |
| "learning_rate": 8.808664259927798e-06, | |
| "loss": 1.9638, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.1263537906137184, | |
| "grad_norm": 0.43987998366355896, | |
| "learning_rate": 8.772563176895308e-06, | |
| "loss": 2.0338, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1299638989169675, | |
| "grad_norm": 0.4130140244960785, | |
| "learning_rate": 8.736462093862817e-06, | |
| "loss": 1.9947, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.13357400722021662, | |
| "grad_norm": 0.4416247010231018, | |
| "learning_rate": 8.700361010830326e-06, | |
| "loss": 2.0282, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.1371841155234657, | |
| "grad_norm": 0.3853282034397125, | |
| "learning_rate": 8.664259927797834e-06, | |
| "loss": 1.946, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.1407942238267148, | |
| "grad_norm": 0.4156254827976227, | |
| "learning_rate": 8.628158844765343e-06, | |
| "loss": 1.9734, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.1444043321299639, | |
| "grad_norm": 0.4245947003364563, | |
| "learning_rate": 8.592057761732853e-06, | |
| "loss": 2.0125, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.148014440433213, | |
| "grad_norm": 0.3952503502368927, | |
| "learning_rate": 8.55595667870036e-06, | |
| "loss": 1.9628, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.15162454873646208, | |
| "grad_norm": 0.4200592637062073, | |
| "learning_rate": 8.519855595667871e-06, | |
| "loss": 1.958, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.1552346570397112, | |
| "grad_norm": 0.414566308259964, | |
| "learning_rate": 8.483754512635379e-06, | |
| "loss": 1.9864, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.1588447653429603, | |
| "grad_norm": 0.38787394762039185, | |
| "learning_rate": 8.447653429602888e-06, | |
| "loss": 1.9111, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1624548736462094, | |
| "grad_norm": 0.3872066140174866, | |
| "learning_rate": 8.411552346570398e-06, | |
| "loss": 1.9536, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.16606498194945848, | |
| "grad_norm": 0.3785644769668579, | |
| "learning_rate": 8.375451263537907e-06, | |
| "loss": 1.9339, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.16967509025270758, | |
| "grad_norm": 0.3656150698661804, | |
| "learning_rate": 8.339350180505416e-06, | |
| "loss": 1.898, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.17328519855595667, | |
| "grad_norm": 0.355506032705307, | |
| "learning_rate": 8.303249097472926e-06, | |
| "loss": 1.8982, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.17689530685920576, | |
| "grad_norm": 0.3854127824306488, | |
| "learning_rate": 8.267148014440433e-06, | |
| "loss": 1.9163, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.18050541516245489, | |
| "grad_norm": 0.3902463912963867, | |
| "learning_rate": 8.231046931407943e-06, | |
| "loss": 1.8881, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.18411552346570398, | |
| "grad_norm": 0.38277703523635864, | |
| "learning_rate": 8.194945848375452e-06, | |
| "loss": 1.9762, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.18772563176895307, | |
| "grad_norm": 0.3710600435733795, | |
| "learning_rate": 8.158844765342961e-06, | |
| "loss": 1.902, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.19133574007220217, | |
| "grad_norm": 0.4147663116455078, | |
| "learning_rate": 8.12274368231047e-06, | |
| "loss": 1.9431, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.19494584837545126, | |
| "grad_norm": 0.40643152594566345, | |
| "learning_rate": 8.086642599277978e-06, | |
| "loss": 1.9789, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.19855595667870035, | |
| "grad_norm": 0.3781704902648926, | |
| "learning_rate": 8.050541516245488e-06, | |
| "loss": 1.9064, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.20216606498194944, | |
| "grad_norm": 0.3825320303440094, | |
| "learning_rate": 8.014440433212997e-06, | |
| "loss": 1.8291, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.20577617328519857, | |
| "grad_norm": 0.3635013997554779, | |
| "learning_rate": 7.978339350180506e-06, | |
| "loss": 1.8746, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.20938628158844766, | |
| "grad_norm": 0.3744393289089203, | |
| "learning_rate": 7.942238267148014e-06, | |
| "loss": 1.8553, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.21299638989169675, | |
| "grad_norm": 0.3601558804512024, | |
| "learning_rate": 7.906137184115525e-06, | |
| "loss": 1.829, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.21660649819494585, | |
| "grad_norm": 0.37339311838150024, | |
| "learning_rate": 7.870036101083033e-06, | |
| "loss": 1.8379, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.22021660649819494, | |
| "grad_norm": 0.36436107754707336, | |
| "learning_rate": 7.833935018050542e-06, | |
| "loss": 1.8441, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.22382671480144403, | |
| "grad_norm": 0.3610350787639618, | |
| "learning_rate": 7.797833935018051e-06, | |
| "loss": 1.8066, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.22743682310469315, | |
| "grad_norm": 0.3610876202583313, | |
| "learning_rate": 7.76173285198556e-06, | |
| "loss": 1.8493, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.23104693140794225, | |
| "grad_norm": 0.3644372522830963, | |
| "learning_rate": 7.72563176895307e-06, | |
| "loss": 1.8307, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.23465703971119134, | |
| "grad_norm": 0.34159204363822937, | |
| "learning_rate": 7.68953068592058e-06, | |
| "loss": 1.7994, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.23826714801444043, | |
| "grad_norm": 0.3353036046028137, | |
| "learning_rate": 7.653429602888087e-06, | |
| "loss": 1.7779, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.24187725631768953, | |
| "grad_norm": 0.3560786843299866, | |
| "learning_rate": 7.617328519855596e-06, | |
| "loss": 1.8221, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.24548736462093862, | |
| "grad_norm": 0.3391997814178467, | |
| "learning_rate": 7.5812274368231055e-06, | |
| "loss": 1.812, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.2490974729241877, | |
| "grad_norm": 0.3948769271373749, | |
| "learning_rate": 7.545126353790614e-06, | |
| "loss": 1.8202, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.2527075812274368, | |
| "grad_norm": 0.3438505232334137, | |
| "learning_rate": 7.509025270758123e-06, | |
| "loss": 1.7723, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.2563176895306859, | |
| "grad_norm": 0.38049331307411194, | |
| "learning_rate": 7.472924187725632e-06, | |
| "loss": 1.7967, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.259927797833935, | |
| "grad_norm": 0.3355187773704529, | |
| "learning_rate": 7.436823104693142e-06, | |
| "loss": 1.78, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.26353790613718414, | |
| "grad_norm": 0.3481115698814392, | |
| "learning_rate": 7.40072202166065e-06, | |
| "loss": 1.8357, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.26714801444043323, | |
| "grad_norm": 0.3510536849498749, | |
| "learning_rate": 7.36462093862816e-06, | |
| "loss": 1.8351, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.27075812274368233, | |
| "grad_norm": 0.3468690514564514, | |
| "learning_rate": 7.328519855595668e-06, | |
| "loss": 1.8028, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.2743682310469314, | |
| "grad_norm": 0.34152960777282715, | |
| "learning_rate": 7.2924187725631776e-06, | |
| "loss": 1.7492, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.2779783393501805, | |
| "grad_norm": 0.3646658957004547, | |
| "learning_rate": 7.256317689530686e-06, | |
| "loss": 1.8215, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.2815884476534296, | |
| "grad_norm": 0.3506210148334503, | |
| "learning_rate": 7.220216606498196e-06, | |
| "loss": 1.7737, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.2851985559566787, | |
| "grad_norm": 0.3491074740886688, | |
| "learning_rate": 7.184115523465705e-06, | |
| "loss": 1.7873, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.2888086642599278, | |
| "grad_norm": 0.3594192564487457, | |
| "learning_rate": 7.148014440433214e-06, | |
| "loss": 1.7768, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.2924187725631769, | |
| "grad_norm": 0.3775302469730377, | |
| "learning_rate": 7.1119133574007225e-06, | |
| "loss": 1.815, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.296028880866426, | |
| "grad_norm": 0.3560231626033783, | |
| "learning_rate": 7.075812274368231e-06, | |
| "loss": 1.7757, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.2996389891696751, | |
| "grad_norm": 0.3486398458480835, | |
| "learning_rate": 7.039711191335741e-06, | |
| "loss": 1.8329, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.30324909747292417, | |
| "grad_norm": 0.43985557556152344, | |
| "learning_rate": 7.00361010830325e-06, | |
| "loss": 1.7659, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.30685920577617326, | |
| "grad_norm": 0.3351285755634308, | |
| "learning_rate": 6.967509025270759e-06, | |
| "loss": 1.7274, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.3104693140794224, | |
| "grad_norm": 0.3412298858165741, | |
| "learning_rate": 6.9314079422382674e-06, | |
| "loss": 1.7834, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.3140794223826715, | |
| "grad_norm": 0.34331032633781433, | |
| "learning_rate": 6.895306859205777e-06, | |
| "loss": 1.7608, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.3176895306859206, | |
| "grad_norm": 0.33744579553604126, | |
| "learning_rate": 6.859205776173285e-06, | |
| "loss": 1.6853, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3212996389891697, | |
| "grad_norm": 0.3468646705150604, | |
| "learning_rate": 6.8231046931407954e-06, | |
| "loss": 1.7376, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3249097472924188, | |
| "grad_norm": 0.3477930724620819, | |
| "learning_rate": 6.787003610108304e-06, | |
| "loss": 1.6986, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.3285198555956679, | |
| "grad_norm": 0.32764512300491333, | |
| "learning_rate": 6.750902527075813e-06, | |
| "loss": 1.7399, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.33212996389891697, | |
| "grad_norm": 0.37140509486198425, | |
| "learning_rate": 6.714801444043322e-06, | |
| "loss": 1.7066, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.33574007220216606, | |
| "grad_norm": 0.3442263603210449, | |
| "learning_rate": 6.678700361010831e-06, | |
| "loss": 1.7482, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.33935018050541516, | |
| "grad_norm": 0.335580050945282, | |
| "learning_rate": 6.6425992779783395e-06, | |
| "loss": 1.6956, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.34296028880866425, | |
| "grad_norm": 0.3378622531890869, | |
| "learning_rate": 6.606498194945848e-06, | |
| "loss": 1.7578, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.34657039711191334, | |
| "grad_norm": 0.3609424829483032, | |
| "learning_rate": 6.570397111913358e-06, | |
| "loss": 1.7366, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.35018050541516244, | |
| "grad_norm": 0.36335399746894836, | |
| "learning_rate": 6.534296028880867e-06, | |
| "loss": 1.7315, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.35379061371841153, | |
| "grad_norm": 0.31665071845054626, | |
| "learning_rate": 6.498194945848376e-06, | |
| "loss": 1.6854, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.3574007220216607, | |
| "grad_norm": 0.32897570729255676, | |
| "learning_rate": 6.4620938628158845e-06, | |
| "loss": 1.7269, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.36101083032490977, | |
| "grad_norm": 0.31995493173599243, | |
| "learning_rate": 6.425992779783395e-06, | |
| "loss": 1.7104, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.36462093862815886, | |
| "grad_norm": 0.3097304701805115, | |
| "learning_rate": 6.389891696750903e-06, | |
| "loss": 1.7223, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.36823104693140796, | |
| "grad_norm": 0.31838735938072205, | |
| "learning_rate": 6.3537906137184125e-06, | |
| "loss": 1.7211, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.37184115523465705, | |
| "grad_norm": 0.3398403823375702, | |
| "learning_rate": 6.317689530685921e-06, | |
| "loss": 1.7942, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.37545126353790614, | |
| "grad_norm": 0.3619037866592407, | |
| "learning_rate": 6.28158844765343e-06, | |
| "loss": 1.7575, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.37906137184115524, | |
| "grad_norm": 0.3106141984462738, | |
| "learning_rate": 6.245487364620939e-06, | |
| "loss": 1.7166, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.38267148014440433, | |
| "grad_norm": 0.33814042806625366, | |
| "learning_rate": 6.209386281588449e-06, | |
| "loss": 1.7097, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.3862815884476534, | |
| "grad_norm": 0.31877759099006653, | |
| "learning_rate": 6.173285198555957e-06, | |
| "loss": 1.7069, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.3898916967509025, | |
| "grad_norm": 0.31549936532974243, | |
| "learning_rate": 6.137184115523466e-06, | |
| "loss": 1.6778, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.3935018050541516, | |
| "grad_norm": 0.2897959351539612, | |
| "learning_rate": 6.101083032490975e-06, | |
| "loss": 1.6649, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.3971119133574007, | |
| "grad_norm": 0.3396201431751251, | |
| "learning_rate": 6.064981949458484e-06, | |
| "loss": 1.6476, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.4007220216606498, | |
| "grad_norm": 0.2955738604068756, | |
| "learning_rate": 6.028880866425994e-06, | |
| "loss": 1.7084, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.4043321299638989, | |
| "grad_norm": 0.29493531584739685, | |
| "learning_rate": 5.992779783393502e-06, | |
| "loss": 1.6688, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.40794223826714804, | |
| "grad_norm": 0.29845130443573, | |
| "learning_rate": 5.956678700361012e-06, | |
| "loss": 1.7013, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.41155234657039713, | |
| "grad_norm": 0.3014790713787079, | |
| "learning_rate": 5.92057761732852e-06, | |
| "loss": 1.7228, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.4151624548736462, | |
| "grad_norm": 0.2831939160823822, | |
| "learning_rate": 5.8844765342960295e-06, | |
| "loss": 1.6686, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.4187725631768953, | |
| "grad_norm": 0.29137012362480164, | |
| "learning_rate": 5.848375451263538e-06, | |
| "loss": 1.7134, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.4223826714801444, | |
| "grad_norm": 0.2850097715854645, | |
| "learning_rate": 5.812274368231048e-06, | |
| "loss": 1.6714, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.4259927797833935, | |
| "grad_norm": 0.28906336426734924, | |
| "learning_rate": 5.776173285198557e-06, | |
| "loss": 1.6804, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.4296028880866426, | |
| "grad_norm": 0.2782924473285675, | |
| "learning_rate": 5.740072202166066e-06, | |
| "loss": 1.692, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.4332129963898917, | |
| "grad_norm": 0.2898225486278534, | |
| "learning_rate": 5.7039711191335744e-06, | |
| "loss": 1.699, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.4368231046931408, | |
| "grad_norm": 0.30253660678863525, | |
| "learning_rate": 5.667870036101083e-06, | |
| "loss": 1.6563, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.4404332129963899, | |
| "grad_norm": 0.3063725531101227, | |
| "learning_rate": 5.631768953068592e-06, | |
| "loss": 1.7102, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.44404332129963897, | |
| "grad_norm": 0.2866223156452179, | |
| "learning_rate": 5.595667870036101e-06, | |
| "loss": 1.6335, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.44765342960288806, | |
| "grad_norm": 0.2892088294029236, | |
| "learning_rate": 5.559566787003611e-06, | |
| "loss": 1.6983, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.45126353790613716, | |
| "grad_norm": 0.29790347814559937, | |
| "learning_rate": 5.523465703971119e-06, | |
| "loss": 1.6526, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.4548736462093863, | |
| "grad_norm": 0.3047894239425659, | |
| "learning_rate": 5.487364620938629e-06, | |
| "loss": 1.6814, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.4584837545126354, | |
| "grad_norm": 0.2874986529350281, | |
| "learning_rate": 5.451263537906137e-06, | |
| "loss": 1.6889, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.4620938628158845, | |
| "grad_norm": 0.29620295763015747, | |
| "learning_rate": 5.415162454873647e-06, | |
| "loss": 1.7049, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.4657039711191336, | |
| "grad_norm": 0.29222729802131653, | |
| "learning_rate": 5.379061371841156e-06, | |
| "loss": 1.6348, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.4693140794223827, | |
| "grad_norm": 0.29496854543685913, | |
| "learning_rate": 5.342960288808665e-06, | |
| "loss": 1.6994, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.4729241877256318, | |
| "grad_norm": 0.28537270426750183, | |
| "learning_rate": 5.306859205776174e-06, | |
| "loss": 1.6879, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.47653429602888087, | |
| "grad_norm": 0.3027176558971405, | |
| "learning_rate": 5.270758122743683e-06, | |
| "loss": 1.6945, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.48014440433212996, | |
| "grad_norm": 0.29232731461524963, | |
| "learning_rate": 5.2346570397111915e-06, | |
| "loss": 1.7217, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.48375451263537905, | |
| "grad_norm": 0.3224429488182068, | |
| "learning_rate": 5.1985559566787e-06, | |
| "loss": 1.6952, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.48736462093862815, | |
| "grad_norm": 0.30147045850753784, | |
| "learning_rate": 5.16245487364621e-06, | |
| "loss": 1.6578, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.49097472924187724, | |
| "grad_norm": 0.3153703212738037, | |
| "learning_rate": 5.126353790613719e-06, | |
| "loss": 1.6692, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.49458483754512633, | |
| "grad_norm": 0.285895437002182, | |
| "learning_rate": 5.090252707581228e-06, | |
| "loss": 1.6915, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.4981949458483754, | |
| "grad_norm": 0.290146142244339, | |
| "learning_rate": 5.054151624548736e-06, | |
| "loss": 1.6339, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.5018050541516246, | |
| "grad_norm": 0.29954054951667786, | |
| "learning_rate": 5.018050541516246e-06, | |
| "loss": 1.6713, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.5054151624548736, | |
| "grad_norm": 0.2817152738571167, | |
| "learning_rate": 4.981949458483755e-06, | |
| "loss": 1.6127, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.5090252707581228, | |
| "grad_norm": 0.29228702187538147, | |
| "learning_rate": 4.9458483754512636e-06, | |
| "loss": 1.6296, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.5126353790613718, | |
| "grad_norm": 0.28209373354911804, | |
| "learning_rate": 4.909747292418773e-06, | |
| "loss": 1.6397, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.516245487364621, | |
| "grad_norm": 0.2972622811794281, | |
| "learning_rate": 4.873646209386282e-06, | |
| "loss": 1.6726, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.51985559566787, | |
| "grad_norm": 0.28512051701545715, | |
| "learning_rate": 4.837545126353791e-06, | |
| "loss": 1.629, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.5234657039711191, | |
| "grad_norm": 0.30687668919563293, | |
| "learning_rate": 4.8014440433213e-06, | |
| "loss": 1.6447, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.5270758122743683, | |
| "grad_norm": 0.30815911293029785, | |
| "learning_rate": 4.765342960288809e-06, | |
| "loss": 1.6755, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.5306859205776173, | |
| "grad_norm": 0.27116310596466064, | |
| "learning_rate": 4.729241877256318e-06, | |
| "loss": 1.6323, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.5342960288808665, | |
| "grad_norm": 0.3312828242778778, | |
| "learning_rate": 4.693140794223827e-06, | |
| "loss": 1.6279, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.5379061371841155, | |
| "grad_norm": 0.2883269786834717, | |
| "learning_rate": 4.6570397111913365e-06, | |
| "loss": 1.6407, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.5415162454873647, | |
| "grad_norm": 0.2799761891365051, | |
| "learning_rate": 4.620938628158845e-06, | |
| "loss": 1.5842, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.5451263537906137, | |
| "grad_norm": 0.30059683322906494, | |
| "learning_rate": 4.584837545126354e-06, | |
| "loss": 1.6363, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.5487364620938628, | |
| "grad_norm": 0.2806866466999054, | |
| "learning_rate": 4.548736462093864e-06, | |
| "loss": 1.6374, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.5523465703971119, | |
| "grad_norm": 0.28719037771224976, | |
| "learning_rate": 4.512635379061372e-06, | |
| "loss": 1.652, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.555956678700361, | |
| "grad_norm": 0.29331767559051514, | |
| "learning_rate": 4.4765342960288814e-06, | |
| "loss": 1.6597, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.5595667870036101, | |
| "grad_norm": 0.29668062925338745, | |
| "learning_rate": 4.44043321299639e-06, | |
| "loss": 1.6513, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.5631768953068592, | |
| "grad_norm": 0.2959443926811218, | |
| "learning_rate": 4.404332129963899e-06, | |
| "loss": 1.6144, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.5667870036101083, | |
| "grad_norm": 0.28033116459846497, | |
| "learning_rate": 4.3682310469314086e-06, | |
| "loss": 1.6395, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.5703971119133574, | |
| "grad_norm": 0.334304541349411, | |
| "learning_rate": 4.332129963898917e-06, | |
| "loss": 1.6872, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.5740072202166066, | |
| "grad_norm": 0.31229716539382935, | |
| "learning_rate": 4.296028880866426e-06, | |
| "loss": 1.7021, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.5776173285198556, | |
| "grad_norm": 0.29688820242881775, | |
| "learning_rate": 4.259927797833936e-06, | |
| "loss": 1.5845, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.5812274368231047, | |
| "grad_norm": 0.3158322274684906, | |
| "learning_rate": 4.223826714801444e-06, | |
| "loss": 1.7025, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.5848375451263538, | |
| "grad_norm": 0.30644160509109497, | |
| "learning_rate": 4.1877256317689535e-06, | |
| "loss": 1.6314, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.5884476534296029, | |
| "grad_norm": 0.286468505859375, | |
| "learning_rate": 4.151624548736463e-06, | |
| "loss": 1.6037, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.592057761732852, | |
| "grad_norm": 0.30735689401626587, | |
| "learning_rate": 4.115523465703971e-06, | |
| "loss": 1.6586, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.5956678700361011, | |
| "grad_norm": 0.3296254575252533, | |
| "learning_rate": 4.079422382671481e-06, | |
| "loss": 1.6132, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.5992779783393501, | |
| "grad_norm": 0.3069722354412079, | |
| "learning_rate": 4.043321299638989e-06, | |
| "loss": 1.5489, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.6028880866425993, | |
| "grad_norm": 0.28502118587493896, | |
| "learning_rate": 4.0072202166064985e-06, | |
| "loss": 1.6295, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.6064981949458483, | |
| "grad_norm": 0.3035149574279785, | |
| "learning_rate": 3.971119133574007e-06, | |
| "loss": 1.6124, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.6101083032490975, | |
| "grad_norm": 0.2957877218723297, | |
| "learning_rate": 3.935018050541516e-06, | |
| "loss": 1.5872, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.6137184115523465, | |
| "grad_norm": 0.3029181957244873, | |
| "learning_rate": 3.898916967509026e-06, | |
| "loss": 1.6383, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.6173285198555957, | |
| "grad_norm": 0.2973797619342804, | |
| "learning_rate": 3.862815884476535e-06, | |
| "loss": 1.5865, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.6209386281588448, | |
| "grad_norm": 0.3023606240749359, | |
| "learning_rate": 3.826714801444043e-06, | |
| "loss": 1.5918, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.6245487364620939, | |
| "grad_norm": 0.284014493227005, | |
| "learning_rate": 3.7906137184115527e-06, | |
| "loss": 1.617, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.628158844765343, | |
| "grad_norm": 0.2972394824028015, | |
| "learning_rate": 3.7545126353790616e-06, | |
| "loss": 1.5863, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.631768953068592, | |
| "grad_norm": 0.3185518980026245, | |
| "learning_rate": 3.718411552346571e-06, | |
| "loss": 1.6378, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.6353790613718412, | |
| "grad_norm": 0.28479403257369995, | |
| "learning_rate": 3.68231046931408e-06, | |
| "loss": 1.6447, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.6389891696750902, | |
| "grad_norm": 0.280767560005188, | |
| "learning_rate": 3.6462093862815888e-06, | |
| "loss": 1.5842, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.6425992779783394, | |
| "grad_norm": 0.3037751019001007, | |
| "learning_rate": 3.610108303249098e-06, | |
| "loss": 1.6133, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.6462093862815884, | |
| "grad_norm": 0.2837255895137787, | |
| "learning_rate": 3.574007220216607e-06, | |
| "loss": 1.612, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.6498194945848376, | |
| "grad_norm": 0.32686418294906616, | |
| "learning_rate": 3.5379061371841155e-06, | |
| "loss": 1.6702, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.6534296028880866, | |
| "grad_norm": 0.28730490803718567, | |
| "learning_rate": 3.501805054151625e-06, | |
| "loss": 1.5823, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.6570397111913358, | |
| "grad_norm": 0.3138350546360016, | |
| "learning_rate": 3.4657039711191337e-06, | |
| "loss": 1.6215, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.6606498194945848, | |
| "grad_norm": 0.30949413776397705, | |
| "learning_rate": 3.4296028880866426e-06, | |
| "loss": 1.6279, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.6642599277978339, | |
| "grad_norm": 0.29190826416015625, | |
| "learning_rate": 3.393501805054152e-06, | |
| "loss": 1.6123, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.6678700361010831, | |
| "grad_norm": 0.295973539352417, | |
| "learning_rate": 3.357400722021661e-06, | |
| "loss": 1.6119, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.6714801444043321, | |
| "grad_norm": 0.3496643006801605, | |
| "learning_rate": 3.3212996389891698e-06, | |
| "loss": 1.6278, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.6750902527075813, | |
| "grad_norm": 0.28338801860809326, | |
| "learning_rate": 3.285198555956679e-06, | |
| "loss": 1.6153, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.6787003610108303, | |
| "grad_norm": 0.31767937541007996, | |
| "learning_rate": 3.249097472924188e-06, | |
| "loss": 1.6194, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.6823104693140795, | |
| "grad_norm": 0.31142091751098633, | |
| "learning_rate": 3.2129963898916973e-06, | |
| "loss": 1.556, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.6859205776173285, | |
| "grad_norm": 0.2969271242618561, | |
| "learning_rate": 3.1768953068592062e-06, | |
| "loss": 1.5854, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.6895306859205776, | |
| "grad_norm": 0.2818593382835388, | |
| "learning_rate": 3.140794223826715e-06, | |
| "loss": 1.5868, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.6931407942238267, | |
| "grad_norm": 0.27337560057640076, | |
| "learning_rate": 3.1046931407942245e-06, | |
| "loss": 1.6015, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.6967509025270758, | |
| "grad_norm": 0.2905227541923523, | |
| "learning_rate": 3.068592057761733e-06, | |
| "loss": 1.5871, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.7003610108303249, | |
| "grad_norm": 0.2861591875553131, | |
| "learning_rate": 3.032490974729242e-06, | |
| "loss": 1.591, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.703971119133574, | |
| "grad_norm": 0.28847527503967285, | |
| "learning_rate": 2.996389891696751e-06, | |
| "loss": 1.6101, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.7075812274368231, | |
| "grad_norm": 0.2815534770488739, | |
| "learning_rate": 2.96028880866426e-06, | |
| "loss": 1.5857, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.7111913357400722, | |
| "grad_norm": 0.30834150314331055, | |
| "learning_rate": 2.924187725631769e-06, | |
| "loss": 1.6362, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.7148014440433214, | |
| "grad_norm": 0.28502917289733887, | |
| "learning_rate": 2.8880866425992783e-06, | |
| "loss": 1.6403, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.7184115523465704, | |
| "grad_norm": 0.29509761929512024, | |
| "learning_rate": 2.8519855595667872e-06, | |
| "loss": 1.6124, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.7220216606498195, | |
| "grad_norm": 0.321684330701828, | |
| "learning_rate": 2.815884476534296e-06, | |
| "loss": 1.6086, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.7256317689530686, | |
| "grad_norm": 0.29435068368911743, | |
| "learning_rate": 2.7797833935018055e-06, | |
| "loss": 1.5909, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.7292418772563177, | |
| "grad_norm": 0.2958379089832306, | |
| "learning_rate": 2.7436823104693144e-06, | |
| "loss": 1.5553, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.7328519855595668, | |
| "grad_norm": 0.2923933267593384, | |
| "learning_rate": 2.7075812274368237e-06, | |
| "loss": 1.586, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.7364620938628159, | |
| "grad_norm": 0.28724294900894165, | |
| "learning_rate": 2.6714801444043326e-06, | |
| "loss": 1.6265, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.740072202166065, | |
| "grad_norm": 0.27974867820739746, | |
| "learning_rate": 2.6353790613718415e-06, | |
| "loss": 1.6058, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.7436823104693141, | |
| "grad_norm": 0.2806769609451294, | |
| "learning_rate": 2.59927797833935e-06, | |
| "loss": 1.5748, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.7472924187725631, | |
| "grad_norm": 0.30230197310447693, | |
| "learning_rate": 2.5631768953068593e-06, | |
| "loss": 1.6434, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.7509025270758123, | |
| "grad_norm": 0.29202568531036377, | |
| "learning_rate": 2.527075812274368e-06, | |
| "loss": 1.5673, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.7545126353790613, | |
| "grad_norm": 0.299363911151886, | |
| "learning_rate": 2.4909747292418775e-06, | |
| "loss": 1.5546, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.7581227436823105, | |
| "grad_norm": 0.3040112853050232, | |
| "learning_rate": 2.4548736462093864e-06, | |
| "loss": 1.6415, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.7617328519855595, | |
| "grad_norm": 0.3127659261226654, | |
| "learning_rate": 2.4187725631768953e-06, | |
| "loss": 1.5328, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.7653429602888087, | |
| "grad_norm": 0.2943485379219055, | |
| "learning_rate": 2.3826714801444047e-06, | |
| "loss": 1.5685, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.7689530685920578, | |
| "grad_norm": 0.2981877326965332, | |
| "learning_rate": 2.3465703971119136e-06, | |
| "loss": 1.6391, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.7725631768953068, | |
| "grad_norm": 0.2939170002937317, | |
| "learning_rate": 2.3104693140794225e-06, | |
| "loss": 1.6518, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.776173285198556, | |
| "grad_norm": 0.29882779717445374, | |
| "learning_rate": 2.274368231046932e-06, | |
| "loss": 1.6313, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.779783393501805, | |
| "grad_norm": 0.30276381969451904, | |
| "learning_rate": 2.2382671480144407e-06, | |
| "loss": 1.6487, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.7833935018050542, | |
| "grad_norm": 0.2941838502883911, | |
| "learning_rate": 2.2021660649819496e-06, | |
| "loss": 1.629, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.7870036101083032, | |
| "grad_norm": 0.2928179204463959, | |
| "learning_rate": 2.1660649819494585e-06, | |
| "loss": 1.6191, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.7906137184115524, | |
| "grad_norm": 0.3069646656513214, | |
| "learning_rate": 2.129963898916968e-06, | |
| "loss": 1.6167, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.7942238267148014, | |
| "grad_norm": 0.2801755666732788, | |
| "learning_rate": 2.0938628158844768e-06, | |
| "loss": 1.6065, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.7978339350180506, | |
| "grad_norm": 0.2858894467353821, | |
| "learning_rate": 2.0577617328519857e-06, | |
| "loss": 1.5903, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.8014440433212996, | |
| "grad_norm": 0.3028116524219513, | |
| "learning_rate": 2.0216606498194946e-06, | |
| "loss": 1.5893, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.8050541516245487, | |
| "grad_norm": 0.2898472845554352, | |
| "learning_rate": 1.9855595667870035e-06, | |
| "loss": 1.6318, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.8086642599277978, | |
| "grad_norm": 0.2822205424308777, | |
| "learning_rate": 1.949458483754513e-06, | |
| "loss": 1.6005, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.8122743682310469, | |
| "grad_norm": 0.29625725746154785, | |
| "learning_rate": 1.9133574007220217e-06, | |
| "loss": 1.5658, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.8158844765342961, | |
| "grad_norm": 0.2854407727718353, | |
| "learning_rate": 1.8772563176895308e-06, | |
| "loss": 1.5803, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.8194945848375451, | |
| "grad_norm": 0.3293429911136627, | |
| "learning_rate": 1.84115523465704e-06, | |
| "loss": 1.5214, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.8231046931407943, | |
| "grad_norm": 0.31535983085632324, | |
| "learning_rate": 1.805054151624549e-06, | |
| "loss": 1.5533, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.8267148014440433, | |
| "grad_norm": 0.2959997057914734, | |
| "learning_rate": 1.7689530685920577e-06, | |
| "loss": 1.6071, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.8303249097472925, | |
| "grad_norm": 0.3083932399749756, | |
| "learning_rate": 1.7328519855595669e-06, | |
| "loss": 1.5714, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.8339350180505415, | |
| "grad_norm": 0.2812926471233368, | |
| "learning_rate": 1.696750902527076e-06, | |
| "loss": 1.5719, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.8375451263537906, | |
| "grad_norm": 0.30061206221580505, | |
| "learning_rate": 1.6606498194945849e-06, | |
| "loss": 1.6485, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.8411552346570397, | |
| "grad_norm": 0.29850760102272034, | |
| "learning_rate": 1.624548736462094e-06, | |
| "loss": 1.6151, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.8447653429602888, | |
| "grad_norm": 0.2823867201805115, | |
| "learning_rate": 1.5884476534296031e-06, | |
| "loss": 1.5873, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.8483754512635379, | |
| "grad_norm": 0.35106369853019714, | |
| "learning_rate": 1.5523465703971122e-06, | |
| "loss": 1.6039, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.851985559566787, | |
| "grad_norm": 0.30508509278297424, | |
| "learning_rate": 1.516245487364621e-06, | |
| "loss": 1.6638, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.855595667870036, | |
| "grad_norm": 0.2848837673664093, | |
| "learning_rate": 1.48014440433213e-06, | |
| "loss": 1.5795, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.8592057761732852, | |
| "grad_norm": 0.29718756675720215, | |
| "learning_rate": 1.4440433212996392e-06, | |
| "loss": 1.5963, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.8628158844765343, | |
| "grad_norm": 0.2890275716781616, | |
| "learning_rate": 1.407942238267148e-06, | |
| "loss": 1.5765, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.8664259927797834, | |
| "grad_norm": 0.31825488805770874, | |
| "learning_rate": 1.3718411552346572e-06, | |
| "loss": 1.616, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.8700361010830325, | |
| "grad_norm": 0.29879897832870483, | |
| "learning_rate": 1.3357400722021663e-06, | |
| "loss": 1.6033, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.8736462093862816, | |
| "grad_norm": 0.3011474013328552, | |
| "learning_rate": 1.299638989169675e-06, | |
| "loss": 1.5859, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.8772563176895307, | |
| "grad_norm": 0.29284024238586426, | |
| "learning_rate": 1.263537906137184e-06, | |
| "loss": 1.5689, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.8808664259927798, | |
| "grad_norm": 0.31116312742233276, | |
| "learning_rate": 1.2274368231046932e-06, | |
| "loss": 1.6243, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.8844765342960289, | |
| "grad_norm": 0.2952096164226532, | |
| "learning_rate": 1.1913357400722023e-06, | |
| "loss": 1.5763, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.8880866425992779, | |
| "grad_norm": 0.2993987798690796, | |
| "learning_rate": 1.1552346570397112e-06, | |
| "loss": 1.573, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.8916967509025271, | |
| "grad_norm": 0.3527011573314667, | |
| "learning_rate": 1.1191335740072204e-06, | |
| "loss": 1.6556, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.8953068592057761, | |
| "grad_norm": 0.317192405462265, | |
| "learning_rate": 1.0830324909747293e-06, | |
| "loss": 1.5407, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.8989169675090253, | |
| "grad_norm": 0.28880420327186584, | |
| "learning_rate": 1.0469314079422384e-06, | |
| "loss": 1.5866, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.9025270758122743, | |
| "grad_norm": 0.2954920828342438, | |
| "learning_rate": 1.0108303249097473e-06, | |
| "loss": 1.5643, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.9061371841155235, | |
| "grad_norm": 0.2832193970680237, | |
| "learning_rate": 9.747292418772564e-07, | |
| "loss": 1.6216, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.9097472924187726, | |
| "grad_norm": 0.2810640335083008, | |
| "learning_rate": 9.386281588447654e-07, | |
| "loss": 1.5439, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.9133574007220217, | |
| "grad_norm": 0.3256809413433075, | |
| "learning_rate": 9.025270758122745e-07, | |
| "loss": 1.637, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.9169675090252708, | |
| "grad_norm": 0.3034754991531372, | |
| "learning_rate": 8.664259927797834e-07, | |
| "loss": 1.5583, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.9205776173285198, | |
| "grad_norm": 0.30552029609680176, | |
| "learning_rate": 8.303249097472924e-07, | |
| "loss": 1.6058, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.924187725631769, | |
| "grad_norm": 0.30193251371383667, | |
| "learning_rate": 7.942238267148016e-07, | |
| "loss": 1.5893, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.927797833935018, | |
| "grad_norm": 0.29293292760849, | |
| "learning_rate": 7.581227436823105e-07, | |
| "loss": 1.5992, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.9314079422382672, | |
| "grad_norm": 0.3184659779071808, | |
| "learning_rate": 7.220216606498196e-07, | |
| "loss": 1.5516, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.9350180505415162, | |
| "grad_norm": 0.28600752353668213, | |
| "learning_rate": 6.859205776173286e-07, | |
| "loss": 1.5821, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.9386281588447654, | |
| "grad_norm": 0.3145878314971924, | |
| "learning_rate": 6.498194945848375e-07, | |
| "loss": 1.5855, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.9422382671480144, | |
| "grad_norm": 0.2760341167449951, | |
| "learning_rate": 6.137184115523466e-07, | |
| "loss": 1.5639, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.9458483754512635, | |
| "grad_norm": 0.316036194562912, | |
| "learning_rate": 5.776173285198556e-07, | |
| "loss": 1.559, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.9494584837545126, | |
| "grad_norm": 0.304881751537323, | |
| "learning_rate": 5.415162454873646e-07, | |
| "loss": 1.5951, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.9530685920577617, | |
| "grad_norm": 0.31182485818862915, | |
| "learning_rate": 5.054151624548736e-07, | |
| "loss": 1.5896, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.9566787003610109, | |
| "grad_norm": 0.29128628969192505, | |
| "learning_rate": 4.693140794223827e-07, | |
| "loss": 1.6025, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.9602888086642599, | |
| "grad_norm": 0.33992016315460205, | |
| "learning_rate": 4.332129963898917e-07, | |
| "loss": 1.6466, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.9638989169675091, | |
| "grad_norm": 0.395353764295578, | |
| "learning_rate": 3.971119133574008e-07, | |
| "loss": 1.6668, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.9675090252707581, | |
| "grad_norm": 0.3091205954551697, | |
| "learning_rate": 3.610108303249098e-07, | |
| "loss": 1.6403, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.9711191335740073, | |
| "grad_norm": 0.29591071605682373, | |
| "learning_rate": 3.2490974729241875e-07, | |
| "loss": 1.5891, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.9747292418772563, | |
| "grad_norm": 0.3448485732078552, | |
| "learning_rate": 2.888086642599278e-07, | |
| "loss": 1.6936, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.9783393501805054, | |
| "grad_norm": 0.2853164076805115, | |
| "learning_rate": 2.527075812274368e-07, | |
| "loss": 1.5892, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.9819494584837545, | |
| "grad_norm": 0.30147141218185425, | |
| "learning_rate": 2.1660649819494586e-07, | |
| "loss": 1.6227, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.9855595667870036, | |
| "grad_norm": 0.2863829731941223, | |
| "learning_rate": 1.805054151624549e-07, | |
| "loss": 1.5997, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.9891696750902527, | |
| "grad_norm": 0.28152960538864136, | |
| "learning_rate": 1.444043321299639e-07, | |
| "loss": 1.6122, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.9927797833935018, | |
| "grad_norm": 0.3065321743488312, | |
| "learning_rate": 1.0830324909747293e-07, | |
| "loss": 1.5703, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.9963898916967509, | |
| "grad_norm": 0.3063216805458069, | |
| "learning_rate": 7.220216606498195e-08, | |
| "loss": 1.5751, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.29616397619247437, | |
| "learning_rate": 3.6101083032490976e-08, | |
| "loss": 1.5946, | |
| "step": 277 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 277, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 0, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.7564904470872064e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |