| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 5000, | |
| "global_step": 34113, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.997068566235746e-05, | |
| "loss": 0.5849, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.994137132471492e-05, | |
| "loss": 0.4633, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.9912056987072378e-05, | |
| "loss": 0.4247, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9882742649429837e-05, | |
| "loss": 0.4619, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9853428311787297e-05, | |
| "loss": 0.4022, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9824113974144757e-05, | |
| "loss": 0.4302, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 1.9794799636502217e-05, | |
| "loss": 0.3821, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9765485298859673e-05, | |
| "loss": 0.3905, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9736170961217133e-05, | |
| "loss": 0.3973, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9706856623574592e-05, | |
| "loss": 0.3715, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9677542285932052e-05, | |
| "loss": 0.4066, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 1.9648227948289512e-05, | |
| "loss": 0.3592, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.961891361064697e-05, | |
| "loss": 0.3804, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9589599273004428e-05, | |
| "loss": 0.3714, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9560284935361888e-05, | |
| "loss": 0.3693, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9530970597719344e-05, | |
| "loss": 0.3376, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 1.9501656260076804e-05, | |
| "loss": 0.3699, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9472341922434264e-05, | |
| "loss": 0.3919, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 1.9443027584791724e-05, | |
| "loss": 0.3535, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9413713247149183e-05, | |
| "loss": 0.3405, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.938439890950664e-05, | |
| "loss": 0.3694, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.93550845718641e-05, | |
| "loss": 0.3271, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 1.932577023422156e-05, | |
| "loss": 0.3232, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.929645589657902e-05, | |
| "loss": 0.3314, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.926714155893648e-05, | |
| "loss": 0.3622, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.9237827221293935e-05, | |
| "loss": 0.3423, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9208512883651395e-05, | |
| "loss": 0.3359, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.9179198546008855e-05, | |
| "loss": 0.3512, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9149884208366314e-05, | |
| "loss": 0.3704, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9120569870723774e-05, | |
| "loss": 0.3486, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.909125553308123e-05, | |
| "loss": 0.3463, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.906194119543869e-05, | |
| "loss": 0.3341, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.903262685779615e-05, | |
| "loss": 0.3212, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.900331252015361e-05, | |
| "loss": 0.3374, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.897399818251107e-05, | |
| "loss": 0.3435, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.894468384486853e-05, | |
| "loss": 0.3273, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.8915369507225986e-05, | |
| "loss": 0.3198, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8886055169583445e-05, | |
| "loss": 0.3233, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8856740831940902e-05, | |
| "loss": 0.2918, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.882742649429836e-05, | |
| "loss": 0.3092, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.879811215665582e-05, | |
| "loss": 0.3321, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.876879781901328e-05, | |
| "loss": 0.3197, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.873948348137074e-05, | |
| "loss": 0.3332, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8710169143728197e-05, | |
| "loss": 0.3376, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8680854806085657e-05, | |
| "loss": 0.3119, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.8651540468443117e-05, | |
| "loss": 0.3221, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8622226130800576e-05, | |
| "loss": 0.3322, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8592911793158036e-05, | |
| "loss": 0.3107, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8563597455515493e-05, | |
| "loss": 0.3318, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8534283117872952e-05, | |
| "loss": 0.3303, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 1.8504968780230412e-05, | |
| "loss": 0.3135, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.8475654442587872e-05, | |
| "loss": 0.3145, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 1.844634010494533e-05, | |
| "loss": 0.3272, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8417025767302788e-05, | |
| "loss": 0.3114, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8387711429660248e-05, | |
| "loss": 0.3199, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8358397092017708e-05, | |
| "loss": 0.3188, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 1.8329082754375167e-05, | |
| "loss": 0.3245, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8299768416732627e-05, | |
| "loss": 0.331, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8270454079090087e-05, | |
| "loss": 0.3241, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8241139741447543e-05, | |
| "loss": 0.331, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.8211825403805003e-05, | |
| "loss": 0.3274, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 1.8182511066162463e-05, | |
| "loss": 0.3333, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.815319672851992e-05, | |
| "loss": 0.3238, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.812388239087738e-05, | |
| "loss": 0.3256, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.809456805323484e-05, | |
| "loss": 0.3202, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.80652537155923e-05, | |
| "loss": 0.3131, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 1.8035939377949755e-05, | |
| "loss": 0.3199, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.8006625040307214e-05, | |
| "loss": 0.3372, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.7977310702664674e-05, | |
| "loss": 0.3137, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7947996365022134e-05, | |
| "loss": 0.2928, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 1.7918682027379594e-05, | |
| "loss": 0.3158, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.788936768973705e-05, | |
| "loss": 0.3316, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.786005335209451e-05, | |
| "loss": 0.3158, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.783073901445197e-05, | |
| "loss": 0.3086, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.780142467680943e-05, | |
| "loss": 0.3193, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 1.777211033916689e-05, | |
| "loss": 0.3171, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.7742796001524346e-05, | |
| "loss": 0.3035, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.7713481663881805e-05, | |
| "loss": 0.318, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.7684167326239265e-05, | |
| "loss": 0.3084, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 1.7654852988596725e-05, | |
| "loss": 0.3064, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7625538650954185e-05, | |
| "loss": 0.3169, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.7596224313311644e-05, | |
| "loss": 0.2797, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.75669099756691e-05, | |
| "loss": 0.3189, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.753759563802656e-05, | |
| "loss": 0.3024, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.750828130038402e-05, | |
| "loss": 0.3019, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.747896696274148e-05, | |
| "loss": 0.294, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.744965262509894e-05, | |
| "loss": 0.3171, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.7420338287456396e-05, | |
| "loss": 0.3097, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.7391023949813856e-05, | |
| "loss": 0.3315, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.7361709612171312e-05, | |
| "loss": 0.2811, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.7332395274528772e-05, | |
| "loss": 0.316, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.7303080936886232e-05, | |
| "loss": 0.3216, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.727376659924369e-05, | |
| "loss": 0.3247, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.724445226160115e-05, | |
| "loss": 0.3119, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.7215137923958608e-05, | |
| "loss": 0.324, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 1.7185823586316067e-05, | |
| "loss": 0.2838, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.7156509248673527e-05, | |
| "loss": 0.285, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.7127194911030987e-05, | |
| "loss": 0.3189, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.7097880573388447e-05, | |
| "loss": 0.288, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.7068566235745903e-05, | |
| "loss": 0.2833, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "eval_accuracy": 0.8708384862725699, | |
| "eval_combined_score": 0.8462457458685254, | |
| "eval_f1": 0.821653005464481, | |
| "eval_loss": 0.30874383449554443, | |
| "eval_runtime": 108.3745, | |
| "eval_samples_per_second": 373.058, | |
| "eval_steps_per_second": 23.317, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.7039251898103363e-05, | |
| "loss": 0.3055, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.7009937560460823e-05, | |
| "loss": 0.3364, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.6980623222818282e-05, | |
| "loss": 0.3007, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6951308885175742e-05, | |
| "loss": 0.3178, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 1.6921994547533202e-05, | |
| "loss": 0.311, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6892680209890658e-05, | |
| "loss": 0.2985, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6863365872248118e-05, | |
| "loss": 0.3167, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.6834051534605578e-05, | |
| "loss": 0.3162, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.6804737196963038e-05, | |
| "loss": 0.2741, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.6775422859320497e-05, | |
| "loss": 0.273, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6746108521677954e-05, | |
| "loss": 0.2992, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6716794184035413e-05, | |
| "loss": 0.2909, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.668747984639287e-05, | |
| "loss": 0.303, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 1.665816550875033e-05, | |
| "loss": 0.3034, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.662885117110779e-05, | |
| "loss": 0.2962, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.659953683346525e-05, | |
| "loss": 0.3019, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.657022249582271e-05, | |
| "loss": 0.2787, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.6540908158180165e-05, | |
| "loss": 0.2992, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.6511593820537625e-05, | |
| "loss": 0.276, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.6482279482895085e-05, | |
| "loss": 0.2784, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 1.6452965145252545e-05, | |
| "loss": 0.295, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.6423650807610004e-05, | |
| "loss": 0.311, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.639433646996746e-05, | |
| "loss": 0.2837, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.636502213232492e-05, | |
| "loss": 0.3122, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.633570779468238e-05, | |
| "loss": 0.2894, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 1.630639345703984e-05, | |
| "loss": 0.272, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.62770791193973e-05, | |
| "loss": 0.3021, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.624776478175476e-05, | |
| "loss": 0.2769, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6218450444112216e-05, | |
| "loss": 0.3097, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6189136106469676e-05, | |
| "loss": 0.3059, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6159821768827135e-05, | |
| "loss": 0.3084, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6130507431184595e-05, | |
| "loss": 0.3111, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6101193093542055e-05, | |
| "loss": 0.3097, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.607187875589951e-05, | |
| "loss": 0.2773, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 1.604256441825697e-05, | |
| "loss": 0.3035, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.601325008061443e-05, | |
| "loss": 0.2965, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.5983935742971887e-05, | |
| "loss": 0.2795, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.5954621405329347e-05, | |
| "loss": 0.2939, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 1.5925307067686807e-05, | |
| "loss": 0.2988, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.5895992730044266e-05, | |
| "loss": 0.3245, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.5866678392401723e-05, | |
| "loss": 0.3065, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.5837364054759183e-05, | |
| "loss": 0.309, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.5808049717116642e-05, | |
| "loss": 0.292, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 1.5778735379474102e-05, | |
| "loss": 0.3071, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5749421041831562e-05, | |
| "loss": 0.2645, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5720106704189018e-05, | |
| "loss": 0.2933, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5690792366546478e-05, | |
| "loss": 0.3006, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 1.5661478028903938e-05, | |
| "loss": 0.2965, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5632163691261397e-05, | |
| "loss": 0.3085, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5602849353618857e-05, | |
| "loss": 0.2814, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 1.5573535015976317e-05, | |
| "loss": 0.2992, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5544220678333773e-05, | |
| "loss": 0.2831, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5514906340691233e-05, | |
| "loss": 0.2761, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5485592003048693e-05, | |
| "loss": 0.2898, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.5456277665406153e-05, | |
| "loss": 0.2731, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5426963327763612e-05, | |
| "loss": 0.2808, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.539764899012107e-05, | |
| "loss": 0.2646, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.536833465247853e-05, | |
| "loss": 0.2887, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5339020314835988e-05, | |
| "loss": 0.2708, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5309705977193448e-05, | |
| "loss": 0.2852, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5280391639550908e-05, | |
| "loss": 0.2852, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5251077301908364e-05, | |
| "loss": 0.2975, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.5221762964265822e-05, | |
| "loss": 0.2897, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.5192448626623282e-05, | |
| "loss": 0.2799, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.516313428898074e-05, | |
| "loss": 0.2733, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.51338199513382e-05, | |
| "loss": 0.299, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.510450561369566e-05, | |
| "loss": 0.2682, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.5075191276053118e-05, | |
| "loss": 0.3041, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.5045876938410577e-05, | |
| "loss": 0.2798, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5016562600768037e-05, | |
| "loss": 0.2799, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4987248263125495e-05, | |
| "loss": 0.303, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4957933925482955e-05, | |
| "loss": 0.2841, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4928619587840413e-05, | |
| "loss": 0.2907, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4899305250197873e-05, | |
| "loss": 0.26, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4869990912555333e-05, | |
| "loss": 0.262, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.484067657491279e-05, | |
| "loss": 0.287, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.481136223727025e-05, | |
| "loss": 0.2798, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.478204789962771e-05, | |
| "loss": 0.2944, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4752733561985168e-05, | |
| "loss": 0.2785, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4723419224342628e-05, | |
| "loss": 0.286, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4694104886700086e-05, | |
| "loss": 0.268, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4664790549057546e-05, | |
| "loss": 0.2861, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.4635476211415006e-05, | |
| "loss": 0.2612, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.4606161873772464e-05, | |
| "loss": 0.2826, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 1.4576847536129923e-05, | |
| "loss": 0.2857, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.4547533198487383e-05, | |
| "loss": 0.2699, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.451821886084484e-05, | |
| "loss": 0.2788, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4488904523202298e-05, | |
| "loss": 0.3011, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4459590185559757e-05, | |
| "loss": 0.2966, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4430275847917217e-05, | |
| "loss": 0.3035, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4400961510274675e-05, | |
| "loss": 0.2646, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4371647172632135e-05, | |
| "loss": 0.2876, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.4342332834989595e-05, | |
| "loss": 0.2765, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 1.4313018497347053e-05, | |
| "loss": 0.3, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4283704159704513e-05, | |
| "loss": 0.276, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.425438982206197e-05, | |
| "loss": 0.2847, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.422507548441943e-05, | |
| "loss": 0.2828, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.419576114677689e-05, | |
| "loss": 0.2949, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4166446809134348e-05, | |
| "loss": 0.2737, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4137132471491808e-05, | |
| "loss": 0.2702, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "eval_accuracy": 0.8817956962651496, | |
| "eval_combined_score": 0.8619452626377788, | |
| "eval_f1": 0.8420948290104081, | |
| "eval_loss": 0.27627187967300415, | |
| "eval_runtime": 108.3505, | |
| "eval_samples_per_second": 373.141, | |
| "eval_steps_per_second": 23.322, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4107818133849268e-05, | |
| "loss": 0.2976, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.4078503796206726e-05, | |
| "loss": 0.2986, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.4049189458564186e-05, | |
| "loss": 0.2732, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.4019875120921644e-05, | |
| "loss": 0.2875, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.3990560783279103e-05, | |
| "loss": 0.2689, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3961246445636563e-05, | |
| "loss": 0.2715, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3931932107994021e-05, | |
| "loss": 0.2625, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.3902617770351481e-05, | |
| "loss": 0.271, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.387330343270894e-05, | |
| "loss": 0.2777, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.3843989095066399e-05, | |
| "loss": 0.2787, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3814674757423859e-05, | |
| "loss": 0.2588, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3785360419781315e-05, | |
| "loss": 0.289, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.3756046082138775e-05, | |
| "loss": 0.2858, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.3726731744496233e-05, | |
| "loss": 0.2704, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.3697417406853692e-05, | |
| "loss": 0.2946, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.3668103069211152e-05, | |
| "loss": 0.273, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 1.363878873156861e-05, | |
| "loss": 0.2849, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.360947439392607e-05, | |
| "loss": 0.2671, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.3580160056283528e-05, | |
| "loss": 0.295, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3550845718640988e-05, | |
| "loss": 0.2573, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3521531380998448e-05, | |
| "loss": 0.2728, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.3492217043355906e-05, | |
| "loss": 0.285, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.3462902705713365e-05, | |
| "loss": 0.2732, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.3433588368070825e-05, | |
| "loss": 0.2819, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3404274030428283e-05, | |
| "loss": 0.2647, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3374959692785743e-05, | |
| "loss": 0.2676, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.3345645355143201e-05, | |
| "loss": 0.2777, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.3316331017500661e-05, | |
| "loss": 0.2622, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.328701667985812e-05, | |
| "loss": 0.2065, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.3257702342215579e-05, | |
| "loss": 0.2628, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.3228388004573038e-05, | |
| "loss": 0.2213, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.3199073666930498e-05, | |
| "loss": 0.2432, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.3169759329287956e-05, | |
| "loss": 0.2196, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.3140444991645416e-05, | |
| "loss": 0.229, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.3111130654002874e-05, | |
| "loss": 0.2, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.3081816316360334e-05, | |
| "loss": 0.2317, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.305250197871779e-05, | |
| "loss": 0.2475, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.302318764107525e-05, | |
| "loss": 0.2705, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.299387330343271e-05, | |
| "loss": 0.2357, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.2964558965790168e-05, | |
| "loss": 0.2385, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.2935244628147628e-05, | |
| "loss": 0.2449, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.2905930290505086e-05, | |
| "loss": 0.2279, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.2876615952862545e-05, | |
| "loss": 0.2364, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.2847301615220005e-05, | |
| "loss": 0.2434, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.2817987277577463e-05, | |
| "loss": 0.2484, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.2788672939934923e-05, | |
| "loss": 0.2465, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.2759358602292383e-05, | |
| "loss": 0.2381, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.2730044264649841e-05, | |
| "loss": 0.2432, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.27007299270073e-05, | |
| "loss": 0.2117, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.2671415589364759e-05, | |
| "loss": 0.2055, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.2642101251722218e-05, | |
| "loss": 0.2264, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.2612786914079678e-05, | |
| "loss": 0.2369, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 1.2583472576437136e-05, | |
| "loss": 0.2489, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.2554158238794596e-05, | |
| "loss": 0.2084, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.2524843901152056e-05, | |
| "loss": 0.2229, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.2495529563509514e-05, | |
| "loss": 0.2172, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.2466215225866974e-05, | |
| "loss": 0.2234, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 1.2436900888224432e-05, | |
| "loss": 0.2412, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.2407586550581891e-05, | |
| "loss": 0.2285, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.2378272212939351e-05, | |
| "loss": 0.2316, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.234895787529681e-05, | |
| "loss": 0.2379, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2319643537654267e-05, | |
| "loss": 0.2157, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.2290329200011725e-05, | |
| "loss": 0.2359, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.2261014862369185e-05, | |
| "loss": 0.2379, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.2231700524726643e-05, | |
| "loss": 0.2409, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.2202386187084103e-05, | |
| "loss": 0.2239, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 1.2173071849441563e-05, | |
| "loss": 0.2497, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.214375751179902e-05, | |
| "loss": 0.2226, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.211444317415648e-05, | |
| "loss": 0.2456, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.208512883651394e-05, | |
| "loss": 0.2147, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 1.2055814498871398e-05, | |
| "loss": 0.2301, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2026500161228858e-05, | |
| "loss": 0.2486, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.1997185823586316e-05, | |
| "loss": 0.2393, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.1967871485943776e-05, | |
| "loss": 0.2393, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.1938557148301236e-05, | |
| "loss": 0.2242, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.1909242810658694e-05, | |
| "loss": 0.2367, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.1879928473016154e-05, | |
| "loss": 0.2361, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.1850614135373613e-05, | |
| "loss": 0.2173, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.1821299797731071e-05, | |
| "loss": 0.2525, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.1791985460088531e-05, | |
| "loss": 0.2403, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.176267112244599e-05, | |
| "loss": 0.2286, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.1733356784803449e-05, | |
| "loss": 0.253, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.1704042447160909e-05, | |
| "loss": 0.2206, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.1674728109518367e-05, | |
| "loss": 0.2528, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.1645413771875827e-05, | |
| "loss": 0.2042, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.1616099434233283e-05, | |
| "loss": 0.2508, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 1.1586785096590743e-05, | |
| "loss": 0.2314, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.15574707589482e-05, | |
| "loss": 0.2125, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.152815642130566e-05, | |
| "loss": 0.2477, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.149884208366312e-05, | |
| "loss": 0.23, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.1469527746020578e-05, | |
| "loss": 0.2362, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.1440213408378038e-05, | |
| "loss": 0.2587, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.1410899070735498e-05, | |
| "loss": 0.2469, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.1381584733092956e-05, | |
| "loss": 0.2192, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.1352270395450416e-05, | |
| "loss": 0.2523, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.1322956057807874e-05, | |
| "loss": 0.2455, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.1293641720165334e-05, | |
| "loss": 0.2318, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.1264327382522793e-05, | |
| "loss": 0.2357, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.1235013044880251e-05, | |
| "loss": 0.2115, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.1205698707237711e-05, | |
| "loss": 0.2269, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "eval_accuracy": 0.888276032649023, | |
| "eval_combined_score": 0.8675760989713809, | |
| "eval_f1": 0.8468761652937388, | |
| "eval_loss": 0.2819179594516754, | |
| "eval_runtime": 108.337, | |
| "eval_samples_per_second": 373.187, | |
| "eval_steps_per_second": 23.325, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.1176384369595171e-05, | |
| "loss": 0.229, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.1147070031952629e-05, | |
| "loss": 0.2491, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.1117755694310089e-05, | |
| "loss": 0.2175, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.1088441356667547e-05, | |
| "loss": 0.244, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.1059127019025007e-05, | |
| "loss": 0.244, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.1029812681382466e-05, | |
| "loss": 0.2319, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.1000498343739924e-05, | |
| "loss": 0.2064, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.0971184006097384e-05, | |
| "loss": 0.2348, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.0941869668454844e-05, | |
| "loss": 0.251, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.0912555330812302e-05, | |
| "loss": 0.2404, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.0883240993169758e-05, | |
| "loss": 0.2358, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.0853926655527218e-05, | |
| "loss": 0.2143, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.0824612317884678e-05, | |
| "loss": 0.2112, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.0795297980242136e-05, | |
| "loss": 0.2289, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.0765983642599596e-05, | |
| "loss": 0.2097, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.0736669304957055e-05, | |
| "loss": 0.2372, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 1.0707354967314513e-05, | |
| "loss": 0.2426, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.0678040629671973e-05, | |
| "loss": 0.2446, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.0648726292029431e-05, | |
| "loss": 0.2371, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.0619411954386891e-05, | |
| "loss": 0.2438, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.059009761674435e-05, | |
| "loss": 0.2324, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.0560783279101809e-05, | |
| "loss": 0.235, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.0531468941459269e-05, | |
| "loss": 0.2204, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.0502154603816728e-05, | |
| "loss": 0.2203, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.0472840266174186e-05, | |
| "loss": 0.222, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.0443525928531646e-05, | |
| "loss": 0.2566, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.0414211590889104e-05, | |
| "loss": 0.2315, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.0384897253246564e-05, | |
| "loss": 0.2336, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.0355582915604024e-05, | |
| "loss": 0.2375, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.0326268577961482e-05, | |
| "loss": 0.2333, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0296954240318942e-05, | |
| "loss": 0.2355, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0267639902676401e-05, | |
| "loss": 0.2319, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.023832556503386e-05, | |
| "loss": 0.2328, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.020901122739132e-05, | |
| "loss": 0.2279, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.0179696889748777e-05, | |
| "loss": 0.2179, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0150382552106235e-05, | |
| "loss": 0.2064, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0121068214463693e-05, | |
| "loss": 0.1875, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.0091753876821153e-05, | |
| "loss": 0.2398, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 1.0062439539178613e-05, | |
| "loss": 0.2062, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.0033125201536071e-05, | |
| "loss": 0.2487, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.000381086389353e-05, | |
| "loss": 0.2342, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.974496526250989e-06, | |
| "loss": 0.2363, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.945182188608449e-06, | |
| "loss": 0.2114, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 9.915867850965908e-06, | |
| "loss": 0.2273, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.886553513323366e-06, | |
| "loss": 0.2079, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.857239175680826e-06, | |
| "loss": 0.2242, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.827924838038286e-06, | |
| "loss": 0.222, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.798610500395744e-06, | |
| "loss": 0.2155, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.769296162753204e-06, | |
| "loss": 0.2188, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.739981825110662e-06, | |
| "loss": 0.2428, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 9.710667487468122e-06, | |
| "loss": 0.2202, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.681353149825581e-06, | |
| "loss": 0.2133, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.65203881218304e-06, | |
| "loss": 0.2113, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.6227244745405e-06, | |
| "loss": 0.2286, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 9.593410136897957e-06, | |
| "loss": 0.2465, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.564095799255417e-06, | |
| "loss": 0.2044, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.534781461612875e-06, | |
| "loss": 0.2176, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.505467123970335e-06, | |
| "loss": 0.2321, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.476152786327793e-06, | |
| "loss": 0.2215, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.446838448685253e-06, | |
| "loss": 0.2466, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.417524111042712e-06, | |
| "loss": 0.2194, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.38820977340017e-06, | |
| "loss": 0.2555, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.35889543575763e-06, | |
| "loss": 0.2297, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.329581098115088e-06, | |
| "loss": 0.2198, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.300266760472548e-06, | |
| "loss": 0.2292, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.270952422830008e-06, | |
| "loss": 0.2194, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.241638085187466e-06, | |
| "loss": 0.2103, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.212323747544924e-06, | |
| "loss": 0.2466, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.183009409902384e-06, | |
| "loss": 0.229, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.153695072259843e-06, | |
| "loss": 0.2056, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.124380734617302e-06, | |
| "loss": 0.2285, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.095066396974761e-06, | |
| "loss": 0.2169, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.06575205933222e-06, | |
| "loss": 0.2423, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.036437721689679e-06, | |
| "loss": 0.27, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 9.007123384047139e-06, | |
| "loss": 0.2309, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.977809046404597e-06, | |
| "loss": 0.2302, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.948494708762057e-06, | |
| "loss": 0.2252, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.919180371119516e-06, | |
| "loss": 0.2164, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.889866033476975e-06, | |
| "loss": 0.216, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.860551695834433e-06, | |
| "loss": 0.2224, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.831237358191892e-06, | |
| "loss": 0.2277, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.80192302054935e-06, | |
| "loss": 0.2515, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 8.77260868290681e-06, | |
| "loss": 0.2242, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.74329434526427e-06, | |
| "loss": 0.2114, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 8.713980007621728e-06, | |
| "loss": 0.2089, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.684665669979188e-06, | |
| "loss": 0.2169, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.655351332336648e-06, | |
| "loss": 0.2159, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.626036994694106e-06, | |
| "loss": 0.237, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.596722657051565e-06, | |
| "loss": 0.228, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 8.567408319409023e-06, | |
| "loss": 0.2304, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.538093981766483e-06, | |
| "loss": 0.2276, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.508779644123941e-06, | |
| "loss": 0.2271, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.479465306481401e-06, | |
| "loss": 0.2091, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 8.450150968838859e-06, | |
| "loss": 0.2101, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.420836631196319e-06, | |
| "loss": 0.2196, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.391522293553777e-06, | |
| "loss": 0.2106, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.362207955911237e-06, | |
| "loss": 0.2219, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.332893618268696e-06, | |
| "loss": 0.2194, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 8.303579280626155e-06, | |
| "loss": 0.2282, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.274264942983614e-06, | |
| "loss": 0.2182, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "eval_accuracy": 0.892851842691071, | |
| "eval_combined_score": 0.8763560455691379, | |
| "eval_f1": 0.859860248447205, | |
| "eval_loss": 0.27277621626853943, | |
| "eval_runtime": 108.5249, | |
| "eval_samples_per_second": 372.541, | |
| "eval_steps_per_second": 23.285, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.244950605341074e-06, | |
| "loss": 0.2324, | |
| "step": 20050 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 8.215636267698532e-06, | |
| "loss": 0.2174, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 8.186321930055992e-06, | |
| "loss": 0.2112, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.15700759241345e-06, | |
| "loss": 0.2287, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.127693254770908e-06, | |
| "loss": 0.2203, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.098378917128368e-06, | |
| "loss": 0.2167, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.069064579485828e-06, | |
| "loss": 0.2452, | |
| "step": 20350 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.039750241843286e-06, | |
| "loss": 0.2542, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.010435904200745e-06, | |
| "loss": 0.2513, | |
| "step": 20450 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.981121566558205e-06, | |
| "loss": 0.224, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.951807228915663e-06, | |
| "loss": 0.2036, | |
| "step": 20550 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 7.922492891273123e-06, | |
| "loss": 0.2024, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.893178553630581e-06, | |
| "loss": 0.2305, | |
| "step": 20650 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.86386421598804e-06, | |
| "loss": 0.2691, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 7.8345498783455e-06, | |
| "loss": 0.1883, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.805235540702959e-06, | |
| "loss": 0.226, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.775921203060417e-06, | |
| "loss": 0.2196, | |
| "step": 20850 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.746606865417876e-06, | |
| "loss": 0.2322, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 7.717292527775334e-06, | |
| "loss": 0.1952, | |
| "step": 20950 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.687978190132794e-06, | |
| "loss": 0.2394, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.658663852490254e-06, | |
| "loss": 0.2108, | |
| "step": 21050 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 7.629349514847713e-06, | |
| "loss": 0.2373, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 7.600035177205172e-06, | |
| "loss": 0.2263, | |
| "step": 21150 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 7.570720839562631e-06, | |
| "loss": 0.2075, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 7.54140650192009e-06, | |
| "loss": 0.2008, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 7.512092164277549e-06, | |
| "loss": 0.2093, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 7.482777826635008e-06, | |
| "loss": 0.2182, | |
| "step": 21350 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 7.453463488992467e-06, | |
| "loss": 0.2277, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.424149151349925e-06, | |
| "loss": 0.2163, | |
| "step": 21450 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.394834813707384e-06, | |
| "loss": 0.208, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 7.365520476064844e-06, | |
| "loss": 0.2298, | |
| "step": 21550 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 7.336206138422303e-06, | |
| "loss": 0.2268, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 7.306891800779762e-06, | |
| "loss": 0.2359, | |
| "step": 21650 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 7.277577463137221e-06, | |
| "loss": 0.2401, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 7.24826312549468e-06, | |
| "loss": 0.2502, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.218948787852139e-06, | |
| "loss": 0.2345, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.189634450209598e-06, | |
| "loss": 0.2177, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.160320112567057e-06, | |
| "loss": 0.2111, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.131005774924516e-06, | |
| "loss": 0.2534, | |
| "step": 21950 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.101691437281976e-06, | |
| "loss": 0.2255, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.072377099639435e-06, | |
| "loss": 0.2162, | |
| "step": 22050 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 7.043062761996893e-06, | |
| "loss": 0.2019, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 7.013748424354352e-06, | |
| "loss": 0.2259, | |
| "step": 22150 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.984434086711811e-06, | |
| "loss": 0.2122, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.9551197490692704e-06, | |
| "loss": 0.2282, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 6.925805411426729e-06, | |
| "loss": 0.23, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.896491073784188e-06, | |
| "loss": 0.2014, | |
| "step": 22350 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.867176736141647e-06, | |
| "loss": 0.2089, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 6.837862398499107e-06, | |
| "loss": 0.2239, | |
| "step": 22450 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.808548060856566e-06, | |
| "loss": 0.217, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.779233723214025e-06, | |
| "loss": 0.2397, | |
| "step": 22550 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.749919385571484e-06, | |
| "loss": 0.2234, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 6.7206050479289434e-06, | |
| "loss": 0.2158, | |
| "step": 22650 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.6912907102864015e-06, | |
| "loss": 0.2288, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.66197637264386e-06, | |
| "loss": 0.2153, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.632662035001319e-06, | |
| "loss": 0.1533, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.603347697358778e-06, | |
| "loss": 0.1545, | |
| "step": 22850 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 6.574033359716237e-06, | |
| "loss": 0.2142, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.544719022073697e-06, | |
| "loss": 0.1888, | |
| "step": 22950 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.515404684431156e-06, | |
| "loss": 0.1869, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.486090346788615e-06, | |
| "loss": 0.1831, | |
| "step": 23050 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 6.456776009146074e-06, | |
| "loss": 0.1727, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.427461671503533e-06, | |
| "loss": 0.1682, | |
| "step": 23150 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.398147333860992e-06, | |
| "loss": 0.1801, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.368832996218451e-06, | |
| "loss": 0.1928, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.33951865857591e-06, | |
| "loss": 0.1887, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 6.310204320933368e-06, | |
| "loss": 0.197, | |
| "step": 23350 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.280889983290828e-06, | |
| "loss": 0.1789, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.251575645648287e-06, | |
| "loss": 0.1624, | |
| "step": 23450 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.222261308005746e-06, | |
| "loss": 0.211, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 6.192946970363205e-06, | |
| "loss": 0.1852, | |
| "step": 23550 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.1636326327206645e-06, | |
| "loss": 0.1765, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.134318295078123e-06, | |
| "loss": 0.1457, | |
| "step": 23650 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.105003957435582e-06, | |
| "loss": 0.1871, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.075689619793041e-06, | |
| "loss": 0.1792, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 6.046375282150501e-06, | |
| "loss": 0.1884, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 6.01706094450796e-06, | |
| "loss": 0.1897, | |
| "step": 23850 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.987746606865419e-06, | |
| "loss": 0.2121, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.958432269222877e-06, | |
| "loss": 0.1678, | |
| "step": 23950 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.929117931580336e-06, | |
| "loss": 0.1788, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.8998035939377955e-06, | |
| "loss": 0.1722, | |
| "step": 24050 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.8704892562952545e-06, | |
| "loss": 0.2041, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 5.841174918652713e-06, | |
| "loss": 0.1505, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.811860581010172e-06, | |
| "loss": 0.1931, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.782546243367631e-06, | |
| "loss": 0.1806, | |
| "step": 24250 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.753231905725091e-06, | |
| "loss": 0.1732, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 5.72391756808255e-06, | |
| "loss": 0.1724, | |
| "step": 24350 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.694603230440009e-06, | |
| "loss": 0.1829, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.665288892797468e-06, | |
| "loss": 0.1763, | |
| "step": 24450 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.6359745551549274e-06, | |
| "loss": 0.1881, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.606660217512386e-06, | |
| "loss": 0.1708, | |
| "step": 24550 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.5773458798698444e-06, | |
| "loss": 0.1593, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.548031542227303e-06, | |
| "loss": 0.1714, | |
| "step": 24650 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.518717204584762e-06, | |
| "loss": 0.2073, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.489402866942222e-06, | |
| "loss": 0.1764, | |
| "step": 24750 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 5.460088529299681e-06, | |
| "loss": 0.1986, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.43077419165714e-06, | |
| "loss": 0.1809, | |
| "step": 24850 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.401459854014599e-06, | |
| "loss": 0.1715, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.3721455163720585e-06, | |
| "loss": 0.1789, | |
| "step": 24950 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.3428311787295174e-06, | |
| "loss": 0.1682, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "eval_accuracy": 0.8971061093247589, | |
| "eval_combined_score": 0.8792012275084717, | |
| "eval_f1": 0.8612963456921847, | |
| "eval_loss": 0.2921839952468872, | |
| "eval_runtime": 108.5114, | |
| "eval_samples_per_second": 372.587, | |
| "eval_steps_per_second": 23.288, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.313516841086976e-06, | |
| "loss": 0.1766, | |
| "step": 25050 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.284202503444435e-06, | |
| "loss": 0.1648, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.254888165801895e-06, | |
| "loss": 0.1602, | |
| "step": 25150 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.225573828159353e-06, | |
| "loss": 0.1962, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.196259490516812e-06, | |
| "loss": 0.1605, | |
| "step": 25250 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 5.166945152874271e-06, | |
| "loss": 0.1723, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.13763081523173e-06, | |
| "loss": 0.1842, | |
| "step": 25350 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.108316477589189e-06, | |
| "loss": 0.1767, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.0790021399466485e-06, | |
| "loss": 0.1709, | |
| "step": 25450 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.049687802304107e-06, | |
| "loss": 0.1889, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 5.020373464661566e-06, | |
| "loss": 0.1729, | |
| "step": 25550 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.991059127019025e-06, | |
| "loss": 0.1766, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.961744789376484e-06, | |
| "loss": 0.1832, | |
| "step": 25650 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.932430451733943e-06, | |
| "loss": 0.1493, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 4.903116114091403e-06, | |
| "loss": 0.1591, | |
| "step": 25750 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.873801776448862e-06, | |
| "loss": 0.1777, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 4.844487438806321e-06, | |
| "loss": 0.1789, | |
| "step": 25850 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.8151731011637796e-06, | |
| "loss": 0.1729, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.785858763521239e-06, | |
| "loss": 0.1543, | |
| "step": 25950 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.756544425878697e-06, | |
| "loss": 0.1753, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 4.727230088236156e-06, | |
| "loss": 0.1923, | |
| "step": 26050 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.697915750593616e-06, | |
| "loss": 0.169, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.668601412951075e-06, | |
| "loss": 0.194, | |
| "step": 26150 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.639287075308534e-06, | |
| "loss": 0.1677, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.609972737665993e-06, | |
| "loss": 0.1526, | |
| "step": 26250 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 4.580658400023452e-06, | |
| "loss": 0.1594, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.551344062380911e-06, | |
| "loss": 0.192, | |
| "step": 26350 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.5220297247383695e-06, | |
| "loss": 0.1816, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.492715387095829e-06, | |
| "loss": 0.1736, | |
| "step": 26450 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.463401049453288e-06, | |
| "loss": 0.1858, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 4.434086711810747e-06, | |
| "loss": 0.189, | |
| "step": 26550 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.404772374168206e-06, | |
| "loss": 0.1804, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.375458036525665e-06, | |
| "loss": 0.1824, | |
| "step": 26650 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.346143698883124e-06, | |
| "loss": 0.1887, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 4.316829361240583e-06, | |
| "loss": 0.1928, | |
| "step": 26750 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.2875150235980425e-06, | |
| "loss": 0.1691, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.2582006859555014e-06, | |
| "loss": 0.1863, | |
| "step": 26850 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.22888634831296e-06, | |
| "loss": 0.1665, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.199572010670419e-06, | |
| "loss": 0.1657, | |
| "step": 26950 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 4.170257673027878e-06, | |
| "loss": 0.1967, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.140943335385337e-06, | |
| "loss": 0.1775, | |
| "step": 27050 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.111628997742797e-06, | |
| "loss": 0.1769, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.082314660100256e-06, | |
| "loss": 0.1711, | |
| "step": 27150 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 4.053000322457714e-06, | |
| "loss": 0.1637, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.023685984815174e-06, | |
| "loss": 0.2147, | |
| "step": 27250 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.9943716471726325e-06, | |
| "loss": 0.189, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.965057309530091e-06, | |
| "loss": 0.1644, | |
| "step": 27350 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.93574297188755e-06, | |
| "loss": 0.1634, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.90642863424501e-06, | |
| "loss": 0.1792, | |
| "step": 27450 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.877114296602469e-06, | |
| "loss": 0.158, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 3.847799958959927e-06, | |
| "loss": 0.1526, | |
| "step": 27550 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.818485621317387e-06, | |
| "loss": 0.1907, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.7891712836748457e-06, | |
| "loss": 0.1798, | |
| "step": 27650 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.7598569460323047e-06, | |
| "loss": 0.1727, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.730542608389764e-06, | |
| "loss": 0.1754, | |
| "step": 27750 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.701228270747223e-06, | |
| "loss": 0.1563, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 3.6719139331046814e-06, | |
| "loss": 0.2006, | |
| "step": 27850 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 3.6425995954621407e-06, | |
| "loss": 0.1726, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.6132852578195996e-06, | |
| "loss": 0.1851, | |
| "step": 27950 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 3.583970920177059e-06, | |
| "loss": 0.1538, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.554656582534518e-06, | |
| "loss": 0.1685, | |
| "step": 28050 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.5253422448919772e-06, | |
| "loss": 0.1913, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.4960279072494357e-06, | |
| "loss": 0.2014, | |
| "step": 28150 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.466713569606895e-06, | |
| "loss": 0.1738, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.437399231964354e-06, | |
| "loss": 0.1683, | |
| "step": 28250 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.4080848943218133e-06, | |
| "loss": 0.1738, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.3787705566792722e-06, | |
| "loss": 0.1868, | |
| "step": 28350 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 3.3494562190367316e-06, | |
| "loss": 0.192, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 3.32014188139419e-06, | |
| "loss": 0.1733, | |
| "step": 28450 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.290827543751649e-06, | |
| "loss": 0.1772, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.2615132061091083e-06, | |
| "loss": 0.1849, | |
| "step": 28550 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.232198868466567e-06, | |
| "loss": 0.1922, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.2028845308240265e-06, | |
| "loss": 0.1834, | |
| "step": 28650 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.1735701931814855e-06, | |
| "loss": 0.1739, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.144255855538944e-06, | |
| "loss": 0.1975, | |
| "step": 28750 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.1149415178964033e-06, | |
| "loss": 0.1556, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.085627180253862e-06, | |
| "loss": 0.2029, | |
| "step": 28850 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 3.0563128426113215e-06, | |
| "loss": 0.1686, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 3.0269985049687804e-06, | |
| "loss": 0.1692, | |
| "step": 28950 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.9976841673262398e-06, | |
| "loss": 0.1613, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.9683698296836987e-06, | |
| "loss": 0.1567, | |
| "step": 29050 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.939055492041157e-06, | |
| "loss": 0.1899, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.9097411543986165e-06, | |
| "loss": 0.1601, | |
| "step": 29150 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.8804268167560754e-06, | |
| "loss": 0.1773, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.8511124791135348e-06, | |
| "loss": 0.1615, | |
| "step": 29250 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.8217981414709937e-06, | |
| "loss": 0.1696, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.792483803828453e-06, | |
| "loss": 0.1772, | |
| "step": 29350 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.7631694661859115e-06, | |
| "loss": 0.1775, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.733855128543371e-06, | |
| "loss": 0.1744, | |
| "step": 29450 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.7045407909008298e-06, | |
| "loss": 0.1854, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.675226453258289e-06, | |
| "loss": 0.1635, | |
| "step": 29550 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.645912115615748e-06, | |
| "loss": 0.1976, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.616597777973207e-06, | |
| "loss": 0.1896, | |
| "step": 29650 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.587283440330666e-06, | |
| "loss": 0.1705, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.5579691026881247e-06, | |
| "loss": 0.1654, | |
| "step": 29750 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.528654765045584e-06, | |
| "loss": 0.1769, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.499340427403043e-06, | |
| "loss": 0.189, | |
| "step": 29850 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.4700260897605023e-06, | |
| "loss": 0.1885, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.440711752117961e-06, | |
| "loss": 0.1983, | |
| "step": 29950 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.41139741447542e-06, | |
| "loss": 0.175, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "eval_accuracy": 0.8981449418748454, | |
| "eval_combined_score": 0.8808213439154127, | |
| "eval_f1": 0.8634977459559798, | |
| "eval_loss": 0.27546945214271545, | |
| "eval_runtime": 108.4817, | |
| "eval_samples_per_second": 372.689, | |
| "eval_steps_per_second": 23.294, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.382083076832879e-06, | |
| "loss": 0.1559, | |
| "step": 30050 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.352768739190338e-06, | |
| "loss": 0.178, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 2.3234544015477973e-06, | |
| "loss": 0.1692, | |
| "step": 30150 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.2941400639052562e-06, | |
| "loss": 0.168, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.264825726262715e-06, | |
| "loss": 0.1911, | |
| "step": 30250 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.2355113886201745e-06, | |
| "loss": 0.1599, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.2061970509776334e-06, | |
| "loss": 0.1828, | |
| "step": 30350 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.1768827133350923e-06, | |
| "loss": 0.1715, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.1475683756925512e-06, | |
| "loss": 0.1741, | |
| "step": 30450 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.1182540380500106e-06, | |
| "loss": 0.1941, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.0889397004074695e-06, | |
| "loss": 0.1754, | |
| "step": 30550 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 2.0596253627649284e-06, | |
| "loss": 0.1949, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.0303110251223877e-06, | |
| "loss": 0.1752, | |
| "step": 30650 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.0009966874798466e-06, | |
| "loss": 0.1692, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 1.9716823498373055e-06, | |
| "loss": 0.1694, | |
| "step": 30750 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.9423680121947645e-06, | |
| "loss": 0.1882, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.9130536745522234e-06, | |
| "loss": 0.1981, | |
| "step": 30850 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.8837393369096827e-06, | |
| "loss": 0.1724, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.8544249992671418e-06, | |
| "loss": 0.1739, | |
| "step": 30950 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.825110661624601e-06, | |
| "loss": 0.1922, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.7957963239820597e-06, | |
| "loss": 0.1775, | |
| "step": 31050 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.7664819863395188e-06, | |
| "loss": 0.1685, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.737167648696978e-06, | |
| "loss": 0.1481, | |
| "step": 31150 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 1.7078533110544368e-06, | |
| "loss": 0.1634, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.678538973411896e-06, | |
| "loss": 0.1831, | |
| "step": 31250 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.649224635769355e-06, | |
| "loss": 0.1661, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.619910298126814e-06, | |
| "loss": 0.1591, | |
| "step": 31350 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.590595960484273e-06, | |
| "loss": 0.1593, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.561281622841732e-06, | |
| "loss": 0.1637, | |
| "step": 31450 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.531967285199191e-06, | |
| "loss": 0.1647, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.50265294755665e-06, | |
| "loss": 0.1548, | |
| "step": 31550 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.4733386099141092e-06, | |
| "loss": 0.1756, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 1.444024272271568e-06, | |
| "loss": 0.1735, | |
| "step": 31650 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.4147099346290272e-06, | |
| "loss": 0.1735, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.3853955969864864e-06, | |
| "loss": 0.1845, | |
| "step": 31750 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.356081259343945e-06, | |
| "loss": 0.161, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.3267669217014042e-06, | |
| "loss": 0.1789, | |
| "step": 31850 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.2974525840588633e-06, | |
| "loss": 0.1522, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.2681382464163222e-06, | |
| "loss": 0.1866, | |
| "step": 31950 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.2388239087737813e-06, | |
| "loss": 0.1702, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.2095095711312405e-06, | |
| "loss": 0.1787, | |
| "step": 32050 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 1.1801952334886994e-06, | |
| "loss": 0.1858, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.1508808958461585e-06, | |
| "loss": 0.2176, | |
| "step": 32150 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.1215665582036176e-06, | |
| "loss": 0.1612, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.0922522205610765e-06, | |
| "loss": 0.1644, | |
| "step": 32250 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.0629378829185355e-06, | |
| "loss": 0.1634, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.0336235452759946e-06, | |
| "loss": 0.1578, | |
| "step": 32350 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 1.0043092076334537e-06, | |
| "loss": 0.1791, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 9.749948699909126e-07, | |
| "loss": 0.1682, | |
| "step": 32450 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.456805323483717e-07, | |
| "loss": 0.1675, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.163661947058307e-07, | |
| "loss": 0.1825, | |
| "step": 32550 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.870518570632897e-07, | |
| "loss": 0.1627, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 8.577375194207488e-07, | |
| "loss": 0.1656, | |
| "step": 32650 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.284231817782078e-07, | |
| "loss": 0.1714, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.991088441356669e-07, | |
| "loss": 0.1863, | |
| "step": 32750 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.697945064931259e-07, | |
| "loss": 0.1515, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.404801688505849e-07, | |
| "loss": 0.1484, | |
| "step": 32850 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 7.11165831208044e-07, | |
| "loss": 0.179, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.818514935655029e-07, | |
| "loss": 0.1809, | |
| "step": 32950 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.525371559229619e-07, | |
| "loss": 0.1483, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 6.23222818280421e-07, | |
| "loss": 0.1535, | |
| "step": 33050 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 5.939084806378801e-07, | |
| "loss": 0.1857, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.645941429953391e-07, | |
| "loss": 0.2079, | |
| "step": 33150 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.352798053527981e-07, | |
| "loss": 0.1643, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.059654677102571e-07, | |
| "loss": 0.1871, | |
| "step": 33250 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.7665113006771615e-07, | |
| "loss": 0.1981, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 4.473367924251752e-07, | |
| "loss": 0.1952, | |
| "step": 33350 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 4.180224547826342e-07, | |
| "loss": 0.2004, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.8870811714009326e-07, | |
| "loss": 0.1624, | |
| "step": 33450 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.5939377949755233e-07, | |
| "loss": 0.1696, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.300794418550113e-07, | |
| "loss": 0.1533, | |
| "step": 33550 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 3.007651042124703e-07, | |
| "loss": 0.1755, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.714507665699294e-07, | |
| "loss": 0.1636, | |
| "step": 33650 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.421364289273884e-07, | |
| "loss": 0.1827, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.1282209128484745e-07, | |
| "loss": 0.1692, | |
| "step": 33750 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 1.8350775364230647e-07, | |
| "loss": 0.1955, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.5419341599976551e-07, | |
| "loss": 0.1757, | |
| "step": 33850 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.2487907835722453e-07, | |
| "loss": 0.1725, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 9.556474071468355e-08, | |
| "loss": 0.1716, | |
| "step": 33950 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 6.62504030721426e-08, | |
| "loss": 0.1612, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 3.693606542960162e-08, | |
| "loss": 0.1891, | |
| "step": 34050 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.621727787060652e-09, | |
| "loss": 0.1562, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 34113, | |
| "total_flos": 8.382203265765098e+16, | |
| "train_loss": 0.23788358676445095, | |
| "train_runtime": 10630.4787, | |
| "train_samples_per_second": 102.68, | |
| "train_steps_per_second": 3.209 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 34113, | |
| "num_train_epochs": 3, | |
| "save_steps": 5000, | |
| "total_flos": 8.382203265765098e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |