| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.3137151865475565, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.006568575932737782, | |
| "grad_norm": 4.938570976257324, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.1658, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.013137151865475564, | |
| "grad_norm": 3.4416959285736084, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.9316, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.01970572779821335, | |
| "grad_norm": 3.3691599369049072, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.753, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.02627430373095113, | |
| "grad_norm": 3.2769742012023926, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.6576, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.032842879663688915, | |
| "grad_norm": 2.5112249851226807, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.6188, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.0394114555964267, | |
| "grad_norm": 3.0561046600341797, | |
| "learning_rate": 3e-06, | |
| "loss": 0.6021, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.045980031529164474, | |
| "grad_norm": 2.563472032546997, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.564, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.05254860746190226, | |
| "grad_norm": 2.6831862926483154, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.5314, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05911718339464004, | |
| "grad_norm": 2.383794069290161, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.5376, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.06568575932737783, | |
| "grad_norm": 2.8369593620300293, | |
| "learning_rate": 5e-06, | |
| "loss": 0.5404, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.07225433526011561, | |
| "grad_norm": 2.641814708709717, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.5014, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.0788229111928534, | |
| "grad_norm": 2.549522638320923, | |
| "learning_rate": 6e-06, | |
| "loss": 0.4978, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.08539148712559118, | |
| "grad_norm": 2.770048141479492, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.5135, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.09196006305832895, | |
| "grad_norm": 3.170330286026001, | |
| "learning_rate": 7e-06, | |
| "loss": 0.504, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.09852863899106673, | |
| "grad_norm": 2.712066650390625, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.451, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.10509721492380451, | |
| "grad_norm": 2.3632166385650635, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.4533, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.1116657908565423, | |
| "grad_norm": 2.728026866912842, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.4732, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.11823436678928008, | |
| "grad_norm": 2.7299931049346924, | |
| "learning_rate": 9e-06, | |
| "loss": 0.4255, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.12480294272201786, | |
| "grad_norm": 2.4892563819885254, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.4254, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.13137151865475566, | |
| "grad_norm": 2.3553197383880615, | |
| "learning_rate": 1e-05, | |
| "loss": 0.4218, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.13794009458749343, | |
| "grad_norm": 2.1275930404663086, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.4159, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.14450867052023122, | |
| "grad_norm": 2.166019916534424, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.4051, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.151077246452969, | |
| "grad_norm": 2.5494415760040283, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.399, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.1576458223857068, | |
| "grad_norm": 2.482174873352051, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.4013, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.16421439831844456, | |
| "grad_norm": 2.3230655193328857, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.3736, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.17078297425118236, | |
| "grad_norm": 2.1272637844085693, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.3715, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.17735155018392013, | |
| "grad_norm": 2.2144370079040527, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.389, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.1839201261166579, | |
| "grad_norm": 2.156562566757202, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.3779, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.1904887020493957, | |
| "grad_norm": 1.9418479204177856, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.3624, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.19705727798213346, | |
| "grad_norm": 1.9420461654663086, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.3726, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.20362585391487126, | |
| "grad_norm": 1.929376482963562, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.3501, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.21019442984760903, | |
| "grad_norm": 2.257051467895508, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.3755, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.21676300578034682, | |
| "grad_norm": 2.102417230606079, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.3671, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.2233315817130846, | |
| "grad_norm": 2.096370220184326, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.3397, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.2299001576458224, | |
| "grad_norm": 2.0005109310150146, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.3467, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.23646873357856016, | |
| "grad_norm": 2.1975176334381104, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.3341, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.24303730951129796, | |
| "grad_norm": 2.0950164794921875, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.3398, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.24960588544403572, | |
| "grad_norm": 2.0137791633605957, | |
| "learning_rate": 9e-06, | |
| "loss": 0.3183, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.2561744613767735, | |
| "grad_norm": 1.97981595993042, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.3653, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.2627430373095113, | |
| "grad_norm": 1.9213696718215942, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.3429, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.2627430373095113, | |
| "eval_loss": 0.3295910656452179, | |
| "eval_runtime": 594.427, | |
| "eval_samples_per_second": 2.052, | |
| "eval_steps_per_second": 0.13, | |
| "eval_wer": 0.2299665865525529, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.26931161324224906, | |
| "grad_norm": 2.122868299484253, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.318, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.27588018917498686, | |
| "grad_norm": 2.259373426437378, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.3364, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.28244876510772465, | |
| "grad_norm": 1.978367567062378, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.3125, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.28901734104046245, | |
| "grad_norm": 1.8453904390335083, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.3009, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.2955859169732002, | |
| "grad_norm": 2.002053737640381, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.3087, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.302154492905938, | |
| "grad_norm": 2.080690860748291, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.3188, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.3087230688386758, | |
| "grad_norm": 1.6879727840423584, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3043, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.3152916447714136, | |
| "grad_norm": 2.2736353874206543, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.3165, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.3218602207041513, | |
| "grad_norm": 1.8325968980789185, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.3236, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.3284287966368891, | |
| "grad_norm": 1.8235770463943481, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.3057, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.3349973725696269, | |
| "grad_norm": 2.09566068649292, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.2845, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.3415659485023647, | |
| "grad_norm": 1.9097365140914917, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.2833, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.34813452443510245, | |
| "grad_norm": 1.9709367752075195, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.294, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.35470310036784025, | |
| "grad_norm": 1.9940749406814575, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.2789, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.36127167630057805, | |
| "grad_norm": 2.122657299041748, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.3026, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.3678402522333158, | |
| "grad_norm": 1.742019772529602, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.2843, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.3744088281660536, | |
| "grad_norm": 1.8771026134490967, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.2937, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.3809774040987914, | |
| "grad_norm": 1.617311954498291, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.2749, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.3875459800315292, | |
| "grad_norm": 1.902928113937378, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.2929, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.3941145559642669, | |
| "grad_norm": 1.9772895574569702, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.2699, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.4006831318970047, | |
| "grad_norm": 1.7294281721115112, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.288, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.4072517078297425, | |
| "grad_norm": 1.6163759231567383, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.2919, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.4138202837624803, | |
| "grad_norm": 1.9823168516159058, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.2691, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.42038885969521805, | |
| "grad_norm": 2.085510730743408, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.2807, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.42695743562795585, | |
| "grad_norm": 2.1536219120025635, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.2977, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.43352601156069365, | |
| "grad_norm": 2.23018741607666, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.291, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.44009458749343144, | |
| "grad_norm": 1.8097656965255737, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.3125, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.4466631634261692, | |
| "grad_norm": 1.557788372039795, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.2743, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.453231739358907, | |
| "grad_norm": 1.677343726158142, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.2568, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.4598003152916448, | |
| "grad_norm": 1.7231301069259644, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.2701, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.4663688912243826, | |
| "grad_norm": 1.632379412651062, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.2795, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.4729374671571203, | |
| "grad_norm": 2.042989730834961, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.2714, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.4795060430898581, | |
| "grad_norm": 1.8419965505599976, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.2638, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.4860746190225959, | |
| "grad_norm": 1.7692322731018066, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2631, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.4926431949553337, | |
| "grad_norm": 1.8727885484695435, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.2647, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.49921177088807145, | |
| "grad_norm": 1.5731993913650513, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.2765, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.5057803468208093, | |
| "grad_norm": 1.9374446868896484, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.2672, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.512348922753547, | |
| "grad_norm": 1.702355980873108, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.2574, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.5189174986862848, | |
| "grad_norm": 1.539027214050293, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.2651, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.5254860746190226, | |
| "grad_norm": 1.8967093229293823, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2636, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.5254860746190226, | |
| "eval_loss": 0.2615206837654114, | |
| "eval_runtime": 532.8353, | |
| "eval_samples_per_second": 2.29, | |
| "eval_steps_per_second": 0.145, | |
| "eval_wer": 0.1925669734451023, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.5320546505517604, | |
| "grad_norm": 1.688244342803955, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.2683, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.5386232264844981, | |
| "grad_norm": 1.8095322847366333, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.2707, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.545191802417236, | |
| "grad_norm": 1.8945379257202148, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.2535, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.5517603783499737, | |
| "grad_norm": 1.9910788536071777, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.2486, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.5583289542827115, | |
| "grad_norm": 1.3568848371505737, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.2755, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.5648975302154493, | |
| "grad_norm": 1.7561497688293457, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.2679, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.571466106148187, | |
| "grad_norm": 1.6052385568618774, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.2497, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.5780346820809249, | |
| "grad_norm": 1.5882091522216797, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.2673, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.5846032580136626, | |
| "grad_norm": 1.7540193796157837, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.2335, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.5911718339464004, | |
| "grad_norm": 1.4882384538650513, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.2555, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.5977404098791382, | |
| "grad_norm": 2.00600266456604, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.2687, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.604308985811876, | |
| "grad_norm": 1.4347535371780396, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2681, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.6108775617446137, | |
| "grad_norm": 1.8053362369537354, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.2568, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.6174461376773516, | |
| "grad_norm": 1.657359004020691, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.246, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.6240147136100893, | |
| "grad_norm": 1.6917238235473633, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.2333, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.6305832895428272, | |
| "grad_norm": 2.0534191131591797, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.2557, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.6371518654755649, | |
| "grad_norm": 1.6997853517532349, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.2888, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.6437204414083026, | |
| "grad_norm": 1.7183536291122437, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.2364, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.6502890173410405, | |
| "grad_norm": 1.6452535390853882, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.2395, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.6568575932737782, | |
| "grad_norm": 1.7632161378860474, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.2547, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.663426169206516, | |
| "grad_norm": 1.9056440591812134, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.249, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.6699947451392538, | |
| "grad_norm": 2.083576202392578, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.2596, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.6765633210719916, | |
| "grad_norm": 2.1211483478546143, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.2435, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.6831318970047294, | |
| "grad_norm": 1.599048376083374, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.2376, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.6897004729374672, | |
| "grad_norm": 1.6461100578308105, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.2457, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.6962690488702049, | |
| "grad_norm": 1.660069227218628, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.2354, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.7028376248029428, | |
| "grad_norm": 2.221201181411743, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.2794, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.7094062007356805, | |
| "grad_norm": 2.4395995140075684, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.2499, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.7159747766684182, | |
| "grad_norm": 1.4763926267623901, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.2783, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.7225433526011561, | |
| "grad_norm": 1.6573847532272339, | |
| "learning_rate": 5e-06, | |
| "loss": 0.2349, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.7291119285338938, | |
| "grad_norm": 1.8241550922393799, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.2449, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.7356805044666316, | |
| "grad_norm": 1.6332080364227295, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.2413, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.7422490803993694, | |
| "grad_norm": 1.626541256904602, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.2499, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.7488176563321072, | |
| "grad_norm": 1.5557328462600708, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.2534, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.755386232264845, | |
| "grad_norm": 1.932283878326416, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.2379, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.7619548081975828, | |
| "grad_norm": 1.9064332246780396, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.2323, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.7685233841303205, | |
| "grad_norm": 1.5515189170837402, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.2328, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.7750919600630584, | |
| "grad_norm": 2.1047863960266113, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.248, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.7816605359957961, | |
| "grad_norm": 1.8366841077804565, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.2584, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.7882291119285338, | |
| "grad_norm": 1.949874997138977, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.2316, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.7882291119285338, | |
| "eval_loss": 0.23398828506469727, | |
| "eval_runtime": 533.7623, | |
| "eval_samples_per_second": 2.286, | |
| "eval_steps_per_second": 0.144, | |
| "eval_wer": 0.1797877952986693, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.7947976878612717, | |
| "grad_norm": 2.0107836723327637, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.2421, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.8013662637940094, | |
| "grad_norm": 1.959560751914978, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.2319, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.8079348397267473, | |
| "grad_norm": 1.6217644214630127, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.2383, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.814503415659485, | |
| "grad_norm": 2.180353879928589, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.2374, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.8210719915922228, | |
| "grad_norm": 1.7979501485824585, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.2437, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.8276405675249606, | |
| "grad_norm": 1.6101832389831543, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.2142, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.8342091434576984, | |
| "grad_norm": 2.010740280151367, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.2489, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.8407777193904361, | |
| "grad_norm": 1.62699556350708, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.2362, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.847346295323174, | |
| "grad_norm": 1.6320332288742065, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.2134, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.8539148712559117, | |
| "grad_norm": 1.5167447328567505, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.219, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.8604834471886496, | |
| "grad_norm": 1.8277373313903809, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.2593, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.8670520231213873, | |
| "grad_norm": 1.6318010091781616, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.2151, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.873620599054125, | |
| "grad_norm": 1.9252204895019531, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.2509, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.8801891749868629, | |
| "grad_norm": 2.156442880630493, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.2407, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.8867577509196006, | |
| "grad_norm": 1.9615235328674316, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.2299, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.8933263268523384, | |
| "grad_norm": 1.985635757446289, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.2451, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.8998949027850762, | |
| "grad_norm": 1.5059348344802856, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.2389, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.906463478717814, | |
| "grad_norm": 1.8141510486602783, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.2412, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.9130320546505517, | |
| "grad_norm": 1.7190760374069214, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.217, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.9196006305832896, | |
| "grad_norm": 1.7403192520141602, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.2386, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.9261692065160273, | |
| "grad_norm": 1.5826787948608398, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.1965, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.9327377824487652, | |
| "grad_norm": 1.8713033199310303, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.2205, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.9393063583815029, | |
| "grad_norm": 1.6279629468917847, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.2031, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.9458749343142406, | |
| "grad_norm": 1.5993796586990356, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.218, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.9524435102469785, | |
| "grad_norm": 1.8748666048049927, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.2198, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.9590120861797162, | |
| "grad_norm": 2.0783231258392334, | |
| "learning_rate": 3e-06, | |
| "loss": 0.2285, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.965580662112454, | |
| "grad_norm": 1.6366628408432007, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.2244, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.9721492380451918, | |
| "grad_norm": 1.4321212768554688, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.2129, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.9787178139779296, | |
| "grad_norm": 1.9172074794769287, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.2308, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.9852863899106674, | |
| "grad_norm": 1.601163387298584, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.2249, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.9918549658434052, | |
| "grad_norm": 1.920453429222107, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.241, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.9984235417761429, | |
| "grad_norm": 1.421399712562561, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.245, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.0049921177088808, | |
| "grad_norm": 1.2747637033462524, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.183, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 1.0115606936416186, | |
| "grad_norm": 1.276963472366333, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.1669, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.0181292695743562, | |
| "grad_norm": 1.6604379415512085, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.1685, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 1.024697845507094, | |
| "grad_norm": 1.4884883165359497, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.1837, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.031266421439832, | |
| "grad_norm": 1.2239917516708374, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.161, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 1.0378349973725696, | |
| "grad_norm": 1.7419662475585938, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.1903, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.0444035733053074, | |
| "grad_norm": 1.5548486709594727, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.159, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 1.0509721492380453, | |
| "grad_norm": 1.7805577516555786, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.1779, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.0509721492380453, | |
| "eval_loss": 0.21787120401859283, | |
| "eval_runtime": 531.6511, | |
| "eval_samples_per_second": 2.295, | |
| "eval_steps_per_second": 0.145, | |
| "eval_wer": 0.16243625065947595, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.057540725170783, | |
| "grad_norm": 1.646399974822998, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.1891, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 1.0641093011035208, | |
| "grad_norm": 2.159637928009033, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.1712, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.0706778770362586, | |
| "grad_norm": 1.5517977476119995, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.1582, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 1.0772464529689962, | |
| "grad_norm": 1.410233974456787, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.1828, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.083815028901734, | |
| "grad_norm": 1.6741544008255005, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.1597, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 1.090383604834472, | |
| "grad_norm": 1.508376121520996, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.166, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.0969521807672096, | |
| "grad_norm": 1.4631690979003906, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.159, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 1.1035207566999474, | |
| "grad_norm": 1.7129231691360474, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.1682, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.1100893326326853, | |
| "grad_norm": 1.7223049402236938, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.1865, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 1.116657908565423, | |
| "grad_norm": 1.8834460973739624, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.1753, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.1232264844981608, | |
| "grad_norm": 1.6631660461425781, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.1585, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 1.1297950604308986, | |
| "grad_norm": 1.6180058717727661, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.1839, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.1363636363636362, | |
| "grad_norm": 1.635799527168274, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.1685, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 1.142932212296374, | |
| "grad_norm": 1.4893510341644287, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.1678, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.149500788229112, | |
| "grad_norm": 1.4803720712661743, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.1607, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 1.1560693641618498, | |
| "grad_norm": 1.4839766025543213, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.187, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.1626379400945874, | |
| "grad_norm": 1.518723487854004, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.1628, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 1.1692065160273253, | |
| "grad_norm": 1.740968942642212, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.1521, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.1757750919600631, | |
| "grad_norm": 1.941607117652893, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.1558, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 1.1823436678928008, | |
| "grad_norm": 1.47350013256073, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.1758, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.1889122438255386, | |
| "grad_norm": 1.4565749168395996, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.1804, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 1.1954808197582765, | |
| "grad_norm": 1.6238913536071777, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.1596, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.202049395691014, | |
| "grad_norm": 1.6127305030822754, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.1801, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 1.208617971623752, | |
| "grad_norm": 1.5954357385635376, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.1616, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.2151865475564898, | |
| "grad_norm": 1.5588889122009277, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.1727, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 1.2217551234892277, | |
| "grad_norm": 1.2232043743133545, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.1617, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.2283236994219653, | |
| "grad_norm": 1.3071914911270142, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.1623, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 1.2348922753547031, | |
| "grad_norm": 1.1346869468688965, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.155, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.241460851287441, | |
| "grad_norm": 1.572501540184021, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.1742, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 1.2480294272201786, | |
| "grad_norm": 1.6292036771774292, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.1535, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.2545980031529165, | |
| "grad_norm": 1.8764897584915161, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.1564, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 1.2611665790856543, | |
| "grad_norm": 1.7719995975494385, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.1779, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.267735155018392, | |
| "grad_norm": 1.4942494630813599, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.1743, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 1.2743037309511298, | |
| "grad_norm": 1.6207414865493774, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.1731, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.2808723068838677, | |
| "grad_norm": 1.4872610569000244, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.1708, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 1.2874408828166053, | |
| "grad_norm": 1.8898082971572876, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.1819, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.2940094587493431, | |
| "grad_norm": 1.4731193780899048, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.1714, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 1.300578034682081, | |
| "grad_norm": 1.4309362173080444, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.1668, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.3071466106148186, | |
| "grad_norm": 1.6482913494110107, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.1928, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 1.3137151865475565, | |
| "grad_norm": 1.3571888208389282, | |
| "learning_rate": 0.0, | |
| "loss": 0.1626, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.3137151865475565, | |
| "eval_loss": 0.21081987023353577, | |
| "eval_runtime": 712.9205, | |
| "eval_samples_per_second": 1.711, | |
| "eval_steps_per_second": 0.108, | |
| "eval_wer": 0.1561639017527405, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.3137151865475565, | |
| "step": 5000, | |
| "total_flos": 5.435725490631475e+20, | |
| "train_loss": 0.2802161669254303, | |
| "train_runtime": 36290.7707, | |
| "train_samples_per_second": 4.409, | |
| "train_steps_per_second": 0.138 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 2, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5.435725490631475e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |