| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.694760820045558, | |
| "eval_steps": 1000, | |
| "global_step": 5000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02847380410022779, | |
| "grad_norm": 8.5387601852417, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 1.7657, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.05694760820045558, | |
| "grad_norm": 5.922762393951416, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 1.3302, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.08542141230068337, | |
| "grad_norm": 6.087216854095459, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.9581, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.11389521640091116, | |
| "grad_norm": 5.635214328765869, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.8187, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.14236902050113895, | |
| "grad_norm": 4.90886926651001, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.7346, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.17084282460136674, | |
| "grad_norm": 5.0317230224609375, | |
| "learning_rate": 3e-06, | |
| "loss": 0.7124, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19931662870159453, | |
| "grad_norm": 5.094836711883545, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.6898, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.22779043280182232, | |
| "grad_norm": 4.328588962554932, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.649, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.25626423690205014, | |
| "grad_norm": 5.198021411895752, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.6179, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2847380410022779, | |
| "grad_norm": 4.642407417297363, | |
| "learning_rate": 5e-06, | |
| "loss": 0.6448, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.3132118451025057, | |
| "grad_norm": 4.232536315917969, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.6087, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.3416856492027335, | |
| "grad_norm": 4.383147239685059, | |
| "learning_rate": 6e-06, | |
| "loss": 0.5977, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3701594533029613, | |
| "grad_norm": 4.431395530700684, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.6024, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.39863325740318906, | |
| "grad_norm": 5.041561126708984, | |
| "learning_rate": 7e-06, | |
| "loss": 0.5695, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.4271070615034169, | |
| "grad_norm": 4.403401851654053, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.5453, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.45558086560364464, | |
| "grad_norm": 4.56962251663208, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.5746, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.48405466970387245, | |
| "grad_norm": 4.166983127593994, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.5548, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.5125284738041003, | |
| "grad_norm": 3.995704174041748, | |
| "learning_rate": 9e-06, | |
| "loss": 0.5251, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.541002277904328, | |
| "grad_norm": 4.352871417999268, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.529, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.5694760820045558, | |
| "grad_norm": 5.0144219398498535, | |
| "learning_rate": 1e-05, | |
| "loss": 0.5356, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.5979498861047836, | |
| "grad_norm": 4.578596591949463, | |
| "learning_rate": 9.944444444444445e-06, | |
| "loss": 0.5513, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.6264236902050114, | |
| "grad_norm": 4.121158123016357, | |
| "learning_rate": 9.88888888888889e-06, | |
| "loss": 0.5373, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6548974943052391, | |
| "grad_norm": 3.4131343364715576, | |
| "learning_rate": 9.833333333333333e-06, | |
| "loss": 0.5357, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.683371298405467, | |
| "grad_norm": 3.698122501373291, | |
| "learning_rate": 9.777777777777779e-06, | |
| "loss": 0.5219, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.7118451025056948, | |
| "grad_norm": 3.6946487426757812, | |
| "learning_rate": 9.722222222222223e-06, | |
| "loss": 0.525, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.7403189066059226, | |
| "grad_norm": 3.6262104511260986, | |
| "learning_rate": 9.666666666666667e-06, | |
| "loss": 0.5085, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7687927107061503, | |
| "grad_norm": 2.9753031730651855, | |
| "learning_rate": 9.611111111111112e-06, | |
| "loss": 0.4876, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.7972665148063781, | |
| "grad_norm": 3.8146708011627197, | |
| "learning_rate": 9.555555555555556e-06, | |
| "loss": 0.5017, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.8257403189066059, | |
| "grad_norm": 3.672109603881836, | |
| "learning_rate": 9.5e-06, | |
| "loss": 0.4733, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.8542141230068337, | |
| "grad_norm": 4.09952974319458, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.4667, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.8826879271070615, | |
| "grad_norm": 3.828016519546509, | |
| "learning_rate": 9.38888888888889e-06, | |
| "loss": 0.468, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.9111617312072893, | |
| "grad_norm": 3.191138744354248, | |
| "learning_rate": 9.333333333333334e-06, | |
| "loss": 0.4772, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.9396355353075171, | |
| "grad_norm": 4.035594463348389, | |
| "learning_rate": 9.277777777777778e-06, | |
| "loss": 0.4515, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.9681093394077449, | |
| "grad_norm": 3.4418606758117676, | |
| "learning_rate": 9.222222222222224e-06, | |
| "loss": 0.4821, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.9965831435079726, | |
| "grad_norm": 3.5060410499572754, | |
| "learning_rate": 9.166666666666666e-06, | |
| "loss": 0.4541, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 1.0250569476082005, | |
| "grad_norm": 2.4985129833221436, | |
| "learning_rate": 9.111111111111112e-06, | |
| "loss": 0.3697, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 1.0535307517084282, | |
| "grad_norm": 3.486703395843506, | |
| "learning_rate": 9.055555555555556e-06, | |
| "loss": 0.3357, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 1.082004555808656, | |
| "grad_norm": 2.972067356109619, | |
| "learning_rate": 9e-06, | |
| "loss": 0.344, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 1.1104783599088839, | |
| "grad_norm": 3.0161571502685547, | |
| "learning_rate": 8.944444444444446e-06, | |
| "loss": 0.3496, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 1.1389521640091116, | |
| "grad_norm": 3.0906615257263184, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.3452, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.1389521640091116, | |
| "eval_loss": 0.4483538568019867, | |
| "eval_runtime": 2994.5795, | |
| "eval_samples_per_second": 2.345, | |
| "eval_steps_per_second": 0.147, | |
| "eval_wer": 0.31664895030560725, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 1.1674259681093395, | |
| "grad_norm": 3.4060277938842773, | |
| "learning_rate": 8.833333333333334e-06, | |
| "loss": 0.3284, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 1.1958997722095672, | |
| "grad_norm": 3.258347988128662, | |
| "learning_rate": 8.777777777777778e-06, | |
| "loss": 0.342, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 1.224373576309795, | |
| "grad_norm": 3.067859172821045, | |
| "learning_rate": 8.722222222222224e-06, | |
| "loss": 0.3313, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 1.2528473804100229, | |
| "grad_norm": 3.383387804031372, | |
| "learning_rate": 8.666666666666668e-06, | |
| "loss": 0.3345, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 1.2813211845102506, | |
| "grad_norm": 2.9198620319366455, | |
| "learning_rate": 8.611111111111112e-06, | |
| "loss": 0.3493, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 1.3097949886104785, | |
| "grad_norm": 2.705611228942871, | |
| "learning_rate": 8.555555555555556e-06, | |
| "loss": 0.3438, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 1.3382687927107062, | |
| "grad_norm": 3.0571541786193848, | |
| "learning_rate": 8.5e-06, | |
| "loss": 0.3258, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 1.366742596810934, | |
| "grad_norm": 3.117966890335083, | |
| "learning_rate": 8.444444444444446e-06, | |
| "loss": 0.3437, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 1.3952164009111616, | |
| "grad_norm": 2.608323812484741, | |
| "learning_rate": 8.38888888888889e-06, | |
| "loss": 0.3287, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 1.4236902050113895, | |
| "grad_norm": 2.7301223278045654, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.3319, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 1.4521640091116172, | |
| "grad_norm": 2.6983442306518555, | |
| "learning_rate": 8.277777777777778e-06, | |
| "loss": 0.3139, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 1.4806378132118452, | |
| "grad_norm": 3.1055397987365723, | |
| "learning_rate": 8.222222222222222e-06, | |
| "loss": 0.3596, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 1.5091116173120729, | |
| "grad_norm": 2.650721311569214, | |
| "learning_rate": 8.166666666666668e-06, | |
| "loss": 0.3387, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 1.5375854214123006, | |
| "grad_norm": 2.6098825931549072, | |
| "learning_rate": 8.111111111111112e-06, | |
| "loss": 0.3329, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 1.5660592255125285, | |
| "grad_norm": 2.5943613052368164, | |
| "learning_rate": 8.055555555555557e-06, | |
| "loss": 0.34, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 1.5945330296127562, | |
| "grad_norm": 3.092444896697998, | |
| "learning_rate": 8.000000000000001e-06, | |
| "loss": 0.3329, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 1.6230068337129842, | |
| "grad_norm": 2.703401565551758, | |
| "learning_rate": 7.944444444444445e-06, | |
| "loss": 0.3309, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 1.6514806378132119, | |
| "grad_norm": 3.2039897441864014, | |
| "learning_rate": 7.88888888888889e-06, | |
| "loss": 0.3204, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 1.6799544419134396, | |
| "grad_norm": 2.8489060401916504, | |
| "learning_rate": 7.833333333333333e-06, | |
| "loss": 0.341, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 1.7084282460136673, | |
| "grad_norm": 3.1537883281707764, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.3361, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 1.7369020501138952, | |
| "grad_norm": 3.1526262760162354, | |
| "learning_rate": 7.722222222222223e-06, | |
| "loss": 0.3253, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 1.7653758542141231, | |
| "grad_norm": 3.4786808490753174, | |
| "learning_rate": 7.666666666666667e-06, | |
| "loss": 0.3271, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 1.7938496583143508, | |
| "grad_norm": 2.7717833518981934, | |
| "learning_rate": 7.611111111111111e-06, | |
| "loss": 0.3328, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 1.8223234624145785, | |
| "grad_norm": 2.6981112957000732, | |
| "learning_rate": 7.555555555555556e-06, | |
| "loss": 0.3314, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 1.8507972665148062, | |
| "grad_norm": 3.0883076190948486, | |
| "learning_rate": 7.500000000000001e-06, | |
| "loss": 0.35, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 1.8792710706150342, | |
| "grad_norm": 3.0751681327819824, | |
| "learning_rate": 7.444444444444445e-06, | |
| "loss": 0.3256, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 1.907744874715262, | |
| "grad_norm": 2.925309658050537, | |
| "learning_rate": 7.38888888888889e-06, | |
| "loss": 0.3386, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 1.9362186788154898, | |
| "grad_norm": 3.185584545135498, | |
| "learning_rate": 7.333333333333333e-06, | |
| "loss": 0.315, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 1.9646924829157175, | |
| "grad_norm": 3.3560593128204346, | |
| "learning_rate": 7.277777777777778e-06, | |
| "loss": 0.3273, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 1.9931662870159452, | |
| "grad_norm": 2.6927008628845215, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.3162, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 2.021640091116173, | |
| "grad_norm": 2.982485771179199, | |
| "learning_rate": 7.166666666666667e-06, | |
| "loss": 0.2377, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 2.050113895216401, | |
| "grad_norm": 2.3905413150787354, | |
| "learning_rate": 7.111111111111112e-06, | |
| "loss": 0.2009, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 2.078587699316629, | |
| "grad_norm": 2.3832273483276367, | |
| "learning_rate": 7.055555555555557e-06, | |
| "loss": 0.1986, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 2.1070615034168565, | |
| "grad_norm": 2.6094419956207275, | |
| "learning_rate": 7e-06, | |
| "loss": 0.2018, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 2.135535307517084, | |
| "grad_norm": 2.6407711505889893, | |
| "learning_rate": 6.944444444444445e-06, | |
| "loss": 0.1946, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 2.164009111617312, | |
| "grad_norm": 2.179900884628296, | |
| "learning_rate": 6.88888888888889e-06, | |
| "loss": 0.197, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 2.19248291571754, | |
| "grad_norm": 2.5007543563842773, | |
| "learning_rate": 6.833333333333334e-06, | |
| "loss": 0.2, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 2.2209567198177678, | |
| "grad_norm": 2.886491537094116, | |
| "learning_rate": 6.777777777777779e-06, | |
| "loss": 0.2098, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 2.2494305239179955, | |
| "grad_norm": 3.0158777236938477, | |
| "learning_rate": 6.7222222222222235e-06, | |
| "loss": 0.2073, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 2.277904328018223, | |
| "grad_norm": 2.4290666580200195, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.2029, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.277904328018223, | |
| "eval_loss": 0.4125473201274872, | |
| "eval_runtime": 2968.6025, | |
| "eval_samples_per_second": 2.365, | |
| "eval_steps_per_second": 0.148, | |
| "eval_wer": 0.28979537602976346, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 2.306378132118451, | |
| "grad_norm": 2.810905694961548, | |
| "learning_rate": 6.6111111111111115e-06, | |
| "loss": 0.1981, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 2.334851936218679, | |
| "grad_norm": 2.6154942512512207, | |
| "learning_rate": 6.555555555555556e-06, | |
| "loss": 0.2148, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 2.3633257403189067, | |
| "grad_norm": 2.636270761489868, | |
| "learning_rate": 6.5000000000000004e-06, | |
| "loss": 0.1926, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 2.3917995444191344, | |
| "grad_norm": 2.7535417079925537, | |
| "learning_rate": 6.444444444444445e-06, | |
| "loss": 0.2098, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 2.420273348519362, | |
| "grad_norm": 2.86851167678833, | |
| "learning_rate": 6.3888888888888885e-06, | |
| "loss": 0.2182, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 2.44874715261959, | |
| "grad_norm": 2.129084348678589, | |
| "learning_rate": 6.333333333333333e-06, | |
| "loss": 0.2137, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 2.477220956719818, | |
| "grad_norm": 2.456988573074341, | |
| "learning_rate": 6.277777777777778e-06, | |
| "loss": 0.1998, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 2.5056947608200457, | |
| "grad_norm": 2.829132080078125, | |
| "learning_rate": 6.222222222222223e-06, | |
| "loss": 0.2076, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 2.5341685649202734, | |
| "grad_norm": 3.0912461280822754, | |
| "learning_rate": 6.166666666666667e-06, | |
| "loss": 0.2047, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 2.562642369020501, | |
| "grad_norm": 2.693300247192383, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.2043, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 2.591116173120729, | |
| "grad_norm": 2.593743085861206, | |
| "learning_rate": 6.055555555555555e-06, | |
| "loss": 0.2023, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 2.619589977220957, | |
| "grad_norm": 3.2144663333892822, | |
| "learning_rate": 6e-06, | |
| "loss": 0.2013, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 2.6480637813211843, | |
| "grad_norm": 2.123487710952759, | |
| "learning_rate": 5.944444444444445e-06, | |
| "loss": 0.2129, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 2.6765375854214124, | |
| "grad_norm": 2.5355987548828125, | |
| "learning_rate": 5.88888888888889e-06, | |
| "loss": 0.1932, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 2.70501138952164, | |
| "grad_norm": 2.0367653369903564, | |
| "learning_rate": 5.833333333333334e-06, | |
| "loss": 0.2015, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 2.733485193621868, | |
| "grad_norm": 2.442157506942749, | |
| "learning_rate": 5.777777777777778e-06, | |
| "loss": 0.1905, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 2.7619589977220955, | |
| "grad_norm": 2.7985150814056396, | |
| "learning_rate": 5.722222222222222e-06, | |
| "loss": 0.2093, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 2.7904328018223232, | |
| "grad_norm": 2.2095937728881836, | |
| "learning_rate": 5.666666666666667e-06, | |
| "loss": 0.196, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 2.8189066059225514, | |
| "grad_norm": 2.453608751296997, | |
| "learning_rate": 5.611111111111112e-06, | |
| "loss": 0.2113, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 2.847380410022779, | |
| "grad_norm": 2.1963589191436768, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.1997, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 2.875854214123007, | |
| "grad_norm": 2.492926836013794, | |
| "learning_rate": 5.500000000000001e-06, | |
| "loss": 0.2123, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 2.9043280182232345, | |
| "grad_norm": 2.3640942573547363, | |
| "learning_rate": 5.444444444444445e-06, | |
| "loss": 0.1959, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 2.932801822323462, | |
| "grad_norm": 2.2937705516815186, | |
| "learning_rate": 5.388888888888889e-06, | |
| "loss": 0.1879, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 2.9612756264236904, | |
| "grad_norm": 2.901215076446533, | |
| "learning_rate": 5.333333333333334e-06, | |
| "loss": 0.2008, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 2.989749430523918, | |
| "grad_norm": 2.2094128131866455, | |
| "learning_rate": 5.2777777777777785e-06, | |
| "loss": 0.2039, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 3.0182232346241458, | |
| "grad_norm": 1.8943818807601929, | |
| "learning_rate": 5.2222222222222226e-06, | |
| "loss": 0.1459, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 3.0466970387243735, | |
| "grad_norm": 2.121443033218384, | |
| "learning_rate": 5.1666666666666675e-06, | |
| "loss": 0.1153, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 3.075170842824601, | |
| "grad_norm": 2.070572853088379, | |
| "learning_rate": 5.1111111111111115e-06, | |
| "loss": 0.1065, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 3.1036446469248293, | |
| "grad_norm": 1.9575121402740479, | |
| "learning_rate": 5.0555555555555555e-06, | |
| "loss": 0.1104, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 3.132118451025057, | |
| "grad_norm": 2.5714454650878906, | |
| "learning_rate": 5e-06, | |
| "loss": 0.1152, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 3.1605922551252847, | |
| "grad_norm": 1.941654920578003, | |
| "learning_rate": 4.944444444444445e-06, | |
| "loss": 0.1156, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 3.1890660592255125, | |
| "grad_norm": 2.314999580383301, | |
| "learning_rate": 4.888888888888889e-06, | |
| "loss": 0.1079, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 3.21753986332574, | |
| "grad_norm": 2.4355452060699463, | |
| "learning_rate": 4.833333333333333e-06, | |
| "loss": 0.1267, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 3.2460136674259683, | |
| "grad_norm": 2.246598720550537, | |
| "learning_rate": 4.777777777777778e-06, | |
| "loss": 0.1036, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 3.274487471526196, | |
| "grad_norm": 1.9095807075500488, | |
| "learning_rate": 4.722222222222222e-06, | |
| "loss": 0.1124, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 3.3029612756264237, | |
| "grad_norm": 2.485973596572876, | |
| "learning_rate": 4.666666666666667e-06, | |
| "loss": 0.115, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 3.3314350797266514, | |
| "grad_norm": 2.0243821144104004, | |
| "learning_rate": 4.611111111111112e-06, | |
| "loss": 0.1073, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 3.359908883826879, | |
| "grad_norm": 2.4946811199188232, | |
| "learning_rate": 4.555555555555556e-06, | |
| "loss": 0.1127, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 3.3883826879271073, | |
| "grad_norm": 2.0500540733337402, | |
| "learning_rate": 4.5e-06, | |
| "loss": 0.1085, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 3.416856492027335, | |
| "grad_norm": 2.5584237575531006, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.1143, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.416856492027335, | |
| "eval_loss": 0.432865709066391, | |
| "eval_runtime": 2970.1299, | |
| "eval_samples_per_second": 2.364, | |
| "eval_steps_per_second": 0.148, | |
| "eval_wer": 0.28141110815838427, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 3.4453302961275627, | |
| "grad_norm": 2.3055500984191895, | |
| "learning_rate": 4.388888888888889e-06, | |
| "loss": 0.1135, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 3.4738041002277904, | |
| "grad_norm": 2.1304893493652344, | |
| "learning_rate": 4.333333333333334e-06, | |
| "loss": 0.1042, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 3.502277904328018, | |
| "grad_norm": 2.5304181575775146, | |
| "learning_rate": 4.277777777777778e-06, | |
| "loss": 0.1156, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 3.5307517084282463, | |
| "grad_norm": 2.2031352519989014, | |
| "learning_rate": 4.222222222222223e-06, | |
| "loss": 0.1104, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 3.559225512528474, | |
| "grad_norm": 2.125488758087158, | |
| "learning_rate": 4.166666666666667e-06, | |
| "loss": 0.1114, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 3.5876993166287017, | |
| "grad_norm": 2.4652950763702393, | |
| "learning_rate": 4.111111111111111e-06, | |
| "loss": 0.1094, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 3.6161731207289294, | |
| "grad_norm": 1.940301775932312, | |
| "learning_rate": 4.055555555555556e-06, | |
| "loss": 0.1127, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 3.644646924829157, | |
| "grad_norm": 2.5598275661468506, | |
| "learning_rate": 4.000000000000001e-06, | |
| "loss": 0.106, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 3.6731207289293852, | |
| "grad_norm": 2.063333511352539, | |
| "learning_rate": 3.944444444444445e-06, | |
| "loss": 0.1077, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 3.7015945330296125, | |
| "grad_norm": 1.9726035594940186, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.117, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 3.7300683371298406, | |
| "grad_norm": 1.7921056747436523, | |
| "learning_rate": 3.833333333333334e-06, | |
| "loss": 0.1088, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 3.7585421412300684, | |
| "grad_norm": 2.258916139602661, | |
| "learning_rate": 3.777777777777778e-06, | |
| "loss": 0.1092, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 3.787015945330296, | |
| "grad_norm": 2.289665937423706, | |
| "learning_rate": 3.7222222222222225e-06, | |
| "loss": 0.108, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 3.8154897494305238, | |
| "grad_norm": 1.943253755569458, | |
| "learning_rate": 3.6666666666666666e-06, | |
| "loss": 0.111, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 3.8439635535307515, | |
| "grad_norm": 2.662332773208618, | |
| "learning_rate": 3.6111111111111115e-06, | |
| "loss": 0.1176, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 3.8724373576309796, | |
| "grad_norm": 2.1651501655578613, | |
| "learning_rate": 3.555555555555556e-06, | |
| "loss": 0.1169, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 3.9009111617312073, | |
| "grad_norm": 2.1495697498321533, | |
| "learning_rate": 3.5e-06, | |
| "loss": 0.1096, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 3.929384965831435, | |
| "grad_norm": 2.1727662086486816, | |
| "learning_rate": 3.444444444444445e-06, | |
| "loss": 0.1119, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 3.9578587699316627, | |
| "grad_norm": 2.4179928302764893, | |
| "learning_rate": 3.3888888888888893e-06, | |
| "loss": 0.1083, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 3.9863325740318905, | |
| "grad_norm": 2.2201690673828125, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.1097, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 4.014806378132119, | |
| "grad_norm": 1.7109599113464355, | |
| "learning_rate": 3.277777777777778e-06, | |
| "loss": 0.0834, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 4.043280182232346, | |
| "grad_norm": 1.5952197313308716, | |
| "learning_rate": 3.2222222222222227e-06, | |
| "loss": 0.0527, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 4.071753986332574, | |
| "grad_norm": 1.780273675918579, | |
| "learning_rate": 3.1666666666666667e-06, | |
| "loss": 0.0509, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 4.100227790432802, | |
| "grad_norm": 1.4646574258804321, | |
| "learning_rate": 3.1111111111111116e-06, | |
| "loss": 0.0524, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 4.128701594533029, | |
| "grad_norm": 1.4881808757781982, | |
| "learning_rate": 3.055555555555556e-06, | |
| "loss": 0.0486, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 4.157175398633258, | |
| "grad_norm": 1.9114530086517334, | |
| "learning_rate": 3e-06, | |
| "loss": 0.0522, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 4.185649202733485, | |
| "grad_norm": 1.587689995765686, | |
| "learning_rate": 2.944444444444445e-06, | |
| "loss": 0.0549, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 4.214123006833713, | |
| "grad_norm": 2.1305880546569824, | |
| "learning_rate": 2.888888888888889e-06, | |
| "loss": 0.0557, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 4.242596810933941, | |
| "grad_norm": 1.6783413887023926, | |
| "learning_rate": 2.8333333333333335e-06, | |
| "loss": 0.05, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 4.271070615034168, | |
| "grad_norm": 2.365713119506836, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0523, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 4.2995444191343966, | |
| "grad_norm": 1.8531174659729004, | |
| "learning_rate": 2.7222222222222224e-06, | |
| "loss": 0.0555, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 4.328018223234624, | |
| "grad_norm": 1.5612976551055908, | |
| "learning_rate": 2.666666666666667e-06, | |
| "loss": 0.0562, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 4.356492027334852, | |
| "grad_norm": 1.8512299060821533, | |
| "learning_rate": 2.6111111111111113e-06, | |
| "loss": 0.0497, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 4.38496583143508, | |
| "grad_norm": 1.2788305282592773, | |
| "learning_rate": 2.5555555555555557e-06, | |
| "loss": 0.0499, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 4.413439635535307, | |
| "grad_norm": 2.3926126956939697, | |
| "learning_rate": 2.5e-06, | |
| "loss": 0.0516, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 4.4419134396355355, | |
| "grad_norm": 1.7549375295639038, | |
| "learning_rate": 2.4444444444444447e-06, | |
| "loss": 0.0527, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 4.470387243735763, | |
| "grad_norm": 1.603882074356079, | |
| "learning_rate": 2.388888888888889e-06, | |
| "loss": 0.0517, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 4.498861047835991, | |
| "grad_norm": 1.872842788696289, | |
| "learning_rate": 2.3333333333333336e-06, | |
| "loss": 0.047, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 4.527334851936219, | |
| "grad_norm": 1.3543041944503784, | |
| "learning_rate": 2.277777777777778e-06, | |
| "loss": 0.0519, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 4.555808656036446, | |
| "grad_norm": 1.6009862422943115, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0515, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.555808656036446, | |
| "eval_loss": 0.4906230866909027, | |
| "eval_runtime": 2972.4828, | |
| "eval_samples_per_second": 2.362, | |
| "eval_steps_per_second": 0.148, | |
| "eval_wer": 0.28324475152803613, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 4.5842824601366745, | |
| "grad_norm": 2.2875330448150635, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0494, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 4.612756264236902, | |
| "grad_norm": 1.5816611051559448, | |
| "learning_rate": 2.1111111111111114e-06, | |
| "loss": 0.0482, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 4.64123006833713, | |
| "grad_norm": 2.0431058406829834, | |
| "learning_rate": 2.0555555555555555e-06, | |
| "loss": 0.0591, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 4.669703872437358, | |
| "grad_norm": 1.7807472944259644, | |
| "learning_rate": 2.0000000000000003e-06, | |
| "loss": 0.048, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 4.698177676537585, | |
| "grad_norm": 1.7173796892166138, | |
| "learning_rate": 1.944444444444445e-06, | |
| "loss": 0.0506, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 4.7266514806378135, | |
| "grad_norm": 1.8625760078430176, | |
| "learning_rate": 1.888888888888889e-06, | |
| "loss": 0.0493, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 4.755125284738041, | |
| "grad_norm": 1.7857956886291504, | |
| "learning_rate": 1.8333333333333333e-06, | |
| "loss": 0.0507, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 4.783599088838269, | |
| "grad_norm": 1.458436369895935, | |
| "learning_rate": 1.777777777777778e-06, | |
| "loss": 0.0517, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 4.812072892938497, | |
| "grad_norm": 2.7380471229553223, | |
| "learning_rate": 1.7222222222222224e-06, | |
| "loss": 0.0491, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 4.840546697038724, | |
| "grad_norm": 2.323181390762329, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.0513, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 4.8690205011389525, | |
| "grad_norm": 2.195849895477295, | |
| "learning_rate": 1.6111111111111113e-06, | |
| "loss": 0.045, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 4.89749430523918, | |
| "grad_norm": 1.933990478515625, | |
| "learning_rate": 1.5555555555555558e-06, | |
| "loss": 0.0448, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 4.925968109339408, | |
| "grad_norm": 1.2482439279556274, | |
| "learning_rate": 1.5e-06, | |
| "loss": 0.0471, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 4.954441913439636, | |
| "grad_norm": 1.8933014869689941, | |
| "learning_rate": 1.4444444444444445e-06, | |
| "loss": 0.0486, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 4.982915717539863, | |
| "grad_norm": 1.77507746219635, | |
| "learning_rate": 1.3888888888888892e-06, | |
| "loss": 0.0489, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 5.011389521640091, | |
| "grad_norm": 1.1538957357406616, | |
| "learning_rate": 1.3333333333333334e-06, | |
| "loss": 0.0393, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 5.039863325740319, | |
| "grad_norm": 0.9963041543960571, | |
| "learning_rate": 1.2777777777777779e-06, | |
| "loss": 0.0226, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 5.068337129840547, | |
| "grad_norm": 0.8649413585662842, | |
| "learning_rate": 1.2222222222222223e-06, | |
| "loss": 0.0203, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 5.096810933940774, | |
| "grad_norm": 1.663192629814148, | |
| "learning_rate": 1.1666666666666668e-06, | |
| "loss": 0.0209, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 5.125284738041002, | |
| "grad_norm": 2.2397210597991943, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0199, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 5.15375854214123, | |
| "grad_norm": 1.0296459197998047, | |
| "learning_rate": 1.0555555555555557e-06, | |
| "loss": 0.0204, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 5.182232346241458, | |
| "grad_norm": 1.3559978008270264, | |
| "learning_rate": 1.0000000000000002e-06, | |
| "loss": 0.0195, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 5.210706150341686, | |
| "grad_norm": 1.0463148355484009, | |
| "learning_rate": 9.444444444444445e-07, | |
| "loss": 0.0243, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 5.239179954441913, | |
| "grad_norm": 1.8995018005371094, | |
| "learning_rate": 8.88888888888889e-07, | |
| "loss": 0.0204, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 5.267653758542141, | |
| "grad_norm": 0.8292195796966553, | |
| "learning_rate": 8.333333333333333e-07, | |
| "loss": 0.0185, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 5.296127562642369, | |
| "grad_norm": 1.4848933219909668, | |
| "learning_rate": 7.777777777777779e-07, | |
| "loss": 0.0206, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 5.324601366742597, | |
| "grad_norm": 0.9767656922340393, | |
| "learning_rate": 7.222222222222222e-07, | |
| "loss": 0.0202, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 5.353075170842825, | |
| "grad_norm": 0.8986805081367493, | |
| "learning_rate": 6.666666666666667e-07, | |
| "loss": 0.0185, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 5.381548974943052, | |
| "grad_norm": 1.233708143234253, | |
| "learning_rate": 6.111111111111112e-07, | |
| "loss": 0.0178, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 5.41002277904328, | |
| "grad_norm": 1.045877456665039, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0188, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 5.438496583143508, | |
| "grad_norm": 1.4477061033248901, | |
| "learning_rate": 5.000000000000001e-07, | |
| "loss": 0.0217, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 5.466970387243736, | |
| "grad_norm": 1.7844326496124268, | |
| "learning_rate": 4.444444444444445e-07, | |
| "loss": 0.0186, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 5.495444191343964, | |
| "grad_norm": 1.0752902030944824, | |
| "learning_rate": 3.8888888888888895e-07, | |
| "loss": 0.0197, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 5.523917995444191, | |
| "grad_norm": 1.193974256515503, | |
| "learning_rate": 3.3333333333333335e-07, | |
| "loss": 0.0201, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 5.552391799544419, | |
| "grad_norm": 0.7838534116744995, | |
| "learning_rate": 2.7777777777777776e-07, | |
| "loss": 0.0202, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 5.5808656036446465, | |
| "grad_norm": 0.9036350250244141, | |
| "learning_rate": 2.2222222222222224e-07, | |
| "loss": 0.0199, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 5.609339407744875, | |
| "grad_norm": 0.8704236745834351, | |
| "learning_rate": 1.6666666666666668e-07, | |
| "loss": 0.0198, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 5.637813211845103, | |
| "grad_norm": 1.007270336151123, | |
| "learning_rate": 1.1111111111111112e-07, | |
| "loss": 0.0222, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 5.66628701594533, | |
| "grad_norm": 0.9725838899612427, | |
| "learning_rate": 5.555555555555556e-08, | |
| "loss": 0.0196, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 5.694760820045558, | |
| "grad_norm": 0.970182478427887, | |
| "learning_rate": 0.0, | |
| "loss": 0.0193, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 5.694760820045558, | |
| "eval_loss": 0.5499717593193054, | |
| "eval_runtime": 2981.2655, | |
| "eval_samples_per_second": 2.355, | |
| "eval_steps_per_second": 0.147, | |
| "eval_wer": 0.2817565771990433, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 5.694760820045558, | |
| "step": 5000, | |
| "total_flos": 5.43429854134272e+20, | |
| "train_loss": 0.23574846556782722, | |
| "train_runtime": 65192.9802, | |
| "train_samples_per_second": 2.454, | |
| "train_steps_per_second": 0.077 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 5000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 6, | |
| "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": 5.43429854134272e+20, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |