| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.565962534874451, | |
| "global_step": 54000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004991142604074402, | |
| "loss": 2.333, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004982285208148805, | |
| "loss": 2.3899, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0004973427812223207, | |
| "loss": 2.4422, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0004964570416297609, | |
| "loss": 2.4359, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0004955713020372011, | |
| "loss": 2.3725, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0004946855624446413, | |
| "loss": 2.4478, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.0004937998228520815, | |
| "loss": 2.4036, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0004929140832595217, | |
| "loss": 2.3984, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.0004920283436669619, | |
| "loss": 2.3836, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.0004911426040744021, | |
| "loss": 2.3708, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0004902568644818423, | |
| "loss": 2.3009, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.0004893711248892826, | |
| "loss": 2.367, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0004884853852967228, | |
| "loss": 2.3713, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.000487599645704163, | |
| "loss": 2.3609, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00048671390611160323, | |
| "loss": 2.3181, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00048582816651904345, | |
| "loss": 2.3249, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0004849424269264836, | |
| "loss": 2.3505, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00048405668733392384, | |
| "loss": 2.2962, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00048317094774136406, | |
| "loss": 2.3446, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.0004822852081488043, | |
| "loss": 2.308, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00048139946855624445, | |
| "loss": 2.2919, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.0004805137289636847, | |
| "loss": 2.308, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0004796279893711249, | |
| "loss": 2.2863, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0004787422497785651, | |
| "loss": 2.2976, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00047785651018600534, | |
| "loss": 2.288, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.0004769707705934455, | |
| "loss": 2.2771, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00047608503100088573, | |
| "loss": 2.2608, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00047519929140832595, | |
| "loss": 2.2227, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0004743135518157662, | |
| "loss": 2.2823, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0004734278122232064, | |
| "loss": 2.1959, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00047254207263064656, | |
| "loss": 2.2388, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0004716563330380868, | |
| "loss": 2.252, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.000470770593445527, | |
| "loss": 2.2295, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00046988485385296723, | |
| "loss": 2.2035, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0004689991142604074, | |
| "loss": 2.285, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00046811337466784767, | |
| "loss": 2.2439, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.0004672276350752879, | |
| "loss": 2.2378, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0004663418954827281, | |
| "loss": 2.2499, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00046545615589016834, | |
| "loss": 2.1999, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0004645704162976085, | |
| "loss": 2.1862, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00046368467670504873, | |
| "loss": 2.217, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00046279893711248895, | |
| "loss": 2.2083, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00046191319751992917, | |
| "loss": 2.1784, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.0004610274579273694, | |
| "loss": 2.1467, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00046014171833480956, | |
| "loss": 2.2254, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.0004592559787422498, | |
| "loss": 2.2067, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00045837023914969, | |
| "loss": 2.2011, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.0004574844995571302, | |
| "loss": 2.1521, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.0004565987599645704, | |
| "loss": 2.193, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.0004557130203720106, | |
| "loss": 2.1709, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.00045482728077945084, | |
| "loss": 2.1806, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.00045394154118689106, | |
| "loss": 2.1895, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0004530558015943313, | |
| "loss": 2.203, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00045217006200177145, | |
| "loss": 2.1686, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00045128432240921167, | |
| "loss": 2.1461, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.0004503985828166519, | |
| "loss": 2.1239, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.0004495128432240921, | |
| "loss": 2.0857, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.00044862710363153234, | |
| "loss": 2.0997, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0004477413640389725, | |
| "loss": 2.0756, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.0004468556244464128, | |
| "loss": 2.0468, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.000445969884853853, | |
| "loss": 2.0412, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0004450841452612932, | |
| "loss": 2.0289, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0004441984056687334, | |
| "loss": 2.07, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.0004433126660761736, | |
| "loss": 2.039, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00044242692648361384, | |
| "loss": 2.0749, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00044154118689105406, | |
| "loss": 2.0776, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.0004406554472984943, | |
| "loss": 2.0375, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00043976970770593445, | |
| "loss": 2.0512, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00043888396811337467, | |
| "loss": 2.0449, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.0004379982285208149, | |
| "loss": 2.0157, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0004371124889282551, | |
| "loss": 2.0406, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00043622674933569533, | |
| "loss": 2.039, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0004353410097431355, | |
| "loss": 2.0525, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.0004344552701505757, | |
| "loss": 2.0217, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.00043356953055801595, | |
| "loss": 2.0398, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00043268379096545617, | |
| "loss": 2.0426, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00043179805137289634, | |
| "loss": 2.0565, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00043091231178033656, | |
| "loss": 1.9831, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0004300265721877768, | |
| "loss": 2.0202, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.000429140832595217, | |
| "loss": 2.0607, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.0004282550930026572, | |
| "loss": 2.0394, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.0004273693534100974, | |
| "loss": 2.0303, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00042648361381753767, | |
| "loss": 2.0451, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.0004255978742249779, | |
| "loss": 2.0185, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.0004247121346324181, | |
| "loss": 2.0064, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00042382639503985833, | |
| "loss": 2.0093, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.0004229406554472985, | |
| "loss": 1.9924, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0004220549158547387, | |
| "loss": 2.0301, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.00042116917626217894, | |
| "loss": 1.9849, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00042028343666961917, | |
| "loss": 2.0016, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.00041939769707705933, | |
| "loss": 1.9805, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.00041851195748449956, | |
| "loss": 2.0216, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0004176262178919398, | |
| "loss": 1.9921, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00041674047829938, | |
| "loss": 1.9857, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0004158547387068202, | |
| "loss": 2.011, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.0004149689991142604, | |
| "loss": 2.0079, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0004140832595217006, | |
| "loss": 2.0178, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00041319751992914083, | |
| "loss": 1.9978, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00041231178033658105, | |
| "loss": 1.9585, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.0004114260407440213, | |
| "loss": 1.9802, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00041054030115146144, | |
| "loss": 1.9912, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00040965456155890167, | |
| "loss": 2.0093, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0004087688219663419, | |
| "loss": 1.9695, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0004078830823737821, | |
| "loss": 2.0003, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.0004069973427812223, | |
| "loss": 2.0081, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.0004061116031886625, | |
| "loss": 1.9888, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0004052258635961028, | |
| "loss": 2.0072, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.000404340124003543, | |
| "loss": 2.0097, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.0004034543844109832, | |
| "loss": 2.0066, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.0004025686448184234, | |
| "loss": 1.9976, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0004016829052258636, | |
| "loss": 1.9543, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00040079716563330383, | |
| "loss": 1.9676, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00039991142604074405, | |
| "loss": 1.9547, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0003990256864481843, | |
| "loss": 1.8447, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00039813994685562444, | |
| "loss": 1.9108, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.00039725420726306466, | |
| "loss": 1.8746, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.0003963684676705049, | |
| "loss": 1.864, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.0003954827280779451, | |
| "loss": 1.8742, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.0003945969884853853, | |
| "loss": 1.8111, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.0003937112488928255, | |
| "loss": 1.8708, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.0003928255093002657, | |
| "loss": 1.8909, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.00039193976970770594, | |
| "loss": 1.8941, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00039105403011514616, | |
| "loss": 1.8875, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.00039016829052258633, | |
| "loss": 1.8948, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.00038928255093002655, | |
| "loss": 1.8865, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.0003883968113374668, | |
| "loss": 1.8871, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.000387511071744907, | |
| "loss": 1.8811, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.0003866253321523472, | |
| "loss": 1.8876, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.0003857395925597874, | |
| "loss": 1.8702, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.00038485385296722766, | |
| "loss": 1.9055, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.0003839681133746679, | |
| "loss": 1.8816, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.0003830823737821081, | |
| "loss": 1.8652, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.00038219663418954827, | |
| "loss": 1.8891, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.0003813108945969885, | |
| "loss": 1.9039, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 0.0003804251550044287, | |
| "loss": 1.8558, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.00037953941541186894, | |
| "loss": 1.8471, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 0.00037865367581930916, | |
| "loss": 1.8532, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00037776793622674933, | |
| "loss": 1.8786, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.00037688219663418955, | |
| "loss": 1.8285, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.00037599645704162977, | |
| "loss": 1.8505, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.00037511071744907, | |
| "loss": 1.8604, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.0003742249778565102, | |
| "loss": 1.8684, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.0003733392382639504, | |
| "loss": 1.8241, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.0003724534986713906, | |
| "loss": 1.8601, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.0003715677590788308, | |
| "loss": 1.8582, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.00037068201948627105, | |
| "loss": 1.8555, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.0003697962798937112, | |
| "loss": 1.8905, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.00036891054030115144, | |
| "loss": 1.8754, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.00036802480070859166, | |
| "loss": 1.8835, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0003671390611160319, | |
| "loss": 1.847, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.0003662533215234721, | |
| "loss": 1.8621, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 0.00036536758193091227, | |
| "loss": 1.8558, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.0003644818423383525, | |
| "loss": 1.8349, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.00036359610274579277, | |
| "loss": 1.8698, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.000362710363153233, | |
| "loss": 1.8925, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.0003618246235606732, | |
| "loss": 1.8347, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.0003609388839681134, | |
| "loss": 1.869, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.0003600531443755536, | |
| "loss": 1.82, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.0003591674047829938, | |
| "loss": 1.8407, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.00035828166519043405, | |
| "loss": 1.8458, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0003573959255978742, | |
| "loss": 1.8568, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.00035651018600531444, | |
| "loss": 1.855, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.00035562444641275466, | |
| "loss": 1.841, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.0003547387068201949, | |
| "loss": 1.814, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.0003538529672276351, | |
| "loss": 1.8102, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.00035296722763507527, | |
| "loss": 1.8517, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.0003520814880425155, | |
| "loss": 1.8326, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0003511957484499557, | |
| "loss": 1.8514, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.00035031000885739594, | |
| "loss": 1.863, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 0.00034942426926483616, | |
| "loss": 1.779, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 0.0003485385296722763, | |
| "loss": 1.7412, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 0.00034765279007971655, | |
| "loss": 1.7293, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 0.00034676705048715677, | |
| "loss": 1.7102, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 0.000345881310894597, | |
| "loss": 1.7713, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 0.00034499557130203716, | |
| "loss": 1.7884, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 0.0003441098317094774, | |
| "loss": 1.7567, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 0.00034322409211691766, | |
| "loss": 1.7192, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 0.0003423383525243579, | |
| "loss": 1.7205, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 0.0003414526129317981, | |
| "loss": 1.7175, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00034056687333923827, | |
| "loss": 1.7627, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 0.0003396811337466785, | |
| "loss": 1.7404, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.0003387953941541187, | |
| "loss": 1.7664, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 0.00033790965456155893, | |
| "loss": 1.7307, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 0.00033702391496899915, | |
| "loss": 1.7245, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 0.0003361381753764393, | |
| "loss": 1.7643, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 0.00033525243578387954, | |
| "loss": 1.7469, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 0.00033436669619131977, | |
| "loss": 1.7639, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.00033348095659876, | |
| "loss": 1.7364, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 0.00033259521700620016, | |
| "loss": 1.7284, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.0003317094774136404, | |
| "loss": 1.7461, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 0.0003308237378210806, | |
| "loss": 1.7558, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 0.0003299379982285208, | |
| "loss": 1.7578, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 0.00032905225863596104, | |
| "loss": 1.7989, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 0.0003281665190434012, | |
| "loss": 1.7949, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 0.00032728077945084143, | |
| "loss": 1.777, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 0.00032639503985828165, | |
| "loss": 1.777, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 0.0003255093002657219, | |
| "loss": 1.7626, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 0.0003246235606731621, | |
| "loss": 1.7843, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 0.00032373782108060227, | |
| "loss": 1.7301, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 0.0003228520814880425, | |
| "loss": 1.7561, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 0.00032196634189548276, | |
| "loss": 1.7534, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 0.000321080602302923, | |
| "loss": 1.753, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 0.00032019486271036315, | |
| "loss": 1.7259, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 0.0003193091231178034, | |
| "loss": 1.789, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 0.0003184233835252436, | |
| "loss": 1.7771, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 0.0003175376439326838, | |
| "loss": 1.7308, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.00031665190434012404, | |
| "loss": 1.7568, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 0.0003157661647475642, | |
| "loss": 1.7393, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 0.00031488042515500443, | |
| "loss": 1.7593, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 0.00031399468556244465, | |
| "loss": 1.732, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 0.0003131089459698849, | |
| "loss": 1.7342, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 0.0003122232063773251, | |
| "loss": 1.7386, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00031133746678476526, | |
| "loss": 1.7394, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 0.0003104517271922055, | |
| "loss": 1.7439, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.0003095659875996457, | |
| "loss": 1.7053, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 0.00030868024800708593, | |
| "loss": 1.7022, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 0.0003077945084145261, | |
| "loss": 1.7254, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 0.0003069087688219663, | |
| "loss": 1.712, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 0.00030602302922940654, | |
| "loss": 1.7565, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 0.00030513728963684676, | |
| "loss": 1.7495, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.000304251550044287, | |
| "loss": 1.7435, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00030336581045172715, | |
| "loss": 1.7649, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.0003024800708591674, | |
| "loss": 1.743, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 0.00030159433126660765, | |
| "loss": 1.7332, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 0.00030070859167404787, | |
| "loss": 1.6918, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.0002998228520814881, | |
| "loss": 1.6936, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 0.00029893711248892826, | |
| "loss": 1.6353, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 0.0002980513728963685, | |
| "loss": 1.6311, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 0.0002971656333038087, | |
| "loss": 1.6385, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 0.00029627989371124893, | |
| "loss": 1.6188, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.0002953941541186891, | |
| "loss": 1.6043, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 0.0002945084145261293, | |
| "loss": 1.6233, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 0.00029362267493356954, | |
| "loss": 1.6249, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.00029273693534100976, | |
| "loss": 1.6264, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 0.00029185119574845, | |
| "loss": 1.6343, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 0.00029096545615589015, | |
| "loss": 1.6157, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.00029007971656333037, | |
| "loss": 1.6384, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 0.0002891939769707706, | |
| "loss": 1.65, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 0.0002883082373782108, | |
| "loss": 1.6624, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.00028742249778565104, | |
| "loss": 1.6375, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 0.0002865367581930912, | |
| "loss": 1.6722, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 0.00028565101860053143, | |
| "loss": 1.6491, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.00028476527900797165, | |
| "loss": 1.6498, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 0.00028387953941541187, | |
| "loss": 1.6559, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 0.00028299379982285204, | |
| "loss": 1.6044, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 0.00028210806023029226, | |
| "loss": 1.6661, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 0.0002812223206377325, | |
| "loss": 1.6478, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 0.00028033658104517276, | |
| "loss": 1.6382, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 0.000279450841452613, | |
| "loss": 1.6339, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 0.00027856510186005315, | |
| "loss": 1.6223, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 0.00027767936226749337, | |
| "loss": 1.6472, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 0.0002767936226749336, | |
| "loss": 1.6333, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 0.0002759078830823738, | |
| "loss": 1.6384, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 0.00027502214348981404, | |
| "loss": 1.6005, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 0.0002741364038972542, | |
| "loss": 1.6455, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 0.0002732506643046944, | |
| "loss": 1.6201, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 0.00027236492471213465, | |
| "loss": 1.6439, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 0.00027147918511957487, | |
| "loss": 1.6261, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.00027059344552701504, | |
| "loss": 1.6236, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 0.00026970770593445526, | |
| "loss": 1.6156, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 0.0002688219663418955, | |
| "loss": 1.6597, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.0002679362267493357, | |
| "loss": 1.635, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 0.0002670504871567759, | |
| "loss": 1.6202, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 0.0002661647475642161, | |
| "loss": 1.6413, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 0.0002652790079716563, | |
| "loss": 1.6363, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 0.00026439326837909654, | |
| "loss": 1.6283, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 0.00026350752878653676, | |
| "loss": 1.6437, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 0.000262621789193977, | |
| "loss": 1.6456, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 0.00026173604960141715, | |
| "loss": 1.6076, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.00026085031000885737, | |
| "loss": 1.6429, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 0.00025996457041629764, | |
| "loss": 1.6635, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 0.00025907883082373787, | |
| "loss": 1.6269, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 0.00025819309123117803, | |
| "loss": 1.6282, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 0.00025730735163861826, | |
| "loss": 1.5984, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 0.0002564216120460585, | |
| "loss": 1.6434, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 0.0002555358724534987, | |
| "loss": 1.641, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 0.0002546501328609389, | |
| "loss": 1.6232, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 0.0002537643932683791, | |
| "loss": 1.5679, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 0.0002528786536758193, | |
| "loss": 1.6289, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 0.00025199291408325953, | |
| "loss": 1.6466, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 0.00025110717449069975, | |
| "loss": 1.6467, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.00025022143489814, | |
| "loss": 1.6348, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 5.01, | |
| "learning_rate": 0.0002493356953055802, | |
| "loss": 1.58, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 5.03, | |
| "learning_rate": 0.00024844995571302037, | |
| "loss": 1.5345, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 0.0002475642161204606, | |
| "loss": 1.5501, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 5.07, | |
| "learning_rate": 0.0002466784765279008, | |
| "loss": 1.5667, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 5.08, | |
| "learning_rate": 0.00024579273693534103, | |
| "loss": 1.5045, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 0.00024490699734278125, | |
| "loss": 1.5721, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 5.12, | |
| "learning_rate": 0.00024402125775022145, | |
| "loss": 1.5339, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 0.00024313551815766167, | |
| "loss": 1.5543, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 5.15, | |
| "learning_rate": 0.00024224977856510187, | |
| "loss": 1.5568, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 0.0002413640389725421, | |
| "loss": 1.5496, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 5.19, | |
| "learning_rate": 0.00024047829937998228, | |
| "loss": 1.5492, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 5.21, | |
| "learning_rate": 0.0002395925597874225, | |
| "loss": 1.571, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 0.00023870682019486273, | |
| "loss": 1.5495, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 0.00023782108060230292, | |
| "loss": 1.538, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 5.26, | |
| "learning_rate": 0.00023693534100974314, | |
| "loss": 1.5464, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 5.28, | |
| "learning_rate": 0.00023604960141718334, | |
| "loss": 1.5523, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 0.00023516386182462356, | |
| "loss": 1.526, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 0.00023427812223206375, | |
| "loss": 1.5692, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 5.33, | |
| "learning_rate": 0.000233392382639504, | |
| "loss": 1.548, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 5.35, | |
| "learning_rate": 0.00023250664304694422, | |
| "loss": 1.575, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 5.37, | |
| "learning_rate": 0.00023162090345438442, | |
| "loss": 1.5104, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 5.39, | |
| "learning_rate": 0.00023073516386182464, | |
| "loss": 1.558, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 0.00022984942426926484, | |
| "loss": 1.5481, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 5.42, | |
| "learning_rate": 0.00022896368467670506, | |
| "loss": 1.5538, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 5.44, | |
| "learning_rate": 0.00022807794508414525, | |
| "loss": 1.4957, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 0.00022719220549158547, | |
| "loss": 1.5547, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 0.0002263064658990257, | |
| "loss": 1.573, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 5.49, | |
| "learning_rate": 0.0002254207263064659, | |
| "loss": 1.5508, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 5.51, | |
| "learning_rate": 0.0002245349867139061, | |
| "loss": 1.5452, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 0.00022364924712134633, | |
| "loss": 1.5031, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 0.00022276350752878656, | |
| "loss": 1.5238, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 5.56, | |
| "learning_rate": 0.00022187776793622675, | |
| "loss": 1.5511, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 5.58, | |
| "learning_rate": 0.00022099202834366697, | |
| "loss": 1.5784, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 0.0002201062887511072, | |
| "loss": 1.5487, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 0.0002192205491585474, | |
| "loss": 1.5496, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 0.0002183348095659876, | |
| "loss": 1.5414, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 5.65, | |
| "learning_rate": 0.0002174490699734278, | |
| "loss": 1.5677, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 0.00021656333038086803, | |
| "loss": 1.5408, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 5.69, | |
| "learning_rate": 0.00021567759078830822, | |
| "loss": 1.5122, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 0.00021479185119574845, | |
| "loss": 1.5275, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 0.00021390611160318867, | |
| "loss": 1.5314, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 0.0002130203720106289, | |
| "loss": 1.5405, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 5.76, | |
| "learning_rate": 0.0002121346324180691, | |
| "loss": 1.5039, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 0.0002112488928255093, | |
| "loss": 1.5173, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 0.00021036315323294953, | |
| "loss": 1.564, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 0.00020947741364038972, | |
| "loss": 1.5666, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 5.83, | |
| "learning_rate": 0.00020859167404782994, | |
| "loss": 1.5539, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 0.00020770593445527017, | |
| "loss": 1.5436, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 0.00020682019486271036, | |
| "loss": 1.5431, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 0.00020593445527015058, | |
| "loss": 1.5364, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 0.00020504871567759078, | |
| "loss": 1.5186, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 0.000204162976085031, | |
| "loss": 1.5308, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 5.93, | |
| "learning_rate": 0.0002032772364924712, | |
| "loss": 1.5218, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 0.00020239149689991144, | |
| "loss": 1.547, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 5.97, | |
| "learning_rate": 0.00020150575730735166, | |
| "loss": 1.554, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 0.00020062001771479186, | |
| "loss": 1.5563, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 6.01, | |
| "learning_rate": 0.00019973427812223208, | |
| "loss": 1.5311, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 6.02, | |
| "learning_rate": 0.00019884853852967228, | |
| "loss": 1.4537, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 0.0001979627989371125, | |
| "loss": 1.4493, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 6.06, | |
| "learning_rate": 0.0001970770593445527, | |
| "loss": 1.462, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 0.00019619131975199291, | |
| "loss": 1.4879, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 6.09, | |
| "learning_rate": 0.00019530558015943314, | |
| "loss": 1.46, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 0.00019441984056687333, | |
| "loss": 1.4318, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 0.00019353410097431355, | |
| "loss": 1.448, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 0.00019264836138175375, | |
| "loss": 1.4615, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 6.16, | |
| "learning_rate": 0.000191762621789194, | |
| "loss": 1.4537, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 6.18, | |
| "learning_rate": 0.0001908768821966342, | |
| "loss": 1.4697, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 6.2, | |
| "learning_rate": 0.00018999114260407441, | |
| "loss": 1.4502, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 0.00018910540301151464, | |
| "loss": 1.4421, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 0.00018821966341895483, | |
| "loss": 1.456, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 6.25, | |
| "learning_rate": 0.00018733392382639505, | |
| "loss": 1.4647, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 0.00018644818423383525, | |
| "loss": 1.4854, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 6.29, | |
| "learning_rate": 0.00018556244464127547, | |
| "loss": 1.4589, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 0.00018467670504871566, | |
| "loss": 1.4537, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 0.00018379096545615589, | |
| "loss": 1.4602, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 6.34, | |
| "learning_rate": 0.0001829052258635961, | |
| "loss": 1.487, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 6.36, | |
| "learning_rate": 0.00018201948627103633, | |
| "loss": 1.4731, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 6.38, | |
| "learning_rate": 0.00018113374667847655, | |
| "loss": 1.4487, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 0.00018024800708591675, | |
| "loss": 1.4505, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 6.41, | |
| "learning_rate": 0.00017936226749335697, | |
| "loss": 1.4525, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 0.00017847652790079716, | |
| "loss": 1.4717, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 0.00017759078830823738, | |
| "loss": 1.4803, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 6.47, | |
| "learning_rate": 0.0001767050487156776, | |
| "loss": 1.4443, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 6.48, | |
| "learning_rate": 0.0001758193091231178, | |
| "loss": 1.4659, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 6.5, | |
| "learning_rate": 0.00017493356953055802, | |
| "loss": 1.4918, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.00017404782993799822, | |
| "loss": 1.4444, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 6.54, | |
| "learning_rate": 0.00017316209034543844, | |
| "loss": 1.4804, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 6.55, | |
| "learning_rate": 0.00017227635075287863, | |
| "loss": 1.4815, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 6.57, | |
| "learning_rate": 0.00017139061116031888, | |
| "loss": 1.4791, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 0.0001705048715677591, | |
| "loss": 1.4677, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 6.61, | |
| "learning_rate": 0.0001696191319751993, | |
| "loss": 1.4453, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 6.63, | |
| "learning_rate": 0.00016873339238263952, | |
| "loss": 1.4632, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 6.64, | |
| "learning_rate": 0.00016784765279007972, | |
| "loss": 1.4676, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 6.66, | |
| "learning_rate": 0.00016696191319751994, | |
| "loss": 1.4879, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 6.68, | |
| "learning_rate": 0.00016607617360496013, | |
| "loss": 1.4755, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 6.7, | |
| "learning_rate": 0.00016519043401240036, | |
| "loss": 1.4617, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 0.00016430469441984058, | |
| "loss": 1.4597, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 6.73, | |
| "learning_rate": 0.00016341895482728077, | |
| "loss": 1.4442, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 0.000162533215234721, | |
| "loss": 1.4734, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 6.77, | |
| "learning_rate": 0.0001616474756421612, | |
| "loss": 1.471, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 6.78, | |
| "learning_rate": 0.00016076173604960144, | |
| "loss": 1.4606, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 0.00015987599645704163, | |
| "loss": 1.4626, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 0.00015899025686448185, | |
| "loss": 1.4648, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 6.84, | |
| "learning_rate": 0.00015810451727192208, | |
| "loss": 1.4665, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 6.86, | |
| "learning_rate": 0.00015721877767936227, | |
| "loss": 1.4483, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 6.87, | |
| "learning_rate": 0.0001563330380868025, | |
| "loss": 1.4638, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 0.0001554472984942427, | |
| "loss": 1.4695, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 0.0001545615589016829, | |
| "loss": 1.4487, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 6.93, | |
| "learning_rate": 0.0001536758193091231, | |
| "loss": 1.4547, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 0.00015279007971656333, | |
| "loss": 1.4757, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 0.00015190434012400355, | |
| "loss": 1.4709, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 0.00015101860053144374, | |
| "loss": 1.438, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 0.000150132860938884, | |
| "loss": 1.4504, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 7.02, | |
| "learning_rate": 0.00014924712134632419, | |
| "loss": 1.4182, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 0.0001483613817537644, | |
| "loss": 1.3538, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 0.0001474756421612046, | |
| "loss": 1.3627, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 0.00014658990256864482, | |
| "loss": 1.3909, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 7.09, | |
| "learning_rate": 0.00014570416297608505, | |
| "loss": 1.4015, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 0.00014481842338352524, | |
| "loss": 1.4001, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 0.00014393268379096546, | |
| "loss": 1.3875, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 7.14, | |
| "learning_rate": 0.00014304694419840566, | |
| "loss": 1.4087, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 0.00014216120460584588, | |
| "loss": 1.4053, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 7.17, | |
| "learning_rate": 0.00014127546501328607, | |
| "loss": 1.3912, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 7.19, | |
| "learning_rate": 0.00014038972542072632, | |
| "loss": 1.4045, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 0.00013950398582816655, | |
| "loss": 1.3781, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 0.00013861824623560674, | |
| "loss": 1.4017, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 0.00013773250664304696, | |
| "loss": 1.3905, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 7.26, | |
| "learning_rate": 0.00013684676705048716, | |
| "loss": 1.4088, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 7.28, | |
| "learning_rate": 0.00013596102745792738, | |
| "loss": 1.427, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 0.00013507528786536757, | |
| "loss": 1.3799, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 0.0001341895482728078, | |
| "loss": 1.3973, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 7.33, | |
| "learning_rate": 0.00013330380868024802, | |
| "loss": 1.4444, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 0.0001324180690876882, | |
| "loss": 1.3977, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 7.37, | |
| "learning_rate": 0.00013153232949512843, | |
| "loss": 1.413, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 0.00013064658990256863, | |
| "loss": 1.3855, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 7.4, | |
| "learning_rate": 0.00012976085031000888, | |
| "loss": 1.3802, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 0.00012887511071744907, | |
| "loss": 1.3871, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 7.44, | |
| "learning_rate": 0.0001279893711248893, | |
| "loss": 1.3736, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 0.00012710363153232952, | |
| "loss": 1.4049, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 0.0001262178919397697, | |
| "loss": 1.3547, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 7.49, | |
| "learning_rate": 0.00012533215234720993, | |
| "loss": 1.3899, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 7.51, | |
| "learning_rate": 0.00012444641275465013, | |
| "loss": 1.3879, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 7.53, | |
| "learning_rate": 0.00012356067316209035, | |
| "loss": 1.386, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 0.00012267493356953057, | |
| "loss": 1.3889, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 7.56, | |
| "learning_rate": 0.00012178919397697078, | |
| "loss": 1.4047, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 7.58, | |
| "learning_rate": 0.00012090345438441099, | |
| "loss": 1.3835, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 7.6, | |
| "learning_rate": 0.0001200177147918512, | |
| "loss": 1.3969, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 7.62, | |
| "learning_rate": 0.0001191319751992914, | |
| "loss": 1.4, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 7.64, | |
| "learning_rate": 0.00011824623560673161, | |
| "loss": 1.3735, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 0.00011736049601417183, | |
| "loss": 1.3813, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 7.67, | |
| "learning_rate": 0.00011647475642161206, | |
| "loss": 1.3986, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 7.69, | |
| "learning_rate": 0.00011558901682905227, | |
| "loss": 1.3971, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 0.00011470327723649247, | |
| "loss": 1.4061, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 0.00011381753764393268, | |
| "loss": 1.3679, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 7.74, | |
| "learning_rate": 0.00011293179805137289, | |
| "loss": 1.3995, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 7.76, | |
| "learning_rate": 0.0001120460584588131, | |
| "loss": 1.3876, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 7.78, | |
| "learning_rate": 0.00011116031886625333, | |
| "loss": 1.3797, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 0.00011027457927369354, | |
| "loss": 1.3738, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 7.81, | |
| "learning_rate": 0.00010938883968113375, | |
| "loss": 1.3694, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 7.83, | |
| "learning_rate": 0.00010850310008857396, | |
| "loss": 1.4214, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 7.85, | |
| "learning_rate": 0.00010761736049601417, | |
| "loss": 1.4109, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 0.00010673162090345438, | |
| "loss": 1.3824, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 0.0001058458813108946, | |
| "loss": 1.399, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 7.9, | |
| "learning_rate": 0.00010496014171833482, | |
| "loss": 1.3917, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 7.92, | |
| "learning_rate": 0.00010407440212577503, | |
| "loss": 1.422, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 7.94, | |
| "learning_rate": 0.00010318866253321524, | |
| "loss": 1.3856, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 0.00010230292294065544, | |
| "loss": 1.3747, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 0.00010141718334809567, | |
| "loss": 1.3847, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 7.99, | |
| "learning_rate": 0.00010053144375553587, | |
| "loss": 1.3937, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 8.01, | |
| "learning_rate": 9.964570416297608e-05, | |
| "loss": 1.3614, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 9.87599645704163e-05, | |
| "loss": 1.3446, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 9.787422497785651e-05, | |
| "loss": 1.3427, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 9.698848538529672e-05, | |
| "loss": 1.3139, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 8.08, | |
| "learning_rate": 9.610274579273694e-05, | |
| "loss": 1.319, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 9.521700620017715e-05, | |
| "loss": 1.3558, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 9.433126660761736e-05, | |
| "loss": 1.3171, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 8.13, | |
| "learning_rate": 9.344552701505757e-05, | |
| "loss": 1.3402, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 9.255978742249779e-05, | |
| "loss": 1.3222, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 9.1674047829938e-05, | |
| "loss": 1.3373, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 9.078830823737822e-05, | |
| "loss": 1.3553, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 8.990256864481843e-05, | |
| "loss": 1.3406, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 8.22, | |
| "learning_rate": 8.901682905225864e-05, | |
| "loss": 1.3249, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 8.813108945969885e-05, | |
| "loss": 1.3145, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 8.26, | |
| "learning_rate": 8.724534986713905e-05, | |
| "loss": 1.3527, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 8.27, | |
| "learning_rate": 8.635961027457928e-05, | |
| "loss": 1.3283, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 8.54738706820195e-05, | |
| "loss": 1.3233, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 8.31, | |
| "learning_rate": 8.45881310894597e-05, | |
| "loss": 1.3495, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 8.370239149689991e-05, | |
| "loss": 1.3476, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 8.34, | |
| "learning_rate": 8.281665190434012e-05, | |
| "loss": 1.3237, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 8.193091231178033e-05, | |
| "loss": 1.3465, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 8.104517271922054e-05, | |
| "loss": 1.3308, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 8.015943312666077e-05, | |
| "loss": 1.3414, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 8.41, | |
| "learning_rate": 7.927369353410098e-05, | |
| "loss": 1.3329, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 8.43, | |
| "learning_rate": 7.838795394154119e-05, | |
| "loss": 1.346, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 8.45, | |
| "learning_rate": 7.75022143489814e-05, | |
| "loss": 1.3453, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 7.661647475642161e-05, | |
| "loss": 1.3197, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 8.49, | |
| "learning_rate": 7.573073516386182e-05, | |
| "loss": 1.3247, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 8.5, | |
| "learning_rate": 7.484499557130204e-05, | |
| "loss": 1.3389, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 7.395925597874226e-05, | |
| "loss": 1.3421, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 7.307351638618247e-05, | |
| "loss": 1.3203, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 7.218777679362268e-05, | |
| "loss": 1.3218, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 8.57, | |
| "learning_rate": 7.130203720106288e-05, | |
| "loss": 1.3637, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 8.59, | |
| "learning_rate": 7.041629760850309e-05, | |
| "loss": 1.3239, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 6.953055801594331e-05, | |
| "loss": 1.3559, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 8.63, | |
| "learning_rate": 6.864481842338352e-05, | |
| "loss": 1.3263, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 8.64, | |
| "learning_rate": 6.775907883082374e-05, | |
| "loss": 1.3626, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 8.66, | |
| "learning_rate": 6.687333923826395e-05, | |
| "loss": 1.3125, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 6.598759964570416e-05, | |
| "loss": 1.3079, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 6.510186005314437e-05, | |
| "loss": 1.3478, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 8.72, | |
| "learning_rate": 6.421612046058459e-05, | |
| "loss": 1.3252, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 8.73, | |
| "learning_rate": 6.33303808680248e-05, | |
| "loss": 1.3619, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 8.75, | |
| "learning_rate": 6.244464127546502e-05, | |
| "loss": 1.3281, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 6.155890168290523e-05, | |
| "loss": 1.316, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 8.79, | |
| "learning_rate": 6.067316209034544e-05, | |
| "loss": 1.3324, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 8.8, | |
| "learning_rate": 5.9787422497785654e-05, | |
| "loss": 1.3282, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 8.82, | |
| "learning_rate": 5.890168290522586e-05, | |
| "loss": 1.3412, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 5.801594331266608e-05, | |
| "loss": 1.3228, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 8.86, | |
| "learning_rate": 5.713020372010629e-05, | |
| "loss": 1.3367, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 8.88, | |
| "learning_rate": 5.62444641275465e-05, | |
| "loss": 1.3259, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 8.89, | |
| "learning_rate": 5.5358724534986716e-05, | |
| "loss": 1.3245, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 8.91, | |
| "learning_rate": 5.4472984942426924e-05, | |
| "loss": 1.3405, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 8.93, | |
| "learning_rate": 5.358724534986714e-05, | |
| "loss": 1.3278, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 8.95, | |
| "learning_rate": 5.2701505757307354e-05, | |
| "loss": 1.3012, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 5.181576616474756e-05, | |
| "loss": 1.2933, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 8.98, | |
| "learning_rate": 5.093002657218778e-05, | |
| "loss": 1.3753, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 5.004428697962799e-05, | |
| "loss": 1.3211, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 9.02, | |
| "learning_rate": 4.91585473870682e-05, | |
| "loss": 1.3282, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 9.03, | |
| "learning_rate": 4.827280779450841e-05, | |
| "loss": 1.2949, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 4.738706820194863e-05, | |
| "loss": 1.2804, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 4.650132860938884e-05, | |
| "loss": 1.2732, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 9.09, | |
| "learning_rate": 4.561558901682905e-05, | |
| "loss": 1.2743, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 4.472984942426927e-05, | |
| "loss": 1.2978, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 9.12, | |
| "learning_rate": 4.384410983170948e-05, | |
| "loss": 1.3016, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 4.2958370239149686e-05, | |
| "loss": 1.3, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 4.20726306465899e-05, | |
| "loss": 1.2877, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 9.18, | |
| "learning_rate": 4.1186891054030117e-05, | |
| "loss": 1.2664, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 4.030115146147033e-05, | |
| "loss": 1.2769, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 9.21, | |
| "learning_rate": 3.941541186891054e-05, | |
| "loss": 1.2926, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 3.8529672276350755e-05, | |
| "loss": 1.2731, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 9.25, | |
| "learning_rate": 3.764393268379097e-05, | |
| "loss": 1.2806, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 3.675819309123118e-05, | |
| "loss": 1.2822, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 3.587245349867139e-05, | |
| "loss": 1.2866, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 3.498671390611161e-05, | |
| "loss": 1.29, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 3.410097431355182e-05, | |
| "loss": 1.2734, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 9.34, | |
| "learning_rate": 3.3215234720992025e-05, | |
| "loss": 1.309, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 9.35, | |
| "learning_rate": 3.232949512843225e-05, | |
| "loss": 1.2796, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 3.1443755535872456e-05, | |
| "loss": 1.2581, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 9.39, | |
| "learning_rate": 3.0558015943312664e-05, | |
| "loss": 1.2936, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 2.967227635075288e-05, | |
| "loss": 1.3346, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 9.42, | |
| "learning_rate": 2.878653675819309e-05, | |
| "loss": 1.3168, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 9.44, | |
| "learning_rate": 2.7900797165633306e-05, | |
| "loss": 1.3255, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 2.7015057573073517e-05, | |
| "loss": 1.2798, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 2.612931798051373e-05, | |
| "loss": 1.2846, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 2.5243578387953944e-05, | |
| "loss": 1.2615, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 9.51, | |
| "learning_rate": 2.4357838795394153e-05, | |
| "loss": 1.2692, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 9.53, | |
| "learning_rate": 2.3472099202834368e-05, | |
| "loss": 1.2845, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 2.258635961027458e-05, | |
| "loss": 1.284, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 9.57, | |
| "learning_rate": 2.170062001771479e-05, | |
| "loss": 1.3194, | |
| "step": 54000 | |
| } | |
| ], | |
| "max_steps": 56450, | |
| "num_train_epochs": 10, | |
| "total_flos": 1.0185474088615219e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |