| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.9976693857166639, | |
| "eval_steps": 500, | |
| "global_step": 6000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.9583749583749586e-05, | |
| "loss": 0.7997, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.916749916749917e-05, | |
| "loss": 0.5713, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.8751248751248754e-05, | |
| "loss": 0.5638, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.833499833499834e-05, | |
| "loss": 0.5656, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.791874791874792e-05, | |
| "loss": 0.5687, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.7502497502497505e-05, | |
| "loss": 0.59, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.708624708624709e-05, | |
| "loss": 0.5713, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.666999666999667e-05, | |
| "loss": 0.5466, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.6253746253746256e-05, | |
| "loss": 0.5835, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.583749583749584e-05, | |
| "loss": 0.5822, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.5421245421245424e-05, | |
| "loss": 0.5591, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.500499500499501e-05, | |
| "loss": 0.5804, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.458874458874459e-05, | |
| "loss": 0.579, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.4172494172494175e-05, | |
| "loss": 0.5462, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.375624375624376e-05, | |
| "loss": 0.5822, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.333999333999334e-05, | |
| "loss": 0.5678, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.2923742923742926e-05, | |
| "loss": 0.5657, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.250749250749251e-05, | |
| "loss": 0.578, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.2091242091242094e-05, | |
| "loss": 0.5688, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.167499167499168e-05, | |
| "loss": 0.5389, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.125874125874126e-05, | |
| "loss": 0.5079, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.0842490842490845e-05, | |
| "loss": 0.5762, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.042624042624043e-05, | |
| "loss": 0.5532, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.000999000999001e-05, | |
| "loss": 0.5315, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 3.9593739593739596e-05, | |
| "loss": 0.5308, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 3.917748917748918e-05, | |
| "loss": 0.5444, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 3.8761238761238764e-05, | |
| "loss": 0.533, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 3.834498834498835e-05, | |
| "loss": 0.5504, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 3.792873792873793e-05, | |
| "loss": 0.5248, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 3.7512487512487515e-05, | |
| "loss": 0.5588, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 3.70962370962371e-05, | |
| "loss": 0.5198, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 3.667998667998668e-05, | |
| "loss": 0.5509, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 3.6263736263736266e-05, | |
| "loss": 0.5105, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 3.584748584748585e-05, | |
| "loss": 0.5441, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.5431235431235434e-05, | |
| "loss": 0.5263, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.501498501498502e-05, | |
| "loss": 0.5168, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.45987345987346e-05, | |
| "loss": 0.5312, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.4182484182484185e-05, | |
| "loss": 0.5447, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.376623376623377e-05, | |
| "loss": 0.5103, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.334998334998335e-05, | |
| "loss": 0.5377, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.2933732933732936e-05, | |
| "loss": 0.5388, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.251748251748252e-05, | |
| "loss": 0.5226, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.2101232101232104e-05, | |
| "loss": 0.536, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.168498168498169e-05, | |
| "loss": 0.551, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.126873126873127e-05, | |
| "loss": 0.5245, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.0852480852480855e-05, | |
| "loss": 0.5272, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.043623043623044e-05, | |
| "loss": 0.529, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.0019980019980022e-05, | |
| "loss": 0.5391, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 2.9603729603729606e-05, | |
| "loss": 0.5241, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 2.918747918747919e-05, | |
| "loss": 0.4842, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 2.8771228771228777e-05, | |
| "loss": 0.5078, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 2.8354978354978357e-05, | |
| "loss": 0.5291, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 2.7938727938727938e-05, | |
| "loss": 0.5165, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7522477522477525e-05, | |
| "loss": 0.5039, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 2.7106227106227105e-05, | |
| "loss": 0.5038, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 2.6689976689976692e-05, | |
| "loss": 0.5282, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 2.6273726273726273e-05, | |
| "loss": 0.4997, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.585747585747586e-05, | |
| "loss": 0.5096, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 2.544122544122544e-05, | |
| "loss": 0.504, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 2.5024975024975027e-05, | |
| "loss": 0.4911, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 2.4608724608724608e-05, | |
| "loss": 0.3976, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 2.4192474192474195e-05, | |
| "loss": 0.3728, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 2.377622377622378e-05, | |
| "loss": 0.3908, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 2.3359973359973362e-05, | |
| "loss": 0.366, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 2.2943722943722946e-05, | |
| "loss": 0.3728, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 2.252747252747253e-05, | |
| "loss": 0.3972, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 2.2111222111222114e-05, | |
| "loss": 0.3567, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 2.1694971694971697e-05, | |
| "loss": 0.3751, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 2.127872127872128e-05, | |
| "loss": 0.3915, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 2.0862470862470865e-05, | |
| "loss": 0.3891, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 2.0446220446220445e-05, | |
| "loss": 0.3679, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.002997002997003e-05, | |
| "loss": 0.369, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.9613719613719613e-05, | |
| "loss": 0.3613, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 1.9197469197469197e-05, | |
| "loss": 0.3768, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.878121878121878e-05, | |
| "loss": 0.3695, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.8364968364968367e-05, | |
| "loss": 0.36, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.794871794871795e-05, | |
| "loss": 0.3642, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 1.7532467532467535e-05, | |
| "loss": 0.3726, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.711621711621712e-05, | |
| "loss": 0.3851, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.6699966699966702e-05, | |
| "loss": 0.3561, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.6283716283716286e-05, | |
| "loss": 0.3779, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.586746586746587e-05, | |
| "loss": 0.3684, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 1.5451215451215454e-05, | |
| "loss": 0.3671, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.5034965034965034e-05, | |
| "loss": 0.3754, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.4618714618714618e-05, | |
| "loss": 0.3694, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 1.4202464202464203e-05, | |
| "loss": 0.375, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 1.3786213786213787e-05, | |
| "loss": 0.3926, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.336996336996337e-05, | |
| "loss": 0.3766, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.2953712953712954e-05, | |
| "loss": 0.366, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 1.2537462537462538e-05, | |
| "loss": 0.3651, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 1.2121212121212122e-05, | |
| "loss": 0.3738, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 1.1704961704961706e-05, | |
| "loss": 0.3833, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 1.128871128871129e-05, | |
| "loss": 0.3555, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.0872460872460873e-05, | |
| "loss": 0.3889, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 1.0456210456210457e-05, | |
| "loss": 0.3683, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 1.003996003996004e-05, | |
| "loss": 0.381, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.623709623709624e-06, | |
| "loss": 0.3851, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.207459207459208e-06, | |
| "loss": 0.3851, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.791208791208792e-06, | |
| "loss": 0.3776, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.374958374958376e-06, | |
| "loss": 0.3722, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 7.95870795870796e-06, | |
| "loss": 0.3529, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 7.542457542457542e-06, | |
| "loss": 0.3695, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 7.126207126207127e-06, | |
| "loss": 0.3768, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 6.709956709956711e-06, | |
| "loss": 0.3703, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 6.2937062937062944e-06, | |
| "loss": 0.386, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 5.877455877455878e-06, | |
| "loss": 0.3754, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 5.461205461205462e-06, | |
| "loss": 0.3691, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 5.044955044955045e-06, | |
| "loss": 0.3684, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 4.628704628704629e-06, | |
| "loss": 0.3665, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 4.212454212454213e-06, | |
| "loss": 0.3646, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 3.796203796203796e-06, | |
| "loss": 0.3644, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.3799533799533803e-06, | |
| "loss": 0.377, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 2.9637029637029636e-06, | |
| "loss": 0.3781, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 2.5474525474525478e-06, | |
| "loss": 0.3593, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 2.131202131202131e-06, | |
| "loss": 0.3967, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.7149517149517148e-06, | |
| "loss": 0.3754, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.2987012987012988e-06, | |
| "loss": 0.3419, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 8.824508824508825e-07, | |
| "loss": 0.352, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 4.662004662004662e-07, | |
| "loss": 0.3515, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 4.9950049950049954e-08, | |
| "loss": 0.3678, | |
| "step": 6000 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 6006, | |
| "num_train_epochs": 2, | |
| "save_steps": 500, | |
| "total_flos": 0.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |