| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.08594757198109153, | |
| "global_step": 20000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.00040004211081201384, | |
| "loss": 8.3496, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000400168442509171, | |
| "loss": 8.2272, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.000400378992874836, | |
| "loss": 7.6879, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004006737582146567, | |
| "loss": 7.4747, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 0.0004010527333566261, | |
| "loss": 7.2829, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004015159116511832, | |
| "loss": 7.1171, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040206328497132196, | |
| "loss": 6.9445, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004026948437127389, | |
| "loss": 6.8391, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004034105767939909, | |
| "loss": 6.7131, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040421047165670534, | |
| "loss": 6.6113, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004050945142657896, | |
| "loss": 6.4966, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004060626891096795, | |
| "loss": 6.3979, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004071149792006148, | |
| "loss": 6.3116, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00040825136607492915, | |
| "loss": 6.2301, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004094718297933883, | |
| "loss": 6.123, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.0004107763489415231, | |
| "loss": 6.0802, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00041216490063001633, | |
| "loss": 6.0029, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00041363746049510354, | |
| "loss": 5.9471, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004151940026989945, | |
| "loss": 5.9132, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004168344999303346, | |
| "loss": 5.8561, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00041855892340467854, | |
| "loss": 5.8044, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004203672428649916, | |
| "loss": 5.734, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004222594265821944, | |
| "loss": 5.7245, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004242354413557057, | |
| "loss": 5.6867, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.00042629525251402893, | |
| "loss": 5.6387, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004284388239153662, | |
| "loss": 5.6119, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004306661179482429, | |
| "loss": 5.5533, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004329770955321787, | |
| "loss": 5.517, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0004353717161183629, | |
| "loss": 5.4864, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004378499376903721, | |
| "loss": 5.4671, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00044041171676490604, | |
| "loss": 5.4412, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004430570083925455, | |
| "loss": 5.4108, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004457857661585539, | |
| "loss": 5.3807, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004485979421836768, | |
| "loss": 5.3353, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004514934871249904, | |
| "loss": 5.3277, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00045447235017676696, | |
| "loss": 5.2979, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00045753447907136494, | |
| "loss": 5.2791, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.000460679820080143, | |
| "loss": 5.2494, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00046390831801440893, | |
| "loss": 5.2175, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0004672199162263843, | |
| "loss": 5.2038, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004706145566101966, | |
| "loss": 5.1835, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004740921796029061, | |
| "loss": 5.1691, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004776527241855382, | |
| "loss": 5.1582, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004812961278841711, | |
| "loss": 5.1504, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004850223267710129, | |
| "loss": 5.1162, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004888312554655432, | |
| "loss": 5.0957, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004927228471356421, | |
| "loss": 5.079, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0004966970334987757, | |
| "loss": 5.0572, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005007537448231871, | |
| "loss": 5.0342, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005048929099291249, | |
| "loss": 5.0106, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005091144561900837, | |
| "loss": 5.0155, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0005134183095340927, | |
| "loss": 4.9817, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005178043944449977, | |
| "loss": 4.9742, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005222726339638023, | |
| "loss": 4.9299, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005268229496900086, | |
| "loss": 4.9208, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005314552617829947, | |
| "loss": 4.8617, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005361694889634196, | |
| "loss": 4.7952, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005409655485146408, | |
| "loss": 4.7641, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005458433562841782, | |
| "loss": 4.7361, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005508028266851747, | |
| "loss": 4.7023, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.000555843872697916, | |
| "loss": 4.6561, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005609664058713396, | |
| "loss": 4.63, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.0005661703363245996, | |
| "loss": 4.6307, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005714555727486404, | |
| "loss": 4.5881, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005768220224077955, | |
| "loss": 4.5489, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005822695911414169, | |
| "loss": 4.5521, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005877981833655298, | |
| "loss": 4.5165, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005934077020745051, | |
| "loss": 4.505, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0005990980488427659, | |
| "loss": 4.4863, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000604869123826509, | |
| "loss": 4.5071, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0006107208257654633, | |
| "loss": 4.4501, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0006166530519846631, | |
| "loss": 4.4623, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0006226656983962468, | |
| "loss": 4.4336, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.0006287586595012887, | |
| "loss": 4.4335, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.000634931828391647, | |
| "loss": 4.4142, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006411850967518416, | |
| "loss": 4.4145, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006475183548609511, | |
| "loss": 4.3842, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006539314915945428, | |
| "loss": 4.3748, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006604243944266178, | |
| "loss": 4.3815, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006669969494315867, | |
| "loss": 4.352, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006736490412862749, | |
| "loss": 4.3575, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.000680380553271933, | |
| "loss": 4.3416, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006871913672762998, | |
| "loss": 4.341, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0006940813637956594, | |
| "loss": 4.3183, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0007010504219369541, | |
| "loss": 4.3145, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0007080984194198885, | |
| "loss": 4.3065, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0007152252325790948, | |
| "loss": 4.2805, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007224307363662818, | |
| "loss": 4.2804, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007297148043524434, | |
| "loss": 4.2996, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007370773087300737, | |
| "loss": 4.2743, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007445181203154048, | |
| "loss": 4.2621, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007520371085506811, | |
| "loss": 4.2548, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007596341415064441, | |
| "loss": 4.2643, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007673090858838494, | |
| "loss": 4.266, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007750618070170041, | |
| "loss": 4.2503, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007828921688753324, | |
| "loss": 4.2093, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007908000340659631, | |
| "loss": 4.2449, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.0007987852638361333, | |
| "loss": 4.2158, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0008068477180756314, | |
| "loss": 4.202, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0008149872553192515, | |
| "loss": 4.2065, | |
| "step": 20000 | |
| } | |
| ], | |
| "max_steps": 500000, | |
| "num_train_epochs": 3, | |
| "total_flos": 3.187659964416e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |