| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9903536977491961, | |
| "eval_steps": 500, | |
| "global_step": 77, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6351, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 5e-05, | |
| "loss": 1.6528, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001, | |
| "loss": 1.6835, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.995614150494293e-05, | |
| "loss": 1.6183, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.982464296247522e-05, | |
| "loss": 1.6164, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.96057350657239e-05, | |
| "loss": 1.4991, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.929980185352526e-05, | |
| "loss": 1.4829, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.890738003669029e-05, | |
| "loss": 1.4138, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.842915805643155e-05, | |
| "loss": 1.4584, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.786597487660337e-05, | |
| "loss": 1.3482, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.721881851187406e-05, | |
| "loss": 1.3124, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.648882429441257e-05, | |
| "loss": 1.236, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.567727288213005e-05, | |
| "loss": 1.2427, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.478558801197065e-05, | |
| "loss": 1.1855, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.381533400219318e-05, | |
| "loss": 1.1507, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.276821300802534e-05, | |
| "loss": 1.1127, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.164606203550497e-05, | |
| "loss": 1.0321, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 1.0016, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.9184672866292e-05, | |
| "loss": 0.9561, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.784975278258783e-05, | |
| "loss": 0.9562, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.644843137107059e-05, | |
| "loss": 0.8827, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.498316702566828e-05, | |
| "loss": 0.8499, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.345653031794292e-05, | |
| "loss": 0.8233, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.18711994874345e-05, | |
| "loss": 0.8412, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.022995574311876e-05, | |
| "loss": 0.7966, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.85356783842216e-05, | |
| "loss": 0.7664, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.679133974894983e-05, | |
| "loss": 0.7188, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.500000000000001e-05, | |
| "loss": 0.6549, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.316480175599309e-05, | |
| "loss": 0.6638, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.128896457825364e-05, | |
| "loss": 0.6472, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.937577932260515e-05, | |
| "loss": 0.6237, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.742860236609077e-05, | |
| "loss": 0.5706, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.5788, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.344599103076329e-05, | |
| "loss": 0.5543, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.141754350553279e-05, | |
| "loss": 0.5639, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.9369065729286245e-05, | |
| "loss": 0.5404, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.730415142812059e-05, | |
| "loss": 0.5429, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.522642316338268e-05, | |
| "loss": 0.506, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.313952597646568e-05, | |
| "loss": 0.4793, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.104712099416785e-05, | |
| "loss": 0.4886, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.895287900583216e-05, | |
| "loss": 0.4514, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.6860474023534335e-05, | |
| "loss": 0.4984, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.477357683661734e-05, | |
| "loss": 0.4464, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.269584857187943e-05, | |
| "loss": 0.4846, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.063093427071376e-05, | |
| "loss": 0.4491, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.858245649446721e-05, | |
| "loss": 0.4662, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.655400896923672e-05, | |
| "loss": 0.4442, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.454, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.257139763390925e-05, | |
| "loss": 0.3953, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.062422067739485e-05, | |
| "loss": 0.4575, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.8711035421746367e-05, | |
| "loss": 0.4269, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.6835198244006927e-05, | |
| "loss": 0.411, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.500000000000001e-05, | |
| "loss": 0.4304, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.3208660251050158e-05, | |
| "loss": 0.4284, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.1464321615778422e-05, | |
| "loss": 0.4016, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.977004425688126e-05, | |
| "loss": 0.428, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8128800512565513e-05, | |
| "loss": 0.3949, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.6543469682057106e-05, | |
| "loss": 0.4035, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.5016832974331724e-05, | |
| "loss": 0.3929, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3551568628929434e-05, | |
| "loss": 0.37, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2150247217412186e-05, | |
| "loss": 0.3788, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0815327133708015e-05, | |
| "loss": 0.4043, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.4129, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.353937964495029e-06, | |
| "loss": 0.3976, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 7.2317869919746705e-06, | |
| "loss": 0.3999, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.184665997806832e-06, | |
| "loss": 0.3564, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.214411988029355e-06, | |
| "loss": 0.3593, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.322727117869951e-06, | |
| "loss": 0.3916, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.511175705587433e-06, | |
| "loss": 0.3861, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7811814881259503e-06, | |
| "loss": 0.3967, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.134025123396638e-06, | |
| "loss": 0.3624, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.5708419435684462e-06, | |
| "loss": 0.358, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0926199633097157e-06, | |
| "loss": 0.3965, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.001981464747565e-07, | |
| "loss": 0.3942, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.9426493427611177e-07, | |
| "loss": 0.3857, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.753570375247815e-07, | |
| "loss": 0.4294, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.385849505708084e-08, | |
| "loss": 0.3373, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "step": 77, | |
| "total_flos": 543429523144704.0, | |
| "train_loss": 0.7125843113118951, | |
| "train_runtime": 6470.7689, | |
| "train_samples_per_second": 0.769, | |
| "train_steps_per_second": 0.012 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 77, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50000, | |
| "total_flos": 543429523144704.0, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |