| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 100000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.005, | |
| "grad_norm": 0.5228317379951477, | |
| "learning_rate": 3.326666666666667e-06, | |
| "loss": 0.7898, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 1.0322492122650146, | |
| "learning_rate": 6.660000000000001e-06, | |
| "loss": 0.5367, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.015, | |
| "grad_norm": 2.0602364540100098, | |
| "learning_rate": 9.993333333333333e-06, | |
| "loss": 0.4491, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 0.8021494150161743, | |
| "learning_rate": 1.3326666666666667e-05, | |
| "loss": 0.3972, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.025, | |
| "grad_norm": 2.3705506324768066, | |
| "learning_rate": 1.666e-05, | |
| "loss": 0.3854, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 1.653393030166626, | |
| "learning_rate": 1.9993333333333337e-05, | |
| "loss": 0.3723, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.035, | |
| "grad_norm": 2.5747907161712646, | |
| "learning_rate": 1.9998694075951004e-05, | |
| "loss": 0.3747, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 2.956480026245117, | |
| "learning_rate": 1.9994766172915367e-05, | |
| "loss": 0.3435, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.045, | |
| "grad_norm": 1.4719043970108032, | |
| "learning_rate": 1.9988217315672814e-05, | |
| "loss": 0.3456, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 1.938084602355957, | |
| "learning_rate": 1.997904922154766e-05, | |
| "loss": 0.3284, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.055, | |
| "grad_norm": 3.1742770671844482, | |
| "learning_rate": 1.99672642947137e-05, | |
| "loss": 0.3323, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 3.088749647140503, | |
| "learning_rate": 1.995286562556374e-05, | |
| "loss": 0.333, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.065, | |
| "grad_norm": 4.315283298492432, | |
| "learning_rate": 1.993585698989922e-05, | |
| "loss": 0.335, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 1.9591349363327026, | |
| "learning_rate": 1.9916242847940064e-05, | |
| "loss": 0.3301, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.075, | |
| "grad_norm": 2.167367458343506, | |
| "learning_rate": 1.9894028343155063e-05, | |
| "loss": 0.3331, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 1.7126212120056152, | |
| "learning_rate": 1.9869219300913088e-05, | |
| "loss": 0.3305, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.085, | |
| "grad_norm": 2.834191083908081, | |
| "learning_rate": 1.9841822226955496e-05, | |
| "loss": 0.3144, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 1.9557807445526123, | |
| "learning_rate": 1.9811844305690107e-05, | |
| "loss": 0.3165, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.095, | |
| "grad_norm": 2.5267953872680664, | |
| "learning_rate": 1.977929339830722e-05, | |
| "loss": 0.3174, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 1.9610834121704102, | |
| "learning_rate": 1.9744178040718144e-05, | |
| "loss": 0.3093, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.105, | |
| "grad_norm": 3.3462774753570557, | |
| "learning_rate": 1.9706507441316814e-05, | |
| "loss": 0.3191, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 1.884909987449646, | |
| "learning_rate": 1.9666291478565047e-05, | |
| "loss": 0.3254, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.115, | |
| "grad_norm": 2.515130043029785, | |
| "learning_rate": 1.962354069840208e-05, | |
| "loss": 0.3134, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 3.2403035163879395, | |
| "learning_rate": 1.9578266311479082e-05, | |
| "loss": 0.3154, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.125, | |
| "grad_norm": 2.7380077838897705, | |
| "learning_rate": 1.9530480190219365e-05, | |
| "loss": 0.3203, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 3.1931233406066895, | |
| "learning_rate": 1.948019486570503e-05, | |
| "loss": 0.3058, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.135, | |
| "grad_norm": 3.3102500438690186, | |
| "learning_rate": 1.942742352439093e-05, | |
| "loss": 0.3024, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 3.0574374198913574, | |
| "learning_rate": 1.937218000464673e-05, | |
| "loss": 0.3115, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.145, | |
| "grad_norm": 3.5112788677215576, | |
| "learning_rate": 1.931447879312804e-05, | |
| "loss": 0.312, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 2.5837135314941406, | |
| "learning_rate": 1.9254335020977542e-05, | |
| "loss": 0.3095, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.155, | |
| "grad_norm": 2.8150858879089355, | |
| "learning_rate": 1.9191764459857124e-05, | |
| "loss": 0.3038, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 2.605210304260254, | |
| "learning_rate": 1.9126783517812005e-05, | |
| "loss": 0.2889, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.165, | |
| "grad_norm": 2.668278217315674, | |
| "learning_rate": 1.9059409234968034e-05, | |
| "loss": 0.2966, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 1.3171859979629517, | |
| "learning_rate": 1.8989659279063214e-05, | |
| "loss": 0.2891, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.175, | |
| "grad_norm": 3.820676565170288, | |
| "learning_rate": 1.8917551940814615e-05, | |
| "loss": 0.2953, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 2.1207947731018066, | |
| "learning_rate": 1.8843106129121993e-05, | |
| "loss": 0.2982, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.185, | |
| "grad_norm": 19.94713592529297, | |
| "learning_rate": 1.8766341366109214e-05, | |
| "loss": 0.2908, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 2.073420286178589, | |
| "learning_rate": 1.8687277782004954e-05, | |
| "loss": 0.2788, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.195, | |
| "grad_norm": 3.1489017009735107, | |
| "learning_rate": 1.8605936109863872e-05, | |
| "loss": 0.299, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 2.1240925788879395, | |
| "learning_rate": 1.8522337680129728e-05, | |
| "loss": 0.2905, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.205, | |
| "grad_norm": 2.2307348251342773, | |
| "learning_rate": 1.843650441504186e-05, | |
| "loss": 0.2928, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 1.1732121706008911, | |
| "learning_rate": 1.8348458822886437e-05, | |
| "loss": 0.2893, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.215, | |
| "grad_norm": 2.374523639678955, | |
| "learning_rate": 1.8258223992094055e-05, | |
| "loss": 0.2995, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 1.002447247505188, | |
| "learning_rate": 1.816582358518521e-05, | |
| "loss": 0.2884, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.225, | |
| "grad_norm": 2.586608409881592, | |
| "learning_rate": 1.80712818325652e-05, | |
| "loss": 0.2908, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 2.567784070968628, | |
| "learning_rate": 1.7974623526170113e-05, | |
| "loss": 0.2955, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.235, | |
| "grad_norm": 1.543729305267334, | |
| "learning_rate": 1.78758740129656e-05, | |
| "loss": 0.2848, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 4.3870954513549805, | |
| "learning_rate": 1.777505918830004e-05, | |
| "loss": 0.2856, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.245, | |
| "grad_norm": 2.431323528289795, | |
| "learning_rate": 1.767220548911395e-05, | |
| "loss": 0.2699, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 2.8813323974609375, | |
| "learning_rate": 1.756733988700737e-05, | |
| "loss": 0.275, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.255, | |
| "grad_norm": 2.1704187393188477, | |
| "learning_rate": 1.7460489881166986e-05, | |
| "loss": 0.2852, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 1.9194124937057495, | |
| "learning_rate": 1.7351683491154984e-05, | |
| "loss": 0.2963, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.265, | |
| "grad_norm": 2.9418258666992188, | |
| "learning_rate": 1.7240949249561402e-05, | |
| "loss": 0.282, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 2.2254602909088135, | |
| "learning_rate": 1.7128316194521923e-05, | |
| "loss": 0.269, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.275, | |
| "grad_norm": 3.698273181915283, | |
| "learning_rate": 1.7013813862103163e-05, | |
| "loss": 0.2748, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 1.1656277179718018, | |
| "learning_rate": 1.689747227855733e-05, | |
| "loss": 0.2798, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.285, | |
| "grad_norm": 2.6570346355438232, | |
| "learning_rate": 1.677932195244836e-05, | |
| "loss": 0.2791, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 3.1429874897003174, | |
| "learning_rate": 1.6659393866651585e-05, | |
| "loss": 0.2695, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.295, | |
| "grad_norm": 3.390726089477539, | |
| "learning_rate": 1.6537719470228993e-05, | |
| "loss": 0.2749, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 2.2655887603759766, | |
| "learning_rate": 1.6414330670182278e-05, | |
| "loss": 0.2764, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.305, | |
| "grad_norm": 2.555089235305786, | |
| "learning_rate": 1.6289259823085765e-05, | |
| "loss": 0.2834, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 2.014594793319702, | |
| "learning_rate": 1.6162539726601477e-05, | |
| "loss": 0.2801, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.315, | |
| "grad_norm": 3.4530141353607178, | |
| "learning_rate": 1.603420361087849e-05, | |
| "loss": 0.2789, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 1.4499799013137817, | |
| "learning_rate": 1.5904285129838924e-05, | |
| "loss": 0.2799, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.325, | |
| "grad_norm": 2.807821035385132, | |
| "learning_rate": 1.577281835235279e-05, | |
| "loss": 0.2717, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 2.2752676010131836, | |
| "learning_rate": 1.563983775330399e-05, | |
| "loss": 0.2694, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.335, | |
| "grad_norm": 1.8416115045547485, | |
| "learning_rate": 1.5505378204549907e-05, | |
| "loss": 0.2773, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 5.743653297424316, | |
| "learning_rate": 1.536947496577684e-05, | |
| "loss": 0.2786, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.345, | |
| "grad_norm": 2.5975189208984375, | |
| "learning_rate": 1.5232163675253793e-05, | |
| "loss": 0.2825, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 1.8504586219787598, | |
| "learning_rate": 1.5093480340486922e-05, | |
| "loss": 0.2693, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.355, | |
| "grad_norm": 2.5584330558776855, | |
| "learning_rate": 1.495346132877721e-05, | |
| "loss": 0.2688, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 3.1796674728393555, | |
| "learning_rate": 1.4812143357683773e-05, | |
| "loss": 0.2552, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.365, | |
| "grad_norm": 2.1949145793914795, | |
| "learning_rate": 1.4669563485395315e-05, | |
| "loss": 0.2735, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 2.1829614639282227, | |
| "learning_rate": 1.4525759101012249e-05, | |
| "loss": 0.2778, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.375, | |
| "grad_norm": 2.5891077518463135, | |
| "learning_rate": 1.4380767914742074e-05, | |
| "loss": 0.2703, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 1.912104606628418, | |
| "learning_rate": 1.4234627948010497e-05, | |
| "loss": 0.2603, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.385, | |
| "grad_norm": 2.5725393295288086, | |
| "learning_rate": 1.4087377523491016e-05, | |
| "loss": 0.2745, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 2.139127016067505, | |
| "learning_rate": 1.3939055255055419e-05, | |
| "loss": 0.2592, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.395, | |
| "grad_norm": 3.271855115890503, | |
| "learning_rate": 1.3789700037647988e-05, | |
| "loss": 0.2676, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 4.065393924713135, | |
| "learning_rate": 1.363935103708598e-05, | |
| "loss": 0.2582, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.405, | |
| "grad_norm": 4.319698333740234, | |
| "learning_rate": 1.3488047679789052e-05, | |
| "loss": 0.2738, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 1.9484474658966064, | |
| "learning_rate": 1.3335829642440372e-05, | |
| "loss": 0.2625, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.415, | |
| "grad_norm": 2.7909133434295654, | |
| "learning_rate": 1.3182736841582109e-05, | |
| "loss": 0.2732, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 1.9949915409088135, | |
| "learning_rate": 1.3028809423147997e-05, | |
| "loss": 0.2789, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.425, | |
| "grad_norm": 2.6040537357330322, | |
| "learning_rate": 1.2874087751935756e-05, | |
| "loss": 0.27, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 2.1791725158691406, | |
| "learning_rate": 1.2718612401022148e-05, | |
| "loss": 0.2682, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.435, | |
| "grad_norm": 2.334940195083618, | |
| "learning_rate": 1.2562424141123364e-05, | |
| "loss": 0.2702, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 3.011237144470215, | |
| "learning_rate": 1.2405563929903614e-05, | |
| "loss": 0.2684, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.445, | |
| "grad_norm": 3.324738025665283, | |
| "learning_rate": 1.2248072901234715e-05, | |
| "loss": 0.2626, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 2.0907318592071533, | |
| "learning_rate": 1.2089992354409415e-05, | |
| "loss": 0.2673, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.455, | |
| "grad_norm": 2.5054478645324707, | |
| "learning_rate": 1.1931363743311377e-05, | |
| "loss": 0.2634, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 3.015181541442871, | |
| "learning_rate": 1.1772228665544634e-05, | |
| "loss": 0.2623, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.465, | |
| "grad_norm": 1.9461313486099243, | |
| "learning_rate": 1.1612628851525313e-05, | |
| "loss": 0.2685, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 2.5035080909729004, | |
| "learning_rate": 1.1452606153538577e-05, | |
| "loss": 0.2625, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.475, | |
| "grad_norm": 1.4080424308776855, | |
| "learning_rate": 1.129220253476358e-05, | |
| "loss": 0.2651, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 2.9700748920440674, | |
| "learning_rate": 1.1131460058269342e-05, | |
| "loss": 0.2547, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.485, | |
| "grad_norm": 2.252192258834839, | |
| "learning_rate": 1.0970420875984453e-05, | |
| "loss": 0.2629, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 3.6359364986419678, | |
| "learning_rate": 1.0809127217643462e-05, | |
| "loss": 0.2623, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.495, | |
| "grad_norm": 2.2924039363861084, | |
| "learning_rate": 1.0647621379712843e-05, | |
| "loss": 0.2648, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 1.8426686525344849, | |
| "learning_rate": 1.0485945714299487e-05, | |
| "loss": 0.2536, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.505, | |
| "grad_norm": 2.85508131980896, | |
| "learning_rate": 1.0324142618044612e-05, | |
| "loss": 0.2685, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 1.8583101034164429, | |
| "learning_rate": 1.0162254521005956e-05, | |
| "loss": 0.2627, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.515, | |
| "grad_norm": 2.0837361812591553, | |
| "learning_rate": 1.0000323875531241e-05, | |
| "loss": 0.2594, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 2.786468505859375, | |
| "learning_rate": 9.838393145125787e-06, | |
| "loss": 0.2602, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.525, | |
| "grad_norm": 3.2767674922943115, | |
| "learning_rate": 9.676504793317163e-06, | |
| "loss": 0.2575, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 2.087574005126953, | |
| "learning_rate": 9.514701272519918e-06, | |
| "loss": 0.2536, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.535, | |
| "grad_norm": 1.6230825185775757, | |
| "learning_rate": 9.353025012903112e-06, | |
| "loss": 0.2634, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 1.9455654621124268, | |
| "learning_rate": 9.191518411263777e-06, | |
| "loss": 0.2415, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.545, | |
| "grad_norm": 2.6995887756347656, | |
| "learning_rate": 9.03022381990908e-06, | |
| "loss": 0.2687, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 4.489888668060303, | |
| "learning_rate": 8.869183535550167e-06, | |
| "loss": 0.2719, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.555, | |
| "grad_norm": 2.966480016708374, | |
| "learning_rate": 8.708439788210573e-06, | |
| "loss": 0.268, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 3.2469143867492676, | |
| "learning_rate": 8.548034730152153e-06, | |
| "loss": 0.262, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.565, | |
| "grad_norm": 2.2795863151550293, | |
| "learning_rate": 8.388010424821354e-06, | |
| "loss": 0.2644, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 2.264314889907837, | |
| "learning_rate": 8.22840883581883e-06, | |
| "loss": 0.2519, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.575, | |
| "grad_norm": 2.3506600856781006, | |
| "learning_rate": 8.069271815895196e-06, | |
| "loss": 0.2797, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 4.5607428550720215, | |
| "learning_rate": 7.910641095975886e-06, | |
| "loss": 0.2574, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.585, | |
| "grad_norm": 0.9732487201690674, | |
| "learning_rate": 7.752558274217919e-06, | |
| "loss": 0.2529, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 3.291529417037964, | |
| "learning_rate": 7.595064805101538e-06, | |
| "loss": 0.2502, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.595, | |
| "grad_norm": 1.8199131488800049, | |
| "learning_rate": 7.438201988559449e-06, | |
| "loss": 0.2643, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 5.467274188995361, | |
| "learning_rate": 7.282010959146662e-06, | |
| "loss": 0.2636, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.605, | |
| "grad_norm": 3.899115562438965, | |
| "learning_rate": 7.1265326752536545e-06, | |
| "loss": 0.2625, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 2.5981805324554443, | |
| "learning_rate": 6.971807908365737e-06, | |
| "loss": 0.2577, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.615, | |
| "grad_norm": 2.875891923904419, | |
| "learning_rate": 6.817877232371453e-06, | |
| "loss": 0.2649, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 2.9105169773101807, | |
| "learning_rate": 6.66478101292277e-06, | |
| "loss": 0.2496, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.625, | |
| "grad_norm": 3.293264389038086, | |
| "learning_rate": 6.5125593968498855e-06, | |
| "loss": 0.2444, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 3.560056209564209, | |
| "learning_rate": 6.361252301633422e-06, | |
| "loss": 0.2625, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.635, | |
| "grad_norm": 2.613802909851074, | |
| "learning_rate": 6.210899404936761e-06, | |
| "loss": 0.2526, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 3.4607479572296143, | |
| "learning_rate": 6.061540134201248e-06, | |
| "loss": 0.2556, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.645, | |
| "grad_norm": 3.917365074157715, | |
| "learning_rate": 5.913213656307049e-06, | |
| "loss": 0.2601, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 2.4207093715667725, | |
| "learning_rate": 5.765958867302311e-06, | |
| "loss": 0.2512, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.655, | |
| "grad_norm": 2.076265335083008, | |
| "learning_rate": 5.619814382203341e-06, | |
| "loss": 0.2437, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 3.774073600769043, | |
| "learning_rate": 5.474818524868482e-06, | |
| "loss": 0.2726, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.665, | |
| "grad_norm": 3.220529079437256, | |
| "learning_rate": 5.331009317948359e-06, | |
| "loss": 0.2664, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 2.737941026687622, | |
| "learning_rate": 5.188424472915091e-06, | |
| "loss": 0.2703, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.675, | |
| "grad_norm": 2.459578275680542, | |
| "learning_rate": 5.047101380173111e-06, | |
| "loss": 0.244, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 3.3251523971557617, | |
| "learning_rate": 4.907077099254194e-06, | |
| "loss": 0.2597, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.685, | |
| "grad_norm": 2.779625654220581, | |
| "learning_rate": 4.7683883490991985e-06, | |
| "loss": 0.2623, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 1.0731335878372192, | |
| "learning_rate": 4.6310714984292096e-06, | |
| "loss": 0.2616, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.695, | |
| "grad_norm": 1.1064262390136719, | |
| "learning_rate": 4.495162556208438e-06, | |
| "loss": 0.2489, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 6.329731464385986, | |
| "learning_rate": 4.360697162201508e-06, | |
| "loss": 0.254, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.705, | |
| "grad_norm": 3.1738226413726807, | |
| "learning_rate": 4.2277105776275395e-06, | |
| "loss": 0.249, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 2.834291696548462, | |
| "learning_rate": 4.096237675913521e-06, | |
| "loss": 0.2607, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.715, | |
| "grad_norm": 2.2489569187164307, | |
| "learning_rate": 3.966312933549325e-06, | |
| "loss": 0.24, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 4.554741859436035, | |
| "learning_rate": 3.837970421046881e-06, | |
| "loss": 0.2492, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.725, | |
| "grad_norm": 4.065186977386475, | |
| "learning_rate": 3.7112437940057543e-06, | |
| "loss": 0.2491, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 3.260732650756836, | |
| "learning_rate": 3.5861662842875534e-06, | |
| "loss": 0.2413, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.735, | |
| "grad_norm": 4.497494220733643, | |
| "learning_rate": 3.4627706913014503e-06, | |
| "loss": 0.2537, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 2.5874199867248535, | |
| "learning_rate": 3.3410893734031046e-06, | |
| "loss": 0.2627, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.745, | |
| "grad_norm": 2.394810438156128, | |
| "learning_rate": 3.2211542394092443e-06, | |
| "loss": 0.2516, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 2.5389678478240967, | |
| "learning_rate": 3.102996740230131e-06, | |
| "loss": 0.2434, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.755, | |
| "grad_norm": 1.8920937776565552, | |
| "learning_rate": 2.986647860622105e-06, | |
| "loss": 0.2602, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 1.569104790687561, | |
| "learning_rate": 2.8721381110623724e-06, | |
| "loss": 0.248, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.765, | |
| "grad_norm": 2.620171308517456, | |
| "learning_rate": 2.7594975197481546e-06, | |
| "loss": 0.249, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 3.781393527984619, | |
| "learning_rate": 2.64875562472232e-06, | |
| "loss": 0.2475, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.775, | |
| "grad_norm": 4.293469429016113, | |
| "learning_rate": 2.5399414661275444e-06, | |
| "loss": 0.2584, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 1.6750516891479492, | |
| "learning_rate": 2.433083578591029e-06, | |
| "loss": 0.2471, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.785, | |
| "grad_norm": 3.109180450439453, | |
| "learning_rate": 2.3282099837418004e-06, | |
| "loss": 0.257, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 3.492893695831299, | |
| "learning_rate": 2.2253481828625144e-06, | |
| "loss": 0.2644, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.795, | |
| "grad_norm": 3.8731396198272705, | |
| "learning_rate": 2.1245251496777274e-06, | |
| "loss": 0.2546, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 2.1617014408111572, | |
| "learning_rate": 2.025767323280502e-06, | |
| "loss": 0.2542, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.805, | |
| "grad_norm": 5.593191146850586, | |
| "learning_rate": 1.929100601199211e-06, | |
| "loss": 0.2376, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 3.496340036392212, | |
| "learning_rate": 1.834550332606365e-06, | |
| "loss": 0.234, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.815, | |
| "grad_norm": 2.7082982063293457, | |
| "learning_rate": 1.742141311671227e-06, | |
| "loss": 0.2482, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 2.0201199054718018, | |
| "learning_rate": 1.6518977710579708e-06, | |
| "loss": 0.2676, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.825, | |
| "grad_norm": 2.157087564468384, | |
| "learning_rate": 1.5638433755710892e-06, | |
| "loss": 0.2505, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 2.467223882675171, | |
| "learning_rate": 1.4780012159497037e-06, | |
| "loss": 0.2632, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.835, | |
| "grad_norm": 1.9714136123657227, | |
| "learning_rate": 1.3943938028124216e-06, | |
| "loss": 0.2568, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 6.337538242340088, | |
| "learning_rate": 1.3130430607543166e-06, | |
| "loss": 0.2555, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.845, | |
| "grad_norm": 5.382758617401123, | |
| "learning_rate": 1.2339703225975818e-06, | |
| "loss": 0.2508, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 2.141634464263916, | |
| "learning_rate": 1.1571963237973693e-06, | |
| "loss": 0.246, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.855, | |
| "grad_norm": 3.525200128555298, | |
| "learning_rate": 1.0827411970042723e-06, | |
| "loss": 0.2526, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 4.05570650100708, | |
| "learning_rate": 1.010624466784892e-06, | |
| "loss": 0.2479, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.865, | |
| "grad_norm": 4.631308555603027, | |
| "learning_rate": 9.40865044501853e-07, | |
| "loss": 0.2495, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 2.8069100379943848, | |
| "learning_rate": 8.734812233546319e-07, | |
| "loss": 0.252, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.875, | |
| "grad_norm": 3.1960978507995605, | |
| "learning_rate": 8.084906735824771e-07, | |
| "loss": 0.2531, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 5.496489524841309, | |
| "learning_rate": 7.45910437830697e-07, | |
| "loss": 0.2495, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 0.885, | |
| "grad_norm": 3.19916033744812, | |
| "learning_rate": 6.857569266815233e-07, | |
| "loss": 0.2462, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 3.560880422592163, | |
| "learning_rate": 6.2804591435072e-07, | |
| "loss": 0.2611, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 0.895, | |
| "grad_norm": 1.2571769952774048, | |
| "learning_rate": 5.727925345510644e-07, | |
| "loss": 0.2428, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 1.8454118967056274, | |
| "learning_rate": 5.200112765237985e-07, | |
| "loss": 0.25, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 0.905, | |
| "grad_norm": 2.3532028198242188, | |
| "learning_rate": 4.697159812390739e-07, | |
| "loss": 0.2442, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 3.9813899993896484, | |
| "learning_rate": 4.21919837766398e-07, | |
| "loss": 0.2474, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 0.915, | |
| "grad_norm": 2.554281711578369, | |
| "learning_rate": 3.766353798160283e-07, | |
| "loss": 0.2429, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 2.701397657394409, | |
| "learning_rate": 3.338744824522266e-07, | |
| "loss": 0.2477, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 0.925, | |
| "grad_norm": 2.0029845237731934, | |
| "learning_rate": 2.9364835897922606e-07, | |
| "loss": 0.2514, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 5.601722240447998, | |
| "learning_rate": 2.5596755800073593e-07, | |
| "loss": 0.247, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 0.935, | |
| "grad_norm": 2.904860019683838, | |
| "learning_rate": 2.208419606537593e-07, | |
| "loss": 0.2519, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 2.9987542629241943, | |
| "learning_rate": 1.8828077801743006e-07, | |
| "loss": 0.2559, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 0.945, | |
| "grad_norm": 3.036500930786133, | |
| "learning_rate": 1.5829254869756395e-07, | |
| "loss": 0.2558, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 1.084037184715271, | |
| "learning_rate": 1.3088513658756962e-07, | |
| "loss": 0.2485, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 0.955, | |
| "grad_norm": 1.9537458419799805, | |
| "learning_rate": 1.060657288062672e-07, | |
| "loss": 0.2495, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 2.6217870712280273, | |
| "learning_rate": 8.384083381319464e-08, | |
| "loss": 0.2438, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 0.965, | |
| "grad_norm": 4.048638343811035, | |
| "learning_rate": 6.421627970187971e-08, | |
| "loss": 0.2466, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "grad_norm": 2.7693753242492676, | |
| "learning_rate": 4.719721267152033e-08, | |
| "loss": 0.2578, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 0.975, | |
| "grad_norm": 4.596212863922119, | |
| "learning_rate": 3.2788095677480644e-08, | |
| "loss": 0.2506, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "grad_norm": 2.3294010162353516, | |
| "learning_rate": 2.099270726096503e-08, | |
| "loss": 0.2518, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 0.985, | |
| "grad_norm": 4.738352298736572, | |
| "learning_rate": 1.1814140558158438e-08, | |
| "loss": 0.2449, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "grad_norm": 3.600982904434204, | |
| "learning_rate": 5.254802489100819e-09, | |
| "loss": 0.2528, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 0.995, | |
| "grad_norm": 1.760666012763977, | |
| "learning_rate": 1.3164131265197555e-09, | |
| "loss": 0.253, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 2.2342171669006348, | |
| "learning_rate": 5.2447679532718895e-15, | |
| "loss": 0.2454, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 100000, | |
| "total_flos": 1.1225596249876746e+18, | |
| "train_loss": 0.27730819496154785, | |
| "train_runtime": 24748.5341, | |
| "train_samples_per_second": 4.041, | |
| "train_steps_per_second": 4.041 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 100000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 5000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.1225596249876746e+18, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |