|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 1.9937694704049844, |
|
"eval_steps": 500, |
|
"global_step": 80, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 6.666666666666667e-06, |
|
"loss": 1.6772, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.3333333333333333e-05, |
|
"loss": 1.6591, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 2e-05, |
|
"loss": 1.7126, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 1.9991677993445832e-05, |
|
"loss": 1.7077, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.9966725824941933e-05, |
|
"loss": 1.6977, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 1.992518502491028e-05, |
|
"loss": 1.6409, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 1.986712473391289e-05, |
|
"loss": 1.6717, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 1.9792641587574212e-05, |
|
"loss": 1.6683, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 1.9701859555740647e-05, |
|
"loss": 1.6339, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 1.9594929736144978e-05, |
|
"loss": 1.5765, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 1.9472030102919102e-05, |
|
"loss": 1.6116, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 1.9333365210373668e-05, |
|
"loss": 1.6752, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 1.9179165852537596e-05, |
|
"loss": 1.5735, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 1.900968867902419e-05, |
|
"loss": 1.6554, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 1.8825215767863215e-05, |
|
"loss": 1.6075, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 1.8626054156009807e-05, |
|
"loss": 1.6091, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 1.8412535328311813e-05, |
|
"loss": 1.599, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 1.8185014665785936e-05, |
|
"loss": 1.5027, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 1.7943870854121126e-05, |
|
"loss": 1.5883, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 1.768950525339362e-05, |
|
"loss": 1.5739, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 1.74223412300427e-05, |
|
"loss": 1.5958, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 1.7142823452219036e-05, |
|
"loss": 1.489, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 1.6851417149678442e-05, |
|
"loss": 1.5861, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 1.6548607339452853e-05, |
|
"loss": 1.5902, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 1.6234898018587336e-05, |
|
"loss": 1.514, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 1.5910811325286768e-05, |
|
"loss": 1.5592, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 1.5576886669868297e-05, |
|
"loss": 1.5757, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 1.5233679836966122e-05, |
|
"loss": 1.5668, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 1.4881762060482814e-05, |
|
"loss": 1.5971, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 1.4521719072826858e-05, |
|
"loss": 1.5389, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 1.4154150130018867e-05, |
|
"loss": 1.5161, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 1.3779667014289067e-05, |
|
"loss": 1.6, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 1.3398893015826166e-05, |
|
"loss": 1.5598, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.3012461895372343e-05, |
|
"loss": 1.5645, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.2621016829391022e-05, |
|
"loss": 1.8278, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 1.2225209339563144e-05, |
|
"loss": 1.5166, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 1.182569820839362e-05, |
|
"loss": 1.7329, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.1423148382732854e-05, |
|
"loss": 1.5313, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.1018229867038358e-05, |
|
"loss": 1.5852, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 1.0611616608218429e-05, |
|
"loss": 1.5125, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 1.0203985373914056e-05, |
|
"loss": 1.4622, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 9.79601462608595e-06, |
|
"loss": 1.4274, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 9.388383391781576e-06, |
|
"loss": 1.4272, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 8.981770132961649e-06, |
|
"loss": 1.4694, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 8.576851617267151e-06, |
|
"loss": 1.4136, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 1.15, |
|
"learning_rate": 8.174301791606384e-06, |
|
"loss": 1.39, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 7.774790660436857e-06, |
|
"loss": 1.3597, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 7.378983170608982e-06, |
|
"loss": 1.3729, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 6.9875381046276605e-06, |
|
"loss": 1.4594, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 6.601106984173835e-06, |
|
"loss": 1.3709, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 6.220332985710936e-06, |
|
"loss": 1.3582, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 5.845849869981137e-06, |
|
"loss": 1.3942, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 5.478280927173145e-06, |
|
"loss": 1.4468, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 5.11823793951719e-06, |
|
"loss": 1.3877, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 4.766320163033882e-06, |
|
"loss": 1.4887, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 4.423113330131708e-06, |
|
"loss": 1.3755, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 4.0891886747132356e-06, |
|
"loss": 1.3375, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 3.7651019814126656e-06, |
|
"loss": 1.3972, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 3.4513926605471504e-06, |
|
"loss": 1.4204, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 3.1485828503215588e-06, |
|
"loss": 1.4541, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 2.8571765477809645e-06, |
|
"loss": 1.4123, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 2.5776587699573007e-06, |
|
"loss": 1.3873, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 2.3104947466063785e-06, |
|
"loss": 1.4418, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 2.0561291458788736e-06, |
|
"loss": 1.4231, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 1.8149853342140644e-06, |
|
"loss": 1.3752, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 1.587464671688187e-06, |
|
"loss": 1.4395, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 1.373945843990192e-06, |
|
"loss": 1.3792, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 1.1747842321367886e-06, |
|
"loss": 1.4226, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 9.903113209758098e-07, |
|
"loss": 1.4125, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 8.208341474624071e-07, |
|
"loss": 1.4429, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 6.666347896263326e-07, |
|
"loss": 1.3913, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 5.279698970809011e-07, |
|
"loss": 1.4418, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 4.0507026385502747e-07, |
|
"loss": 1.3814, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 2.9814044425935605e-07, |
|
"loss": 1.4249, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 2.0735841242578992e-07, |
|
"loss": 1.4325, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 1.3287526608711132e-07, |
|
"loss": 1.4635, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 1.92, |
|
"learning_rate": 7.481497508972313e-08, |
|
"loss": 1.4079, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 3.327417505806785e-08, |
|
"loss": 1.3911, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 8.322006554171147e-09, |
|
"loss": 1.4636, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 0.0, |
|
"loss": 1.318, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"step": 80, |
|
"total_flos": 2166787537108992.0, |
|
"train_loss": 1.5083298146724702, |
|
"train_runtime": 64.7879, |
|
"train_samples_per_second": 158.363, |
|
"train_steps_per_second": 1.235 |
|
} |
|
], |
|
"logging_steps": 1.0, |
|
"max_steps": 80, |
|
"num_train_epochs": 2, |
|
"save_steps": 500, |
|
"total_flos": 2166787537108992.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|