|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 100.0, |
|
"eval_steps": 100, |
|
"global_step": 10600, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.3773584905660377e-06, |
|
"loss": 126.7325, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 3.2641509433962264e-06, |
|
"loss": 92.4996, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 5.150943396226416e-06, |
|
"loss": 46.8471, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 3.77, |
|
"learning_rate": 7.037735849056604e-06, |
|
"loss": 29.0405, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 4.72, |
|
"learning_rate": 8.924528301886793e-06, |
|
"loss": 21.3856, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 5.66, |
|
"learning_rate": 1.0811320754716984e-05, |
|
"loss": 16.0048, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 6.6, |
|
"learning_rate": 1.2698113207547172e-05, |
|
"loss": 11.667, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 7.55, |
|
"learning_rate": 1.458490566037736e-05, |
|
"loss": 8.5327, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 8.49, |
|
"learning_rate": 1.6471698113207547e-05, |
|
"loss": 6.5511, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 9.43, |
|
"learning_rate": 1.8358490566037736e-05, |
|
"loss": 5.4833, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 10.38, |
|
"learning_rate": 1.9999908365494266e-05, |
|
"loss": 4.8445, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 11.32, |
|
"learning_rate": 1.999307723208382e-05, |
|
"loss": 4.5783, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 12.26, |
|
"learning_rate": 1.9975410248472147e-05, |
|
"loss": 4.458, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 13.21, |
|
"learning_rate": 1.9946926571599945e-05, |
|
"loss": 4.4055, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 14.15, |
|
"learning_rate": 1.9907657087334364e-05, |
|
"loss": 4.3862, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 15.09, |
|
"learning_rate": 1.9857644376978347e-05, |
|
"loss": 4.3751, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 16.04, |
|
"learning_rate": 1.9796942671098212e-05, |
|
"loss": 4.3656, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 16.98, |
|
"learning_rate": 1.9725617790719523e-05, |
|
"loss": 4.3571, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 17.92, |
|
"learning_rate": 1.9643747075954974e-05, |
|
"loss": 4.3475, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 18.87, |
|
"learning_rate": 1.9551419302141784e-05, |
|
"loss": 4.3313, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 19.81, |
|
"learning_rate": 1.9448734583579404e-05, |
|
"loss": 4.3083, |
|
"step": 2100 |
|
}, |
|
{ |
|
"epoch": 20.75, |
|
"learning_rate": 1.9335804264972018e-05, |
|
"loss": 4.2959, |
|
"step": 2200 |
|
}, |
|
{ |
|
"epoch": 21.7, |
|
"learning_rate": 1.9212750800693506e-05, |
|
"loss": 4.2644, |
|
"step": 2300 |
|
}, |
|
{ |
|
"epoch": 22.64, |
|
"learning_rate": 1.90797076220058e-05, |
|
"loss": 4.2342, |
|
"step": 2400 |
|
}, |
|
{ |
|
"epoch": 23.58, |
|
"learning_rate": 1.8936818992374606e-05, |
|
"loss": 4.1978, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 24.53, |
|
"learning_rate": 1.8784239851039375e-05, |
|
"loss": 4.1374, |
|
"step": 2600 |
|
}, |
|
{ |
|
"epoch": 25.47, |
|
"learning_rate": 1.8622135645007168e-05, |
|
"loss": 3.9802, |
|
"step": 2700 |
|
}, |
|
{ |
|
"epoch": 26.42, |
|
"learning_rate": 1.845068214965255e-05, |
|
"loss": 3.6945, |
|
"step": 2800 |
|
}, |
|
{ |
|
"epoch": 27.36, |
|
"learning_rate": 1.8270065278118092e-05, |
|
"loss": 3.3409, |
|
"step": 2900 |
|
}, |
|
{ |
|
"epoch": 28.3, |
|
"learning_rate": 1.8080480879722103e-05, |
|
"loss": 2.8779, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 29.25, |
|
"learning_rate": 1.7882134527592244e-05, |
|
"loss": 2.4274, |
|
"step": 3100 |
|
}, |
|
{ |
|
"epoch": 30.19, |
|
"learning_rate": 1.7679461483820617e-05, |
|
"loss": 2.1192, |
|
"step": 3200 |
|
}, |
|
{ |
|
"epoch": 31.13, |
|
"learning_rate": 1.7464409907419976e-05, |
|
"loss": 1.8713, |
|
"step": 3300 |
|
}, |
|
{ |
|
"epoch": 32.08, |
|
"learning_rate": 1.7241264405022477e-05, |
|
"loss": 1.6762, |
|
"step": 3400 |
|
}, |
|
{ |
|
"epoch": 33.02, |
|
"learning_rate": 1.701026694125833e-05, |
|
"loss": 1.5089, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 33.96, |
|
"learning_rate": 1.6774090768906323e-05, |
|
"loss": 1.3725, |
|
"step": 3600 |
|
}, |
|
{ |
|
"epoch": 34.91, |
|
"learning_rate": 1.6528221181905217e-05, |
|
"loss": 1.2504, |
|
"step": 3700 |
|
}, |
|
{ |
|
"epoch": 35.85, |
|
"learning_rate": 1.6275272811746063e-05, |
|
"loss": 1.1491, |
|
"step": 3800 |
|
}, |
|
{ |
|
"epoch": 36.79, |
|
"learning_rate": 1.6018150231520486e-05, |
|
"loss": 1.0699, |
|
"step": 3900 |
|
}, |
|
{ |
|
"epoch": 37.74, |
|
"learning_rate": 1.575193832697346e-05, |
|
"loss": 0.9869, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 38.68, |
|
"learning_rate": 1.5479489390466186e-05, |
|
"loss": 0.9305, |
|
"step": 4100 |
|
}, |
|
{ |
|
"epoch": 39.62, |
|
"learning_rate": 1.520109884810164e-05, |
|
"loss": 0.8711, |
|
"step": 4200 |
|
}, |
|
{ |
|
"epoch": 40.57, |
|
"learning_rate": 1.4917068568678037e-05, |
|
"loss": 0.8215, |
|
"step": 4300 |
|
}, |
|
{ |
|
"epoch": 41.51, |
|
"learning_rate": 1.4627706536361717e-05, |
|
"loss": 0.7728, |
|
"step": 4400 |
|
}, |
|
{ |
|
"epoch": 42.45, |
|
"learning_rate": 1.4333326516728909e-05, |
|
"loss": 0.7473, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 43.4, |
|
"learning_rate": 1.4034247716538545e-05, |
|
"loss": 0.7033, |
|
"step": 4600 |
|
}, |
|
{ |
|
"epoch": 44.34, |
|
"learning_rate": 1.373079443760497e-05, |
|
"loss": 0.6874, |
|
"step": 4700 |
|
}, |
|
{ |
|
"epoch": 45.28, |
|
"learning_rate": 1.3423295725145908e-05, |
|
"loss": 0.6774, |
|
"step": 4800 |
|
}, |
|
{ |
|
"epoch": 46.23, |
|
"learning_rate": 1.3112085010987002e-05, |
|
"loss": 0.6289, |
|
"step": 4900 |
|
}, |
|
{ |
|
"epoch": 47.17, |
|
"learning_rate": 1.279749975200978e-05, |
|
"loss": 0.6088, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 48.11, |
|
"learning_rate": 1.247988106423513e-05, |
|
"loss": 0.5807, |
|
"step": 5100 |
|
}, |
|
{ |
|
"epoch": 49.06, |
|
"learning_rate": 1.2159573352939014e-05, |
|
"loss": 0.5688, |
|
"step": 5200 |
|
}, |
|
{ |
|
"epoch": 50.0, |
|
"learning_rate": 1.1836923939201543e-05, |
|
"loss": 0.5522, |
|
"step": 5300 |
|
}, |
|
{ |
|
"epoch": 50.94, |
|
"learning_rate": 1.1512282683294342e-05, |
|
"loss": 0.5322, |
|
"step": 5400 |
|
}, |
|
{ |
|
"epoch": 51.89, |
|
"learning_rate": 1.1186001605314571e-05, |
|
"loss": 0.5294, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 52.83, |
|
"learning_rate": 1.0861715375003623e-05, |
|
"loss": 0.5092, |
|
"step": 5600 |
|
}, |
|
{ |
|
"epoch": 53.77, |
|
"learning_rate": 1.0533224988478176e-05, |
|
"loss": 0.5154, |
|
"step": 5700 |
|
}, |
|
{ |
|
"epoch": 54.72, |
|
"learning_rate": 1.0204156407071964e-05, |
|
"loss": 0.5496, |
|
"step": 5800 |
|
}, |
|
{ |
|
"epoch": 55.66, |
|
"learning_rate": 9.874866451589151e-06, |
|
"loss": 0.501, |
|
"step": 5900 |
|
}, |
|
{ |
|
"epoch": 56.6, |
|
"learning_rate": 9.545712182877714e-06, |
|
"loss": 0.489, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 57.55, |
|
"learning_rate": 9.21705051465562e-06, |
|
"loss": 0.4805, |
|
"step": 6100 |
|
}, |
|
{ |
|
"epoch": 58.49, |
|
"learning_rate": 8.889237826496551e-06, |
|
"loss": 0.4786, |
|
"step": 6200 |
|
}, |
|
{ |
|
"epoch": 59.43, |
|
"learning_rate": 8.562629577394817e-06, |
|
"loss": 0.4576, |
|
"step": 6300 |
|
}, |
|
{ |
|
"epoch": 60.38, |
|
"learning_rate": 8.237579920328478e-06, |
|
"loss": 0.4606, |
|
"step": 6400 |
|
}, |
|
{ |
|
"epoch": 61.32, |
|
"learning_rate": 7.91444131823864e-06, |
|
"loss": 0.438, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 62.26, |
|
"learning_rate": 7.593564161841318e-06, |
|
"loss": 0.4471, |
|
"step": 6600 |
|
}, |
|
{ |
|
"epoch": 63.21, |
|
"learning_rate": 7.275296389686258e-06, |
|
"loss": 0.4348, |
|
"step": 6700 |
|
}, |
|
{ |
|
"epoch": 64.15, |
|
"learning_rate": 6.959983110874782e-06, |
|
"loss": 0.429, |
|
"step": 6800 |
|
}, |
|
{ |
|
"epoch": 65.09, |
|
"learning_rate": 6.647966230845655e-06, |
|
"loss": 0.4101, |
|
"step": 6900 |
|
}, |
|
{ |
|
"epoch": 66.04, |
|
"learning_rate": 6.339584080634824e-06, |
|
"loss": 0.4191, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 66.98, |
|
"learning_rate": 6.03517105001101e-06, |
|
"loss": 0.422, |
|
"step": 7100 |
|
}, |
|
{ |
|
"epoch": 67.92, |
|
"learning_rate": 5.735057224884939e-06, |
|
"loss": 0.418, |
|
"step": 7200 |
|
}, |
|
{ |
|
"epoch": 68.87, |
|
"learning_rate": 5.439568029385422e-06, |
|
"loss": 0.3971, |
|
"step": 7300 |
|
}, |
|
{ |
|
"epoch": 69.81, |
|
"learning_rate": 5.14902387299036e-06, |
|
"loss": 0.3938, |
|
"step": 7400 |
|
}, |
|
{ |
|
"epoch": 70.75, |
|
"learning_rate": 4.863739803095299e-06, |
|
"loss": 0.387, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 71.7, |
|
"learning_rate": 4.584025163396323e-06, |
|
"loss": 0.3915, |
|
"step": 7600 |
|
}, |
|
{ |
|
"epoch": 72.64, |
|
"learning_rate": 4.310183258457632e-06, |
|
"loss": 0.3865, |
|
"step": 7700 |
|
}, |
|
{ |
|
"epoch": 73.58, |
|
"learning_rate": 4.042511024827573e-06, |
|
"loss": 0.3771, |
|
"step": 7800 |
|
}, |
|
{ |
|
"epoch": 74.53, |
|
"learning_rate": 3.7838779196384036e-06, |
|
"loss": 0.3776, |
|
"step": 7900 |
|
}, |
|
{ |
|
"epoch": 75.47, |
|
"learning_rate": 3.5293399536806903e-06, |
|
"loss": 0.3716, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 76.42, |
|
"learning_rate": 3.2818183552879e-06, |
|
"loss": 0.3717, |
|
"step": 8100 |
|
}, |
|
{ |
|
"epoch": 77.36, |
|
"learning_rate": 3.041581520960881e-06, |
|
"loss": 0.3701, |
|
"step": 8200 |
|
}, |
|
{ |
|
"epoch": 78.3, |
|
"learning_rate": 2.8088899480708677e-06, |
|
"loss": 0.369, |
|
"step": 8300 |
|
}, |
|
{ |
|
"epoch": 79.25, |
|
"learning_rate": 2.5839959523929126e-06, |
|
"loss": 0.3556, |
|
"step": 8400 |
|
}, |
|
{ |
|
"epoch": 80.19, |
|
"learning_rate": 2.3671433945108712e-06, |
|
"loss": 0.3618, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 81.13, |
|
"learning_rate": 2.15856741539067e-06, |
|
"loss": 0.367, |
|
"step": 8600 |
|
}, |
|
{ |
|
"epoch": 82.08, |
|
"learning_rate": 1.9604521050536506e-06, |
|
"loss": 0.3557, |
|
"step": 8700 |
|
}, |
|
{ |
|
"epoch": 83.02, |
|
"learning_rate": 1.7690103228094845e-06, |
|
"loss": 0.3637, |
|
"step": 8800 |
|
}, |
|
{ |
|
"epoch": 83.96, |
|
"learning_rate": 1.5864936963585164e-06, |
|
"loss": 0.3524, |
|
"step": 8900 |
|
}, |
|
{ |
|
"epoch": 84.91, |
|
"learning_rate": 1.413100134991111e-06, |
|
"loss": 0.3581, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 85.85, |
|
"learning_rate": 1.2490176555326306e-06, |
|
"loss": 0.3546, |
|
"step": 9100 |
|
}, |
|
{ |
|
"epoch": 86.79, |
|
"learning_rate": 1.0944241784701025e-06, |
|
"loss": 0.352, |
|
"step": 9200 |
|
}, |
|
{ |
|
"epoch": 87.74, |
|
"learning_rate": 9.49487335026692e-07, |
|
"loss": 0.3451, |
|
"step": 9300 |
|
}, |
|
{ |
|
"epoch": 88.68, |
|
"learning_rate": 8.143642853931688e-07, |
|
"loss": 0.3632, |
|
"step": 9400 |
|
}, |
|
{ |
|
"epoch": 89.62, |
|
"learning_rate": 6.892015483135095e-07, |
|
"loss": 0.3469, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 90.57, |
|
"learning_rate": 5.741348422094128e-07, |
|
"loss": 0.339, |
|
"step": 9600 |
|
}, |
|
{ |
|
"epoch": 91.51, |
|
"learning_rate": 4.69288938015946e-07, |
|
"loss": 0.3671, |
|
"step": 9700 |
|
}, |
|
{ |
|
"epoch": 92.45, |
|
"learning_rate": 3.747775238879847e-07, |
|
"loss": 0.3441, |
|
"step": 9800 |
|
}, |
|
{ |
|
"epoch": 93.4, |
|
"learning_rate": 2.907030819240908e-07, |
|
"loss": 0.3582, |
|
"step": 9900 |
|
}, |
|
{ |
|
"epoch": 94.34, |
|
"learning_rate": 2.1715677704150395e-07, |
|
"loss": 0.3402, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 95.28, |
|
"learning_rate": 1.5421835812278408e-07, |
|
"loss": 0.3499, |
|
"step": 10100 |
|
}, |
|
{ |
|
"epoch": 96.23, |
|
"learning_rate": 1.0195607154123932e-07, |
|
"loss": 0.3519, |
|
"step": 10200 |
|
}, |
|
{ |
|
"epoch": 97.17, |
|
"learning_rate": 6.042658715895932e-08, |
|
"loss": 0.348, |
|
"step": 10300 |
|
}, |
|
{ |
|
"epoch": 98.11, |
|
"learning_rate": 2.967493687765277e-08, |
|
"loss": 0.3553, |
|
"step": 10400 |
|
}, |
|
{ |
|
"epoch": 99.06, |
|
"learning_rate": 9.734465808957095e-09, |
|
"loss": 0.3431, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 100.0, |
|
"learning_rate": 6.267961171368963e-10, |
|
"loss": 0.3459, |
|
"step": 10600 |
|
}, |
|
{ |
|
"epoch": 100.0, |
|
"step": 10600, |
|
"total_flos": 6.514875776263782e+16, |
|
"train_loss": 4.663586199778431, |
|
"train_runtime": 6099.7809, |
|
"train_samples_per_second": 55.477, |
|
"train_steps_per_second": 1.738 |
|
} |
|
], |
|
"logging_steps": 100, |
|
"max_steps": 10600, |
|
"num_train_epochs": 100, |
|
"save_steps": 100, |
|
"total_flos": 6.514875776263782e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|