|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 2.0, |
|
"eval_steps": 1000, |
|
"global_step": 1922, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0, |
|
"learning_rate": 3.0927835051546394e-07, |
|
"loss": 0.5792, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 3.092783505154639e-06, |
|
"loss": 0.7297, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.02, |
|
"learning_rate": 6.185567010309278e-06, |
|
"loss": 0.7078, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 9.278350515463918e-06, |
|
"loss": 0.6125, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 1.2371134020618556e-05, |
|
"loss": 0.5099, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.05, |
|
"learning_rate": 1.5463917525773194e-05, |
|
"loss": 0.4724, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 1.8556701030927837e-05, |
|
"loss": 0.4138, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 2.1649484536082473e-05, |
|
"loss": 0.4252, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 2.4742268041237112e-05, |
|
"loss": 0.3907, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 2.7835051546391755e-05, |
|
"loss": 0.3657, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 2.999979997843754e-05, |
|
"loss": 0.4024, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 2.999624418794702e-05, |
|
"loss": 0.3996, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 2.998824468665815e-05, |
|
"loss": 0.3838, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 2.997580384499581e-05, |
|
"loss": 0.3463, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 0.15, |
|
"learning_rate": 2.9958925349449884e-05, |
|
"loss": 0.3673, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 2.9937614201482864e-05, |
|
"loss": 0.3746, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 2.9911876716047828e-05, |
|
"loss": 0.3571, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 0.18, |
|
"learning_rate": 2.988172051971717e-05, |
|
"loss": 0.3634, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 2.9847154548422685e-05, |
|
"loss": 0.3422, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 2.9808189044807664e-05, |
|
"loss": 0.346, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 2.976483555519177e-05, |
|
"loss": 0.3678, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 0.22, |
|
"learning_rate": 2.9717106926149626e-05, |
|
"loss": 0.3453, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 2.9665017300704087e-05, |
|
"loss": 0.3406, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.9608582114135352e-05, |
|
"loss": 0.333, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 0.25, |
|
"learning_rate": 2.954781808940717e-05, |
|
"loss": 0.367, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 2.9482743232211473e-05, |
|
"loss": 0.3626, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 2.9413376825632873e-05, |
|
"loss": 0.3564, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.9339739424434687e-05, |
|
"loss": 0.3645, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 2.9261852848968123e-05, |
|
"loss": 0.3692, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 2.9179740178706412e-05, |
|
"loss": 0.371, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 2.9093425745405897e-05, |
|
"loss": 0.3524, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.9002935125895976e-05, |
|
"loss": 0.3338, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 2.890829513450016e-05, |
|
"loss": 0.3698, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 2.8809533815090403e-05, |
|
"loss": 0.3597, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 0.35, |
|
"learning_rate": 2.8706680432777084e-05, |
|
"loss": 0.3626, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 2.859976546523714e-05, |
|
"loss": 0.3353, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 2.848882059368287e-05, |
|
"loss": 0.3644, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 2.8373878693474103e-05, |
|
"loss": 0.3462, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 2.825497382437652e-05, |
|
"loss": 0.357, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 2.8132141220469005e-05, |
|
"loss": 0.3381, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 0.42, |
|
"learning_rate": 2.8005417279703043e-05, |
|
"loss": 0.3552, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 2.787483955311722e-05, |
|
"loss": 0.3445, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 2.774044673371001e-05, |
|
"loss": 0.329, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 0.45, |
|
"learning_rate": 2.7602278644974242e-05, |
|
"loss": 0.3453, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 2.7460376229096517e-05, |
|
"loss": 0.3499, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 2.7314781534825174e-05, |
|
"loss": 0.3651, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 0.48, |
|
"learning_rate": 2.7165537705010303e-05, |
|
"loss": 0.3457, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 2.701268896381963e-05, |
|
"loss": 0.3405, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 2.685628060363392e-05, |
|
"loss": 0.3496, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 2.6696358971625862e-05, |
|
"loss": 0.3612, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 0.52, |
|
"learning_rate": 2.653297145602641e-05, |
|
"loss": 0.3231, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 2.6366166472082628e-05, |
|
"loss": 0.3293, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 2.6195993447711203e-05, |
|
"loss": 0.3475, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 0.55, |
|
"learning_rate": 2.602250280885189e-05, |
|
"loss": 0.3295, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 2.584574596452522e-05, |
|
"loss": 0.3402, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 2.5665775291598897e-05, |
|
"loss": 0.3076, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 0.58, |
|
"learning_rate": 2.5482644119267395e-05, |
|
"loss": 0.3495, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 2.529640671324936e-05, |
|
"loss": 0.3512, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 2.5107118259707533e-05, |
|
"loss": 0.3348, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.4914834848895875e-05, |
|
"loss": 0.3379, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 0.62, |
|
"learning_rate": 2.471961345853883e-05, |
|
"loss": 0.3359, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 2.452151193694758e-05, |
|
"loss": 0.3439, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.4320588985878357e-05, |
|
"loss": 0.3524, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 2.4116904143137835e-05, |
|
"loss": 0.3245, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 2.3910517764940795e-05, |
|
"loss": 0.3308, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 0.68, |
|
"learning_rate": 2.3701491008025273e-05, |
|
"loss": 0.333, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 2.3489885811530526e-05, |
|
"loss": 0.3226, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 2.327576487864307e-05, |
|
"loss": 0.3461, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 2.3059191658016444e-05, |
|
"loss": 0.3427, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 0.72, |
|
"learning_rate": 2.2840230324969947e-05, |
|
"loss": 0.3201, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 2.261894576247213e-05, |
|
"loss": 0.3151, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 2.239540354191458e-05, |
|
"loss": 0.3433, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 0.75, |
|
"learning_rate": 2.2169669903681678e-05, |
|
"loss": 0.326, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 2.1941811737522166e-05, |
|
"loss": 0.3183, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.171189656272826e-05, |
|
"loss": 0.3469, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 0.78, |
|
"learning_rate": 2.1479992508128224e-05, |
|
"loss": 0.3384, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 2.1246168291898337e-05, |
|
"loss": 0.337, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 2.1010493201200207e-05, |
|
"loss": 0.321, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.07730370716495e-05, |
|
"loss": 0.3184, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 0.82, |
|
"learning_rate": 2.0533870266622117e-05, |
|
"loss": 0.3214, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 2.0293063656404028e-05, |
|
"loss": 0.337, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 2.005068859719088e-05, |
|
"loss": 0.3218, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 1.9806816909943594e-05, |
|
"loss": 0.3519, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 1.956152085910629e-05, |
|
"loss": 0.3189, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 1.931487313119275e-05, |
|
"loss": 0.3288, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 0.88, |
|
"learning_rate": 1.9066946813247847e-05, |
|
"loss": 0.3351, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 1.881781537119028e-05, |
|
"loss": 0.3397, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 1.856755262804302e-05, |
|
"loss": 0.3376, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 0.92, |
|
"learning_rate": 1.831623274205799e-05, |
|
"loss": 0.3277, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 1.806393018474131e-05, |
|
"loss": 0.3301, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 1.781071971878587e-05, |
|
"loss": 0.3263, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 0.95, |
|
"learning_rate": 1.7556676375917456e-05, |
|
"loss": 0.3452, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 1.7301875434661255e-05, |
|
"loss": 0.3164, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 1.704639239803523e-05, |
|
"loss": 0.3357, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 0.98, |
|
"learning_rate": 1.67903029711769e-05, |
|
"loss": 0.3285, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 1.6533683038910306e-05, |
|
"loss": 0.3147, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 1.6276608643259695e-05, |
|
"loss": 0.3313, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 1.6019155960916613e-05, |
|
"loss": 0.287, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 1.02, |
|
"learning_rate": 1.5761401280667126e-05, |
|
"loss": 0.3042, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 1.5503420980785788e-05, |
|
"loss": 0.2909, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 1.5245291506403133e-05, |
|
"loss": 0.2966, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"eval_loss": 0.4508744180202484, |
|
"eval_runtime": 13.2913, |
|
"eval_samples_per_second": 3.31, |
|
"eval_steps_per_second": 0.828, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 1.4987089346853305e-05, |
|
"loss": 0.2898, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 1.4728891013008638e-05, |
|
"loss": 0.317, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 1.4470773014607832e-05, |
|
"loss": 0.2907, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 1.08, |
|
"learning_rate": 1.4212811837584448e-05, |
|
"loss": 0.3208, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 1.3955083921402494e-05, |
|
"loss": 0.2908, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 1.3697665636405772e-05, |
|
"loss": 0.3029, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 1.3440633261187661e-05, |
|
"loss": 0.305, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 1.12, |
|
"learning_rate": 1.318406295998815e-05, |
|
"loss": 0.2802, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 1.2928030760124725e-05, |
|
"loss": 0.2833, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 1.2672612529463828e-05, |
|
"loss": 0.2886, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 1.241788395393959e-05, |
|
"loss": 0.2751, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 1.216392051512643e-05, |
|
"loss": 0.2883, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 1.18, |
|
"learning_rate": 1.1910797467872246e-05, |
|
"loss": 0.3079, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 1.1658589817998781e-05, |
|
"loss": 0.2702, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 1.1407372300075779e-05, |
|
"loss": 0.304, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 1.1157219355275524e-05, |
|
"loss": 0.3173, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 1.22, |
|
"learning_rate": 1.0908205109314302e-05, |
|
"loss": 0.3235, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 1.0660403350487352e-05, |
|
"loss": 0.2814, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 1.0413887507803817e-05, |
|
"loss": 0.2899, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 1.0168730629228135e-05, |
|
"loss": 0.3036, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 9.925005360034349e-06, |
|
"loss": 0.2942, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 9.682783921279739e-06, |
|
"loss": 0.3041, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 1.28, |
|
"learning_rate": 9.442138088404174e-06, |
|
"loss": 0.2923, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 9.203139169961496e-06, |
|
"loss": 0.2891, |
|
"step": 1240 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 8.96585798648926e-06, |
|
"loss": 0.2841, |
|
"step": 1250 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 8.730364849523085e-06, |
|
"loss": 0.3046, |
|
"step": 1260 |
|
}, |
|
{ |
|
"epoch": 1.32, |
|
"learning_rate": 8.496729540761802e-06, |
|
"loss": 0.3172, |
|
"step": 1270 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 8.265021291389665e-06, |
|
"loss": 0.2828, |
|
"step": 1280 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 8.035308761561632e-06, |
|
"loss": 0.2625, |
|
"step": 1290 |
|
}, |
|
{ |
|
"epoch": 1.35, |
|
"learning_rate": 7.80766002005789e-06, |
|
"loss": 0.2955, |
|
"step": 1300 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 7.582142524113619e-06, |
|
"loss": 0.2769, |
|
"step": 1310 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 7.358823099429932e-06, |
|
"loss": 0.2912, |
|
"step": 1320 |
|
}, |
|
{ |
|
"epoch": 1.38, |
|
"learning_rate": 7.1377679203719866e-06, |
|
"loss": 0.3017, |
|
"step": 1330 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 6.9190424903601e-06, |
|
"loss": 0.283, |
|
"step": 1340 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 6.7027116224596605e-06, |
|
"loss": 0.2855, |
|
"step": 1350 |
|
}, |
|
{ |
|
"epoch": 1.42, |
|
"learning_rate": 6.488839420175642e-06, |
|
"loss": 0.2947, |
|
"step": 1360 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 6.277489258457314e-06, |
|
"loss": 0.2912, |
|
"step": 1370 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 6.068723764918906e-06, |
|
"loss": 0.2863, |
|
"step": 1380 |
|
}, |
|
{ |
|
"epoch": 1.45, |
|
"learning_rate": 5.862604801281679e-06, |
|
"loss": 0.3121, |
|
"step": 1390 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 5.6591934450429655e-06, |
|
"loss": 0.286, |
|
"step": 1400 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 5.458549971377608e-06, |
|
"loss": 0.2897, |
|
"step": 1410 |
|
}, |
|
{ |
|
"epoch": 1.48, |
|
"learning_rate": 5.2607338352770885e-06, |
|
"loss": 0.2831, |
|
"step": 1420 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 5.065803653931754e-06, |
|
"loss": 0.3055, |
|
"step": 1430 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 4.873817189361276e-06, |
|
"loss": 0.2833, |
|
"step": 1440 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 4.6848313312985095e-06, |
|
"loss": 0.2812, |
|
"step": 1450 |
|
}, |
|
{ |
|
"epoch": 1.52, |
|
"learning_rate": 4.498902080331857e-06, |
|
"loss": 0.2604, |
|
"step": 1460 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 4.316084531311057e-06, |
|
"loss": 0.2989, |
|
"step": 1470 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 4.136432857021409e-06, |
|
"loss": 0.2991, |
|
"step": 1480 |
|
}, |
|
{ |
|
"epoch": 1.55, |
|
"learning_rate": 3.960000292131206e-06, |
|
"loss": 0.2907, |
|
"step": 1490 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 3.78683911741715e-06, |
|
"loss": 0.2992, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 3.6170006442724368e-06, |
|
"loss": 0.2926, |
|
"step": 1510 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 3.4505351995020747e-06, |
|
"loss": 0.2971, |
|
"step": 1520 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 3.287492110409963e-06, |
|
"loss": 0.2862, |
|
"step": 1530 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 3.127919690182161e-06, |
|
"loss": 0.3023, |
|
"step": 1540 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 2.9718652235706305e-06, |
|
"loss": 0.2974, |
|
"step": 1550 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 2.819374952881769e-06, |
|
"loss": 0.289, |
|
"step": 1560 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 2.6704940642737883e-06, |
|
"loss": 0.2859, |
|
"step": 1570 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 2.5252666743670937e-06, |
|
"loss": 0.3084, |
|
"step": 1580 |
|
}, |
|
{ |
|
"epoch": 1.65, |
|
"learning_rate": 2.3837358171715804e-06, |
|
"loss": 0.2858, |
|
"step": 1590 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 2.2459434313347153e-06, |
|
"loss": 0.3031, |
|
"step": 1600 |
|
}, |
|
{ |
|
"epoch": 1.68, |
|
"learning_rate": 2.111930347714235e-06, |
|
"loss": 0.3045, |
|
"step": 1610 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 1.9817362772790506e-06, |
|
"loss": 0.2676, |
|
"step": 1620 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 1.8553997993420495e-06, |
|
"loss": 0.2993, |
|
"step": 1630 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 1.7329583501282093e-06, |
|
"loss": 0.2781, |
|
"step": 1640 |
|
}, |
|
{ |
|
"epoch": 1.72, |
|
"learning_rate": 1.614448211681432e-06, |
|
"loss": 0.2943, |
|
"step": 1650 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 1.4999045011134043e-06, |
|
"loss": 0.2783, |
|
"step": 1660 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 1.3893611601976203e-06, |
|
"loss": 0.2784, |
|
"step": 1670 |
|
}, |
|
{ |
|
"epoch": 1.75, |
|
"learning_rate": 1.2828509453117133e-06, |
|
"loss": 0.298, |
|
"step": 1680 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 1.180405417731033e-06, |
|
"loss": 0.2977, |
|
"step": 1690 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 1.0820549342763553e-06, |
|
"loss": 0.3138, |
|
"step": 1700 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 9.878286383185081e-07, |
|
"loss": 0.3143, |
|
"step": 1710 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 8.977544511425539e-07, |
|
"loss": 0.2995, |
|
"step": 1720 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 8.118590636741108e-07, |
|
"loss": 0.2728, |
|
"step": 1730 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 7.301679285702589e-07, |
|
"loss": 0.2746, |
|
"step": 1740 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 6.527052526773742e-07, |
|
"loss": 0.3014, |
|
"step": 1750 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 5.794939898581003e-07, |
|
"loss": 0.2765, |
|
"step": 1760 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 5.105558341896404e-07, |
|
"loss": 0.2985, |
|
"step": 1770 |
|
}, |
|
{ |
|
"epoch": 1.85, |
|
"learning_rate": 4.4591121353532546e-07, |
|
"loss": 0.292, |
|
"step": 1780 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 3.855792834914013e-07, |
|
"loss": 0.2894, |
|
"step": 1790 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 3.295779217108108e-07, |
|
"loss": 0.2815, |
|
"step": 1800 |
|
}, |
|
{ |
|
"epoch": 1.88, |
|
"learning_rate": 2.779237226056569e-07, |
|
"loss": 0.2878, |
|
"step": 1810 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 2.3063199242992215e-07, |
|
"loss": 0.317, |
|
"step": 1820 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 1.8771674474389878e-07, |
|
"loss": 0.2945, |
|
"step": 1830 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 1.4919069626166016e-07, |
|
"loss": 0.2817, |
|
"step": 1840 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 1.1506526308283528e-07, |
|
"loss": 0.3051, |
|
"step": 1850 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 8.535055730976127e-08, |
|
"loss": 0.2636, |
|
"step": 1860 |
|
}, |
|
{ |
|
"epoch": 1.95, |
|
"learning_rate": 6.005538405105371e-08, |
|
"loss": 0.3053, |
|
"step": 1870 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 3.918723881246367e-08, |
|
"loss": 0.2882, |
|
"step": 1880 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 2.2752305275797748e-08, |
|
"loss": 0.272, |
|
"step": 1890 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 1.0755453466555486e-08, |
|
"loss": 0.3028, |
|
"step": 1900 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 3.2002383108437063e-09, |
|
"loss": 0.2992, |
|
"step": 1910 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 8.889858196747813e-11, |
|
"loss": 0.2968, |
|
"step": 1920 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"step": 1922, |
|
"total_flos": 3.4599459174219776e+17, |
|
"train_loss": 0.32601516778337597, |
|
"train_runtime": 7188.9024, |
|
"train_samples_per_second": 1.069, |
|
"train_steps_per_second": 0.267 |
|
} |
|
], |
|
"logging_steps": 10, |
|
"max_steps": 1922, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 2, |
|
"save_steps": 1000, |
|
"total_flos": 3.4599459174219776e+17, |
|
"train_batch_size": 4, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|