|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 9.944416371905003, |
|
"eval_steps": 500, |
|
"global_step": 1230, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.0540540540540545e-06, |
|
"loss": 2998113.6, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.08, |
|
"learning_rate": 8.108108108108109e-06, |
|
"loss": 2797824.6, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.12, |
|
"learning_rate": 1.2162162162162164e-05, |
|
"loss": 2452371.6, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 1.6216216216216218e-05, |
|
"loss": 1871456.8, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 2.027027027027027e-05, |
|
"loss": 1119156.9, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 2.4324324324324327e-05, |
|
"loss": 420413.4, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.28, |
|
"learning_rate": 2.8378378378378378e-05, |
|
"loss": 36048.9469, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.32, |
|
"learning_rate": 2.9924559932942165e-05, |
|
"loss": 866.2308, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 2.9798826487845768e-05, |
|
"loss": 6.5443, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 2.9673093042749373e-05, |
|
"loss": 0.5418, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 2.9547359597652975e-05, |
|
"loss": 0.4141, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 2.942162615255658e-05, |
|
"loss": 0.3744, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 2.9295892707460186e-05, |
|
"loss": 0.3621, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 2.917015926236379e-05, |
|
"loss": 0.3423, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 2.9044425817267394e-05, |
|
"loss": 0.3333, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.65, |
|
"learning_rate": 2.8918692372171e-05, |
|
"loss": 0.3359, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 2.87929589270746e-05, |
|
"loss": 0.3301, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 2.8667225481978207e-05, |
|
"loss": 0.3238, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 2.854149203688181e-05, |
|
"loss": 0.3221, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 2.8415758591785418e-05, |
|
"loss": 0.3132, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.85, |
|
"learning_rate": 2.829002514668902e-05, |
|
"loss": 0.3141, |
|
"step": 105 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 2.8164291701592623e-05, |
|
"loss": 0.3019, |
|
"step": 110 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 2.8038558256496228e-05, |
|
"loss": 0.324, |
|
"step": 115 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 2.7912824811399834e-05, |
|
"loss": 0.3118, |
|
"step": 120 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"eval_f1": 0.3004154682007031, |
|
"eval_loss": 0.2937778830528259, |
|
"eval_runtime": 30.7677, |
|
"eval_samples_per_second": 32.144, |
|
"eval_steps_per_second": 8.06, |
|
"step": 123 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 2.778709136630344e-05, |
|
"loss": 0.3159, |
|
"step": 125 |
|
}, |
|
{ |
|
"epoch": 1.05, |
|
"learning_rate": 2.766135792120704e-05, |
|
"loss": 0.2949, |
|
"step": 130 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 2.7535624476110643e-05, |
|
"loss": 0.3043, |
|
"step": 135 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 2.7409891031014252e-05, |
|
"loss": 0.3076, |
|
"step": 140 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 2.7284157585917854e-05, |
|
"loss": 0.301, |
|
"step": 145 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 2.715842414082146e-05, |
|
"loss": 0.3006, |
|
"step": 150 |
|
}, |
|
{ |
|
"epoch": 1.25, |
|
"learning_rate": 2.7032690695725062e-05, |
|
"loss": 0.2929, |
|
"step": 155 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 2.6906957250628668e-05, |
|
"loss": 0.3107, |
|
"step": 160 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 2.6781223805532273e-05, |
|
"loss": 0.2988, |
|
"step": 165 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 2.6655490360435875e-05, |
|
"loss": 0.3097, |
|
"step": 170 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 2.6529756915339484e-05, |
|
"loss": 0.2967, |
|
"step": 175 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 2.6404023470243086e-05, |
|
"loss": 0.2918, |
|
"step": 180 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 2.627829002514669e-05, |
|
"loss": 0.287, |
|
"step": 185 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 2.6152556580050294e-05, |
|
"loss": 0.3104, |
|
"step": 190 |
|
}, |
|
{ |
|
"epoch": 1.58, |
|
"learning_rate": 2.60268231349539e-05, |
|
"loss": 0.2859, |
|
"step": 195 |
|
}, |
|
{ |
|
"epoch": 1.62, |
|
"learning_rate": 2.59010896898575e-05, |
|
"loss": 0.2938, |
|
"step": 200 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 2.5775356244761107e-05, |
|
"loss": 0.2884, |
|
"step": 205 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 2.564962279966471e-05, |
|
"loss": 0.2858, |
|
"step": 210 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 2.5523889354568318e-05, |
|
"loss": 0.2956, |
|
"step": 215 |
|
}, |
|
{ |
|
"epoch": 1.78, |
|
"learning_rate": 2.539815590947192e-05, |
|
"loss": 0.2916, |
|
"step": 220 |
|
}, |
|
{ |
|
"epoch": 1.82, |
|
"learning_rate": 2.5272422464375523e-05, |
|
"loss": 0.2956, |
|
"step": 225 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 2.5146689019279128e-05, |
|
"loss": 0.2905, |
|
"step": 230 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 2.5020955574182734e-05, |
|
"loss": 0.2843, |
|
"step": 235 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 2.489522212908634e-05, |
|
"loss": 0.2878, |
|
"step": 240 |
|
}, |
|
{ |
|
"epoch": 1.98, |
|
"learning_rate": 2.476948868398994e-05, |
|
"loss": 0.2732, |
|
"step": 245 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_f1": 0.46014293567894443, |
|
"eval_loss": 0.27736958861351013, |
|
"eval_runtime": 34.3798, |
|
"eval_samples_per_second": 28.767, |
|
"eval_steps_per_second": 7.214, |
|
"step": 247 |
|
}, |
|
{ |
|
"epoch": 2.02, |
|
"learning_rate": 2.4643755238893543e-05, |
|
"loss": 0.2978, |
|
"step": 250 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 2.4518021793797152e-05, |
|
"loss": 0.287, |
|
"step": 255 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 2.4392288348700754e-05, |
|
"loss": 0.2841, |
|
"step": 260 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 2.426655490360436e-05, |
|
"loss": 0.2755, |
|
"step": 265 |
|
}, |
|
{ |
|
"epoch": 2.18, |
|
"learning_rate": 2.4140821458507962e-05, |
|
"loss": 0.2786, |
|
"step": 270 |
|
}, |
|
{ |
|
"epoch": 2.22, |
|
"learning_rate": 2.4015088013411568e-05, |
|
"loss": 0.2837, |
|
"step": 275 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 2.3889354568315173e-05, |
|
"loss": 0.2691, |
|
"step": 280 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 2.3763621123218775e-05, |
|
"loss": 0.2785, |
|
"step": 285 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 2.363788767812238e-05, |
|
"loss": 0.2863, |
|
"step": 290 |
|
}, |
|
{ |
|
"epoch": 2.39, |
|
"learning_rate": 2.3512154233025986e-05, |
|
"loss": 0.277, |
|
"step": 295 |
|
}, |
|
{ |
|
"epoch": 2.43, |
|
"learning_rate": 2.338642078792959e-05, |
|
"loss": 0.2755, |
|
"step": 300 |
|
}, |
|
{ |
|
"epoch": 2.47, |
|
"learning_rate": 2.3260687342833194e-05, |
|
"loss": 0.2726, |
|
"step": 305 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 2.3134953897736796e-05, |
|
"loss": 0.2732, |
|
"step": 310 |
|
}, |
|
{ |
|
"epoch": 2.55, |
|
"learning_rate": 2.3009220452640405e-05, |
|
"loss": 0.2772, |
|
"step": 315 |
|
}, |
|
{ |
|
"epoch": 2.59, |
|
"learning_rate": 2.2883487007544007e-05, |
|
"loss": 0.2853, |
|
"step": 320 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"learning_rate": 2.275775356244761e-05, |
|
"loss": 0.2803, |
|
"step": 325 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 2.263202011735122e-05, |
|
"loss": 0.2777, |
|
"step": 330 |
|
}, |
|
{ |
|
"epoch": 2.71, |
|
"learning_rate": 2.250628667225482e-05, |
|
"loss": 0.2728, |
|
"step": 335 |
|
}, |
|
{ |
|
"epoch": 2.75, |
|
"learning_rate": 2.2380553227158426e-05, |
|
"loss": 0.2793, |
|
"step": 340 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 2.2254819782062028e-05, |
|
"loss": 0.2574, |
|
"step": 345 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 2.2129086336965634e-05, |
|
"loss": 0.2722, |
|
"step": 350 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 2.200335289186924e-05, |
|
"loss": 0.2847, |
|
"step": 355 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 2.187761944677284e-05, |
|
"loss": 0.2808, |
|
"step": 360 |
|
}, |
|
{ |
|
"epoch": 2.95, |
|
"learning_rate": 2.1751886001676447e-05, |
|
"loss": 0.2837, |
|
"step": 365 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 2.1626152556580052e-05, |
|
"loss": 0.2705, |
|
"step": 370 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"eval_f1": 0.4702956617850236, |
|
"eval_loss": 0.2660216987133026, |
|
"eval_runtime": 30.702, |
|
"eval_samples_per_second": 32.213, |
|
"eval_steps_per_second": 8.078, |
|
"step": 371 |
|
}, |
|
{ |
|
"epoch": 3.03, |
|
"learning_rate": 2.1500419111483655e-05, |
|
"loss": 0.2717, |
|
"step": 375 |
|
}, |
|
{ |
|
"epoch": 3.07, |
|
"learning_rate": 2.137468566638726e-05, |
|
"loss": 0.2599, |
|
"step": 380 |
|
}, |
|
{ |
|
"epoch": 3.11, |
|
"learning_rate": 2.1248952221290862e-05, |
|
"loss": 0.2668, |
|
"step": 385 |
|
}, |
|
{ |
|
"epoch": 3.15, |
|
"learning_rate": 2.112321877619447e-05, |
|
"loss": 0.2689, |
|
"step": 390 |
|
}, |
|
{ |
|
"epoch": 3.19, |
|
"learning_rate": 2.0997485331098073e-05, |
|
"loss": 0.2767, |
|
"step": 395 |
|
}, |
|
{ |
|
"epoch": 3.23, |
|
"learning_rate": 2.0871751886001675e-05, |
|
"loss": 0.2755, |
|
"step": 400 |
|
}, |
|
{ |
|
"epoch": 3.27, |
|
"learning_rate": 2.074601844090528e-05, |
|
"loss": 0.2645, |
|
"step": 405 |
|
}, |
|
{ |
|
"epoch": 3.31, |
|
"learning_rate": 2.0620284995808886e-05, |
|
"loss": 0.2656, |
|
"step": 410 |
|
}, |
|
{ |
|
"epoch": 3.36, |
|
"learning_rate": 2.0494551550712492e-05, |
|
"loss": 0.2691, |
|
"step": 415 |
|
}, |
|
{ |
|
"epoch": 3.4, |
|
"learning_rate": 2.0368818105616094e-05, |
|
"loss": 0.2641, |
|
"step": 420 |
|
}, |
|
{ |
|
"epoch": 3.44, |
|
"learning_rate": 2.0243084660519696e-05, |
|
"loss": 0.2576, |
|
"step": 425 |
|
}, |
|
{ |
|
"epoch": 3.48, |
|
"learning_rate": 2.0117351215423305e-05, |
|
"loss": 0.2577, |
|
"step": 430 |
|
}, |
|
{ |
|
"epoch": 3.52, |
|
"learning_rate": 1.9991617770326907e-05, |
|
"loss": 0.2609, |
|
"step": 435 |
|
}, |
|
{ |
|
"epoch": 3.56, |
|
"learning_rate": 1.986588432523051e-05, |
|
"loss": 0.2558, |
|
"step": 440 |
|
}, |
|
{ |
|
"epoch": 3.6, |
|
"learning_rate": 1.9740150880134115e-05, |
|
"loss": 0.2645, |
|
"step": 445 |
|
}, |
|
{ |
|
"epoch": 3.64, |
|
"learning_rate": 1.961441743503772e-05, |
|
"loss": 0.2698, |
|
"step": 450 |
|
}, |
|
{ |
|
"epoch": 3.68, |
|
"learning_rate": 1.9488683989941326e-05, |
|
"loss": 0.26, |
|
"step": 455 |
|
}, |
|
{ |
|
"epoch": 3.72, |
|
"learning_rate": 1.9362950544844928e-05, |
|
"loss": 0.2641, |
|
"step": 460 |
|
}, |
|
{ |
|
"epoch": 3.76, |
|
"learning_rate": 1.923721709974853e-05, |
|
"loss": 0.2672, |
|
"step": 465 |
|
}, |
|
{ |
|
"epoch": 3.8, |
|
"learning_rate": 1.911148365465214e-05, |
|
"loss": 0.2592, |
|
"step": 470 |
|
}, |
|
{ |
|
"epoch": 3.84, |
|
"learning_rate": 1.898575020955574e-05, |
|
"loss": 0.2662, |
|
"step": 475 |
|
}, |
|
{ |
|
"epoch": 3.88, |
|
"learning_rate": 1.8860016764459347e-05, |
|
"loss": 0.26, |
|
"step": 480 |
|
}, |
|
{ |
|
"epoch": 3.92, |
|
"learning_rate": 1.8734283319362952e-05, |
|
"loss": 0.264, |
|
"step": 485 |
|
}, |
|
{ |
|
"epoch": 3.96, |
|
"learning_rate": 1.8608549874266555e-05, |
|
"loss": 0.2671, |
|
"step": 490 |
|
}, |
|
{ |
|
"epoch": 3.99, |
|
"eval_f1": 0.4754279403644394, |
|
"eval_loss": 0.2679407000541687, |
|
"eval_runtime": 34.3196, |
|
"eval_samples_per_second": 28.817, |
|
"eval_steps_per_second": 7.226, |
|
"step": 494 |
|
}, |
|
{ |
|
"epoch": 4.0, |
|
"learning_rate": 1.848281642917016e-05, |
|
"loss": 0.2603, |
|
"step": 495 |
|
}, |
|
{ |
|
"epoch": 4.04, |
|
"learning_rate": 1.8357082984073762e-05, |
|
"loss": 0.2476, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 4.08, |
|
"learning_rate": 1.823134953897737e-05, |
|
"loss": 0.2623, |
|
"step": 505 |
|
}, |
|
{ |
|
"epoch": 4.12, |
|
"learning_rate": 1.8105616093880973e-05, |
|
"loss": 0.2614, |
|
"step": 510 |
|
}, |
|
{ |
|
"epoch": 4.16, |
|
"learning_rate": 1.7979882648784575e-05, |
|
"loss": 0.2405, |
|
"step": 515 |
|
}, |
|
{ |
|
"epoch": 4.2, |
|
"learning_rate": 1.785414920368818e-05, |
|
"loss": 0.2519, |
|
"step": 520 |
|
}, |
|
{ |
|
"epoch": 4.24, |
|
"learning_rate": 1.7728415758591787e-05, |
|
"loss": 0.2499, |
|
"step": 525 |
|
}, |
|
{ |
|
"epoch": 4.28, |
|
"learning_rate": 1.7602682313495392e-05, |
|
"loss": 0.2678, |
|
"step": 530 |
|
}, |
|
{ |
|
"epoch": 4.33, |
|
"learning_rate": 1.7476948868398994e-05, |
|
"loss": 0.2639, |
|
"step": 535 |
|
}, |
|
{ |
|
"epoch": 4.37, |
|
"learning_rate": 1.7351215423302596e-05, |
|
"loss": 0.2554, |
|
"step": 540 |
|
}, |
|
{ |
|
"epoch": 4.41, |
|
"learning_rate": 1.7225481978206205e-05, |
|
"loss": 0.2615, |
|
"step": 545 |
|
}, |
|
{ |
|
"epoch": 4.45, |
|
"learning_rate": 1.7099748533109807e-05, |
|
"loss": 0.2518, |
|
"step": 550 |
|
}, |
|
{ |
|
"epoch": 4.49, |
|
"learning_rate": 1.6974015088013413e-05, |
|
"loss": 0.2612, |
|
"step": 555 |
|
}, |
|
{ |
|
"epoch": 4.53, |
|
"learning_rate": 1.6848281642917015e-05, |
|
"loss": 0.2608, |
|
"step": 560 |
|
}, |
|
{ |
|
"epoch": 4.57, |
|
"learning_rate": 1.672254819782062e-05, |
|
"loss": 0.2446, |
|
"step": 565 |
|
}, |
|
{ |
|
"epoch": 4.61, |
|
"learning_rate": 1.6596814752724226e-05, |
|
"loss": 0.2556, |
|
"step": 570 |
|
}, |
|
{ |
|
"epoch": 4.65, |
|
"learning_rate": 1.6471081307627828e-05, |
|
"loss": 0.2477, |
|
"step": 575 |
|
}, |
|
{ |
|
"epoch": 4.69, |
|
"learning_rate": 1.6345347862531434e-05, |
|
"loss": 0.2649, |
|
"step": 580 |
|
}, |
|
{ |
|
"epoch": 4.73, |
|
"learning_rate": 1.621961441743504e-05, |
|
"loss": 0.2611, |
|
"step": 585 |
|
}, |
|
{ |
|
"epoch": 4.77, |
|
"learning_rate": 1.609388097233864e-05, |
|
"loss": 0.259, |
|
"step": 590 |
|
}, |
|
{ |
|
"epoch": 4.81, |
|
"learning_rate": 1.5968147527242247e-05, |
|
"loss": 0.2618, |
|
"step": 595 |
|
}, |
|
{ |
|
"epoch": 4.85, |
|
"learning_rate": 1.584241408214585e-05, |
|
"loss": 0.2478, |
|
"step": 600 |
|
}, |
|
{ |
|
"epoch": 4.89, |
|
"learning_rate": 1.5716680637049458e-05, |
|
"loss": 0.2542, |
|
"step": 605 |
|
}, |
|
{ |
|
"epoch": 4.93, |
|
"learning_rate": 1.559094719195306e-05, |
|
"loss": 0.2488, |
|
"step": 610 |
|
}, |
|
{ |
|
"epoch": 4.97, |
|
"learning_rate": 1.5465213746856662e-05, |
|
"loss": 0.2529, |
|
"step": 615 |
|
}, |
|
{ |
|
"epoch": 5.0, |
|
"eval_f1": 0.5162986330178759, |
|
"eval_loss": 0.2654779851436615, |
|
"eval_runtime": 32.5281, |
|
"eval_samples_per_second": 30.405, |
|
"eval_steps_per_second": 7.624, |
|
"step": 618 |
|
}, |
|
{ |
|
"epoch": 5.01, |
|
"learning_rate": 1.533948030176027e-05, |
|
"loss": 0.2474, |
|
"step": 620 |
|
}, |
|
{ |
|
"epoch": 5.05, |
|
"learning_rate": 1.5213746856663873e-05, |
|
"loss": 0.2442, |
|
"step": 625 |
|
}, |
|
{ |
|
"epoch": 5.09, |
|
"learning_rate": 1.5088013411567477e-05, |
|
"loss": 0.2494, |
|
"step": 630 |
|
}, |
|
{ |
|
"epoch": 5.13, |
|
"learning_rate": 1.4962279966471083e-05, |
|
"loss": 0.2505, |
|
"step": 635 |
|
}, |
|
{ |
|
"epoch": 5.17, |
|
"learning_rate": 1.4836546521374687e-05, |
|
"loss": 0.2345, |
|
"step": 640 |
|
}, |
|
{ |
|
"epoch": 5.21, |
|
"learning_rate": 1.471081307627829e-05, |
|
"loss": 0.2491, |
|
"step": 645 |
|
}, |
|
{ |
|
"epoch": 5.26, |
|
"learning_rate": 1.4585079631181894e-05, |
|
"loss": 0.2455, |
|
"step": 650 |
|
}, |
|
{ |
|
"epoch": 5.3, |
|
"learning_rate": 1.44593461860855e-05, |
|
"loss": 0.2483, |
|
"step": 655 |
|
}, |
|
{ |
|
"epoch": 5.34, |
|
"learning_rate": 1.4333612740989104e-05, |
|
"loss": 0.2387, |
|
"step": 660 |
|
}, |
|
{ |
|
"epoch": 5.38, |
|
"learning_rate": 1.4207879295892709e-05, |
|
"loss": 0.2568, |
|
"step": 665 |
|
}, |
|
{ |
|
"epoch": 5.42, |
|
"learning_rate": 1.4082145850796311e-05, |
|
"loss": 0.2539, |
|
"step": 670 |
|
}, |
|
{ |
|
"epoch": 5.46, |
|
"learning_rate": 1.3956412405699917e-05, |
|
"loss": 0.2306, |
|
"step": 675 |
|
}, |
|
{ |
|
"epoch": 5.5, |
|
"learning_rate": 1.383067896060352e-05, |
|
"loss": 0.2453, |
|
"step": 680 |
|
}, |
|
{ |
|
"epoch": 5.54, |
|
"learning_rate": 1.3704945515507126e-05, |
|
"loss": 0.2411, |
|
"step": 685 |
|
}, |
|
{ |
|
"epoch": 5.58, |
|
"learning_rate": 1.357921207041073e-05, |
|
"loss": 0.2526, |
|
"step": 690 |
|
}, |
|
{ |
|
"epoch": 5.62, |
|
"learning_rate": 1.3453478625314334e-05, |
|
"loss": 0.2384, |
|
"step": 695 |
|
}, |
|
{ |
|
"epoch": 5.66, |
|
"learning_rate": 1.3327745180217938e-05, |
|
"loss": 0.2486, |
|
"step": 700 |
|
}, |
|
{ |
|
"epoch": 5.7, |
|
"learning_rate": 1.3202011735121543e-05, |
|
"loss": 0.2357, |
|
"step": 705 |
|
}, |
|
{ |
|
"epoch": 5.74, |
|
"learning_rate": 1.3076278290025147e-05, |
|
"loss": 0.2484, |
|
"step": 710 |
|
}, |
|
{ |
|
"epoch": 5.78, |
|
"learning_rate": 1.295054484492875e-05, |
|
"loss": 0.2405, |
|
"step": 715 |
|
}, |
|
{ |
|
"epoch": 5.82, |
|
"learning_rate": 1.2824811399832355e-05, |
|
"loss": 0.2521, |
|
"step": 720 |
|
}, |
|
{ |
|
"epoch": 5.86, |
|
"learning_rate": 1.269907795473596e-05, |
|
"loss": 0.2575, |
|
"step": 725 |
|
}, |
|
{ |
|
"epoch": 5.9, |
|
"learning_rate": 1.2573344509639564e-05, |
|
"loss": 0.2516, |
|
"step": 730 |
|
}, |
|
{ |
|
"epoch": 5.94, |
|
"learning_rate": 1.244761106454317e-05, |
|
"loss": 0.2353, |
|
"step": 735 |
|
}, |
|
{ |
|
"epoch": 5.98, |
|
"learning_rate": 1.2321877619446772e-05, |
|
"loss": 0.2395, |
|
"step": 740 |
|
}, |
|
{ |
|
"epoch": 6.0, |
|
"eval_f1": 0.5213319458896981, |
|
"eval_loss": 0.26373252272605896, |
|
"eval_runtime": 29.187, |
|
"eval_samples_per_second": 33.885, |
|
"eval_steps_per_second": 8.497, |
|
"step": 742 |
|
}, |
|
{ |
|
"epoch": 6.02, |
|
"learning_rate": 1.2196144174350377e-05, |
|
"loss": 0.256, |
|
"step": 745 |
|
}, |
|
{ |
|
"epoch": 6.06, |
|
"learning_rate": 1.2070410729253981e-05, |
|
"loss": 0.2359, |
|
"step": 750 |
|
}, |
|
{ |
|
"epoch": 6.1, |
|
"learning_rate": 1.1944677284157587e-05, |
|
"loss": 0.239, |
|
"step": 755 |
|
}, |
|
{ |
|
"epoch": 6.14, |
|
"learning_rate": 1.181894383906119e-05, |
|
"loss": 0.2485, |
|
"step": 760 |
|
}, |
|
{ |
|
"epoch": 6.18, |
|
"learning_rate": 1.1693210393964794e-05, |
|
"loss": 0.254, |
|
"step": 765 |
|
}, |
|
{ |
|
"epoch": 6.23, |
|
"learning_rate": 1.1567476948868398e-05, |
|
"loss": 0.2459, |
|
"step": 770 |
|
}, |
|
{ |
|
"epoch": 6.27, |
|
"learning_rate": 1.1441743503772004e-05, |
|
"loss": 0.2328, |
|
"step": 775 |
|
}, |
|
{ |
|
"epoch": 6.31, |
|
"learning_rate": 1.131601005867561e-05, |
|
"loss": 0.2422, |
|
"step": 780 |
|
}, |
|
{ |
|
"epoch": 6.35, |
|
"learning_rate": 1.1190276613579213e-05, |
|
"loss": 0.2296, |
|
"step": 785 |
|
}, |
|
{ |
|
"epoch": 6.39, |
|
"learning_rate": 1.1064543168482817e-05, |
|
"loss": 0.2375, |
|
"step": 790 |
|
}, |
|
{ |
|
"epoch": 6.43, |
|
"learning_rate": 1.093880972338642e-05, |
|
"loss": 0.2341, |
|
"step": 795 |
|
}, |
|
{ |
|
"epoch": 6.47, |
|
"learning_rate": 1.0813076278290026e-05, |
|
"loss": 0.2424, |
|
"step": 800 |
|
}, |
|
{ |
|
"epoch": 6.51, |
|
"learning_rate": 1.068734283319363e-05, |
|
"loss": 0.2253, |
|
"step": 805 |
|
}, |
|
{ |
|
"epoch": 6.55, |
|
"learning_rate": 1.0561609388097236e-05, |
|
"loss": 0.2386, |
|
"step": 810 |
|
}, |
|
{ |
|
"epoch": 6.59, |
|
"learning_rate": 1.0435875943000838e-05, |
|
"loss": 0.2546, |
|
"step": 815 |
|
}, |
|
{ |
|
"epoch": 6.63, |
|
"learning_rate": 1.0310142497904443e-05, |
|
"loss": 0.224, |
|
"step": 820 |
|
}, |
|
{ |
|
"epoch": 6.67, |
|
"learning_rate": 1.0184409052808047e-05, |
|
"loss": 0.2337, |
|
"step": 825 |
|
}, |
|
{ |
|
"epoch": 6.71, |
|
"learning_rate": 1.0058675607711653e-05, |
|
"loss": 0.2415, |
|
"step": 830 |
|
}, |
|
{ |
|
"epoch": 6.75, |
|
"learning_rate": 9.932942162615255e-06, |
|
"loss": 0.2309, |
|
"step": 835 |
|
}, |
|
{ |
|
"epoch": 6.79, |
|
"learning_rate": 9.80720871751886e-06, |
|
"loss": 0.2257, |
|
"step": 840 |
|
}, |
|
{ |
|
"epoch": 6.83, |
|
"learning_rate": 9.681475272422464e-06, |
|
"loss": 0.2373, |
|
"step": 845 |
|
}, |
|
{ |
|
"epoch": 6.87, |
|
"learning_rate": 9.55574182732607e-06, |
|
"loss": 0.2483, |
|
"step": 850 |
|
}, |
|
{ |
|
"epoch": 6.91, |
|
"learning_rate": 9.430008382229673e-06, |
|
"loss": 0.2437, |
|
"step": 855 |
|
}, |
|
{ |
|
"epoch": 6.95, |
|
"learning_rate": 9.304274937133277e-06, |
|
"loss": 0.2344, |
|
"step": 860 |
|
}, |
|
{ |
|
"epoch": 6.99, |
|
"learning_rate": 9.178541492036881e-06, |
|
"loss": 0.2469, |
|
"step": 865 |
|
}, |
|
{ |
|
"epoch": 6.99, |
|
"eval_f1": 0.5234760051880674, |
|
"eval_loss": 0.26504993438720703, |
|
"eval_runtime": 33.4377, |
|
"eval_samples_per_second": 29.577, |
|
"eval_steps_per_second": 7.417, |
|
"step": 865 |
|
}, |
|
{ |
|
"epoch": 7.03, |
|
"learning_rate": 9.052808046940487e-06, |
|
"loss": 0.2417, |
|
"step": 870 |
|
}, |
|
{ |
|
"epoch": 7.07, |
|
"learning_rate": 8.92707460184409e-06, |
|
"loss": 0.2271, |
|
"step": 875 |
|
}, |
|
{ |
|
"epoch": 7.11, |
|
"learning_rate": 8.801341156747696e-06, |
|
"loss": 0.2289, |
|
"step": 880 |
|
}, |
|
{ |
|
"epoch": 7.16, |
|
"learning_rate": 8.675607711651298e-06, |
|
"loss": 0.2285, |
|
"step": 885 |
|
}, |
|
{ |
|
"epoch": 7.2, |
|
"learning_rate": 8.549874266554904e-06, |
|
"loss": 0.2377, |
|
"step": 890 |
|
}, |
|
{ |
|
"epoch": 7.24, |
|
"learning_rate": 8.424140821458508e-06, |
|
"loss": 0.2371, |
|
"step": 895 |
|
}, |
|
{ |
|
"epoch": 7.28, |
|
"learning_rate": 8.298407376362113e-06, |
|
"loss": 0.2407, |
|
"step": 900 |
|
}, |
|
{ |
|
"epoch": 7.32, |
|
"learning_rate": 8.172673931265717e-06, |
|
"loss": 0.2325, |
|
"step": 905 |
|
}, |
|
{ |
|
"epoch": 7.36, |
|
"learning_rate": 8.04694048616932e-06, |
|
"loss": 0.2289, |
|
"step": 910 |
|
}, |
|
{ |
|
"epoch": 7.4, |
|
"learning_rate": 7.921207041072925e-06, |
|
"loss": 0.2383, |
|
"step": 915 |
|
}, |
|
{ |
|
"epoch": 7.44, |
|
"learning_rate": 7.79547359597653e-06, |
|
"loss": 0.2312, |
|
"step": 920 |
|
}, |
|
{ |
|
"epoch": 7.48, |
|
"learning_rate": 7.669740150880136e-06, |
|
"loss": 0.2255, |
|
"step": 925 |
|
}, |
|
{ |
|
"epoch": 7.52, |
|
"learning_rate": 7.544006705783739e-06, |
|
"loss": 0.2387, |
|
"step": 930 |
|
}, |
|
{ |
|
"epoch": 7.56, |
|
"learning_rate": 7.418273260687343e-06, |
|
"loss": 0.2328, |
|
"step": 935 |
|
}, |
|
{ |
|
"epoch": 7.6, |
|
"learning_rate": 7.292539815590947e-06, |
|
"loss": 0.2359, |
|
"step": 940 |
|
}, |
|
{ |
|
"epoch": 7.64, |
|
"learning_rate": 7.166806370494552e-06, |
|
"loss": 0.2322, |
|
"step": 945 |
|
}, |
|
{ |
|
"epoch": 7.68, |
|
"learning_rate": 7.041072925398156e-06, |
|
"loss": 0.2372, |
|
"step": 950 |
|
}, |
|
{ |
|
"epoch": 7.72, |
|
"learning_rate": 6.91533948030176e-06, |
|
"loss": 0.2297, |
|
"step": 955 |
|
}, |
|
{ |
|
"epoch": 7.76, |
|
"learning_rate": 6.789606035205365e-06, |
|
"loss": 0.2345, |
|
"step": 960 |
|
}, |
|
{ |
|
"epoch": 7.8, |
|
"learning_rate": 6.663872590108969e-06, |
|
"loss": 0.2229, |
|
"step": 965 |
|
}, |
|
{ |
|
"epoch": 7.84, |
|
"learning_rate": 6.5381391450125735e-06, |
|
"loss": 0.2395, |
|
"step": 970 |
|
}, |
|
{ |
|
"epoch": 7.88, |
|
"learning_rate": 6.412405699916177e-06, |
|
"loss": 0.2395, |
|
"step": 975 |
|
}, |
|
{ |
|
"epoch": 7.92, |
|
"learning_rate": 6.286672254819782e-06, |
|
"loss": 0.2226, |
|
"step": 980 |
|
}, |
|
{ |
|
"epoch": 7.96, |
|
"learning_rate": 6.160938809723386e-06, |
|
"loss": 0.2301, |
|
"step": 985 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"eval_f1": 0.5250844814140889, |
|
"eval_loss": 0.2636159062385559, |
|
"eval_runtime": 39.4887, |
|
"eval_samples_per_second": 25.045, |
|
"eval_steps_per_second": 6.28, |
|
"step": 989 |
|
}, |
|
{ |
|
"epoch": 8.0, |
|
"learning_rate": 6.0352053646269905e-06, |
|
"loss": 0.2295, |
|
"step": 990 |
|
}, |
|
{ |
|
"epoch": 8.04, |
|
"learning_rate": 5.909471919530595e-06, |
|
"loss": 0.2384, |
|
"step": 995 |
|
}, |
|
{ |
|
"epoch": 8.08, |
|
"learning_rate": 5.783738474434199e-06, |
|
"loss": 0.2277, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 8.13, |
|
"learning_rate": 5.658005029337805e-06, |
|
"loss": 0.2295, |
|
"step": 1005 |
|
}, |
|
{ |
|
"epoch": 8.17, |
|
"learning_rate": 5.532271584241408e-06, |
|
"loss": 0.2323, |
|
"step": 1010 |
|
}, |
|
{ |
|
"epoch": 8.21, |
|
"learning_rate": 5.406538139145013e-06, |
|
"loss": 0.2292, |
|
"step": 1015 |
|
}, |
|
{ |
|
"epoch": 8.25, |
|
"learning_rate": 5.280804694048618e-06, |
|
"loss": 0.2265, |
|
"step": 1020 |
|
}, |
|
{ |
|
"epoch": 8.29, |
|
"learning_rate": 5.155071248952222e-06, |
|
"loss": 0.2213, |
|
"step": 1025 |
|
}, |
|
{ |
|
"epoch": 8.33, |
|
"learning_rate": 5.029337803855826e-06, |
|
"loss": 0.219, |
|
"step": 1030 |
|
}, |
|
{ |
|
"epoch": 8.37, |
|
"learning_rate": 4.90360435875943e-06, |
|
"loss": 0.2201, |
|
"step": 1035 |
|
}, |
|
{ |
|
"epoch": 8.41, |
|
"learning_rate": 4.777870913663035e-06, |
|
"loss": 0.2318, |
|
"step": 1040 |
|
}, |
|
{ |
|
"epoch": 8.45, |
|
"learning_rate": 4.652137468566639e-06, |
|
"loss": 0.2316, |
|
"step": 1045 |
|
}, |
|
{ |
|
"epoch": 8.49, |
|
"learning_rate": 4.526404023470243e-06, |
|
"loss": 0.2279, |
|
"step": 1050 |
|
}, |
|
{ |
|
"epoch": 8.53, |
|
"learning_rate": 4.400670578373848e-06, |
|
"loss": 0.2237, |
|
"step": 1055 |
|
}, |
|
{ |
|
"epoch": 8.57, |
|
"learning_rate": 4.274937133277452e-06, |
|
"loss": 0.2469, |
|
"step": 1060 |
|
}, |
|
{ |
|
"epoch": 8.61, |
|
"learning_rate": 4.1492036881810565e-06, |
|
"loss": 0.224, |
|
"step": 1065 |
|
}, |
|
{ |
|
"epoch": 8.65, |
|
"learning_rate": 4.02347024308466e-06, |
|
"loss": 0.2277, |
|
"step": 1070 |
|
}, |
|
{ |
|
"epoch": 8.69, |
|
"learning_rate": 3.897736797988265e-06, |
|
"loss": 0.2334, |
|
"step": 1075 |
|
}, |
|
{ |
|
"epoch": 8.73, |
|
"learning_rate": 3.7720033528918693e-06, |
|
"loss": 0.2291, |
|
"step": 1080 |
|
}, |
|
{ |
|
"epoch": 8.77, |
|
"learning_rate": 3.6462699077954736e-06, |
|
"loss": 0.2333, |
|
"step": 1085 |
|
}, |
|
{ |
|
"epoch": 8.81, |
|
"learning_rate": 3.520536462699078e-06, |
|
"loss": 0.2358, |
|
"step": 1090 |
|
}, |
|
{ |
|
"epoch": 8.85, |
|
"learning_rate": 3.3948030176026825e-06, |
|
"loss": 0.2314, |
|
"step": 1095 |
|
}, |
|
{ |
|
"epoch": 8.89, |
|
"learning_rate": 3.2690695725062868e-06, |
|
"loss": 0.2263, |
|
"step": 1100 |
|
}, |
|
{ |
|
"epoch": 8.93, |
|
"learning_rate": 3.143336127409891e-06, |
|
"loss": 0.2274, |
|
"step": 1105 |
|
}, |
|
{ |
|
"epoch": 8.97, |
|
"learning_rate": 3.0176026823134953e-06, |
|
"loss": 0.2273, |
|
"step": 1110 |
|
}, |
|
{ |
|
"epoch": 9.0, |
|
"eval_f1": 0.5394001016776816, |
|
"eval_loss": 0.26480013132095337, |
|
"eval_runtime": 32.8402, |
|
"eval_samples_per_second": 30.116, |
|
"eval_steps_per_second": 7.552, |
|
"step": 1113 |
|
}, |
|
{ |
|
"epoch": 9.01, |
|
"learning_rate": 2.8918692372170995e-06, |
|
"loss": 0.2256, |
|
"step": 1115 |
|
}, |
|
{ |
|
"epoch": 9.06, |
|
"learning_rate": 2.766135792120704e-06, |
|
"loss": 0.2162, |
|
"step": 1120 |
|
}, |
|
{ |
|
"epoch": 9.1, |
|
"learning_rate": 2.640402347024309e-06, |
|
"loss": 0.231, |
|
"step": 1125 |
|
}, |
|
{ |
|
"epoch": 9.14, |
|
"learning_rate": 2.514668901927913e-06, |
|
"loss": 0.2298, |
|
"step": 1130 |
|
}, |
|
{ |
|
"epoch": 9.18, |
|
"learning_rate": 2.3889354568315174e-06, |
|
"loss": 0.2291, |
|
"step": 1135 |
|
}, |
|
{ |
|
"epoch": 9.22, |
|
"learning_rate": 2.2632020117351217e-06, |
|
"loss": 0.2263, |
|
"step": 1140 |
|
}, |
|
{ |
|
"epoch": 9.26, |
|
"learning_rate": 2.137468566638726e-06, |
|
"loss": 0.2211, |
|
"step": 1145 |
|
}, |
|
{ |
|
"epoch": 9.3, |
|
"learning_rate": 2.01173512154233e-06, |
|
"loss": 0.2302, |
|
"step": 1150 |
|
}, |
|
{ |
|
"epoch": 9.34, |
|
"learning_rate": 1.8860016764459346e-06, |
|
"loss": 0.2218, |
|
"step": 1155 |
|
}, |
|
{ |
|
"epoch": 9.38, |
|
"learning_rate": 1.760268231349539e-06, |
|
"loss": 0.2366, |
|
"step": 1160 |
|
}, |
|
{ |
|
"epoch": 9.42, |
|
"learning_rate": 1.6345347862531434e-06, |
|
"loss": 0.2307, |
|
"step": 1165 |
|
}, |
|
{ |
|
"epoch": 9.46, |
|
"learning_rate": 1.5088013411567476e-06, |
|
"loss": 0.234, |
|
"step": 1170 |
|
}, |
|
{ |
|
"epoch": 9.5, |
|
"learning_rate": 1.383067896060352e-06, |
|
"loss": 0.2296, |
|
"step": 1175 |
|
}, |
|
{ |
|
"epoch": 9.54, |
|
"learning_rate": 1.2573344509639566e-06, |
|
"loss": 0.2307, |
|
"step": 1180 |
|
}, |
|
{ |
|
"epoch": 9.58, |
|
"learning_rate": 1.1316010058675608e-06, |
|
"loss": 0.2234, |
|
"step": 1185 |
|
}, |
|
{ |
|
"epoch": 9.62, |
|
"learning_rate": 1.005867560771165e-06, |
|
"loss": 0.2299, |
|
"step": 1190 |
|
}, |
|
{ |
|
"epoch": 9.66, |
|
"learning_rate": 8.801341156747695e-07, |
|
"loss": 0.2325, |
|
"step": 1195 |
|
}, |
|
{ |
|
"epoch": 9.7, |
|
"learning_rate": 7.544006705783738e-07, |
|
"loss": 0.2238, |
|
"step": 1200 |
|
}, |
|
{ |
|
"epoch": 9.74, |
|
"learning_rate": 6.286672254819783e-07, |
|
"loss": 0.2293, |
|
"step": 1205 |
|
}, |
|
{ |
|
"epoch": 9.78, |
|
"learning_rate": 5.029337803855825e-07, |
|
"loss": 0.2265, |
|
"step": 1210 |
|
}, |
|
{ |
|
"epoch": 9.82, |
|
"learning_rate": 3.772003352891869e-07, |
|
"loss": 0.2245, |
|
"step": 1215 |
|
}, |
|
{ |
|
"epoch": 9.86, |
|
"learning_rate": 2.5146689019279127e-07, |
|
"loss": 0.2133, |
|
"step": 1220 |
|
}, |
|
{ |
|
"epoch": 9.9, |
|
"learning_rate": 1.2573344509639564e-07, |
|
"loss": 0.2202, |
|
"step": 1225 |
|
}, |
|
{ |
|
"epoch": 9.94, |
|
"learning_rate": 0.0, |
|
"loss": 0.2169, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 9.94, |
|
"eval_f1": 0.5395372489193999, |
|
"eval_loss": 0.2657696008682251, |
|
"eval_runtime": 31.9099, |
|
"eval_samples_per_second": 30.993, |
|
"eval_steps_per_second": 7.772, |
|
"step": 1230 |
|
}, |
|
{ |
|
"epoch": 9.94, |
|
"step": 1230, |
|
"total_flos": 462869108028288.0, |
|
"train_loss": 47546.01565268776, |
|
"train_runtime": 9264.1071, |
|
"train_samples_per_second": 8.543, |
|
"train_steps_per_second": 0.133 |
|
} |
|
], |
|
"logging_steps": 5, |
|
"max_steps": 1230, |
|
"num_train_epochs": 10, |
|
"save_steps": 500, |
|
"total_flos": 462869108028288.0, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|