| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 1110, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.504504504504505e-07, | |
| "loss": 3.2559, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.00900900900901e-07, | |
| "loss": 3.2281, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 1.3513513513513515e-06, | |
| "loss": 3.1953, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 1.801801801801802e-06, | |
| "loss": 3.0844, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.2522522522522524e-06, | |
| "loss": 3.2563, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 2.702702702702703e-06, | |
| "loss": 3.098, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 3.1531531531531532e-06, | |
| "loss": 2.9754, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 3.603603603603604e-06, | |
| "loss": 3.1168, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.0540540540540545e-06, | |
| "loss": 3.2441, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.504504504504505e-06, | |
| "loss": 3.084, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.954954954954955e-06, | |
| "loss": 3.1965, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 5.405405405405406e-06, | |
| "loss": 2.9336, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 5.855855855855856e-06, | |
| "loss": 2.977, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 6.3063063063063065e-06, | |
| "loss": 3.0297, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 6.7567567567567575e-06, | |
| "loss": 2.9363, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 7.207207207207208e-06, | |
| "loss": 2.9477, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 7.657657657657658e-06, | |
| "loss": 3.0219, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.108108108108109e-06, | |
| "loss": 2.9004, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 8.55855855855856e-06, | |
| "loss": 2.991, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 9.00900900900901e-06, | |
| "loss": 2.8973, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 9.45945945945946e-06, | |
| "loss": 2.9184, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 9.90990990990991e-06, | |
| "loss": 2.868, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 9.95995995995996e-06, | |
| "loss": 2.8301, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 9.90990990990991e-06, | |
| "loss": 2.716, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 9.85985985985986e-06, | |
| "loss": 2.8859, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 9.80980980980981e-06, | |
| "loss": 2.8418, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 9.75975975975976e-06, | |
| "loss": 2.8004, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 9.70970970970971e-06, | |
| "loss": 2.8203, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 9.65965965965966e-06, | |
| "loss": 2.8668, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 9.60960960960961e-06, | |
| "loss": 2.8055, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 9.55955955955956e-06, | |
| "loss": 2.7934, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 9.50950950950951e-06, | |
| "loss": 2.9164, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 9.45945945945946e-06, | |
| "loss": 2.9328, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.40940940940941e-06, | |
| "loss": 2.9363, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 9.35935935935936e-06, | |
| "loss": 2.841, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 9.30930930930931e-06, | |
| "loss": 2.7398, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 9.25925925925926e-06, | |
| "loss": 2.8695, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 9.20920920920921e-06, | |
| "loss": 2.8363, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 9.15915915915916e-06, | |
| "loss": 2.8844, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 9.10910910910911e-06, | |
| "loss": 2.7594, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 9.05905905905906e-06, | |
| "loss": 2.8078, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 9.00900900900901e-06, | |
| "loss": 2.9926, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 8.95895895895896e-06, | |
| "loss": 2.8766, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 8.90890890890891e-06, | |
| "loss": 2.8145, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 8.85885885885886e-06, | |
| "loss": 2.7383, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 8.80880880880881e-06, | |
| "loss": 2.7613, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 8.75875875875876e-06, | |
| "loss": 2.9582, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 8.70870870870871e-06, | |
| "loss": 2.8043, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 8.65865865865866e-06, | |
| "loss": 2.7922, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 8.60860860860861e-06, | |
| "loss": 2.9848, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 8.55855855855856e-06, | |
| "loss": 2.8777, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 8.50850850850851e-06, | |
| "loss": 2.9402, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 8.45845845845846e-06, | |
| "loss": 2.7785, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 8.408408408408409e-06, | |
| "loss": 2.8871, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.358358358358359e-06, | |
| "loss": 2.8348, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.308308308308309e-06, | |
| "loss": 2.7664, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 8.258258258258259e-06, | |
| "loss": 2.7973, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 8.208208208208209e-06, | |
| "loss": 2.7527, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 8.158158158158159e-06, | |
| "loss": 2.7844, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 8.108108108108109e-06, | |
| "loss": 2.8102, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 8.058058058058059e-06, | |
| "loss": 2.923, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 8.00800800800801e-06, | |
| "loss": 2.7535, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 7.95795795795796e-06, | |
| "loss": 2.8164, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 7.90790790790791e-06, | |
| "loss": 2.8312, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 7.85785785785786e-06, | |
| "loss": 2.8027, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 7.807807807807808e-06, | |
| "loss": 2.7281, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 7.757757757757758e-06, | |
| "loss": 2.8328, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 7.707707707707708e-06, | |
| "loss": 2.8695, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 7.657657657657658e-06, | |
| "loss": 2.8, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 7.607607607607608e-06, | |
| "loss": 2.8832, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 7.557557557557558e-06, | |
| "loss": 2.7445, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 7.507507507507507e-06, | |
| "loss": 2.7359, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 7.457457457457457e-06, | |
| "loss": 2.6559, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "loss": 2.7598, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 7.3573573573573575e-06, | |
| "loss": 2.8637, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 7.307307307307308e-06, | |
| "loss": 2.7266, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 7.257257257257258e-06, | |
| "loss": 2.7289, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 7.207207207207208e-06, | |
| "loss": 2.7895, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 7.157157157157158e-06, | |
| "loss": 2.7336, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 7.107107107107107e-06, | |
| "loss": 2.6094, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 7.057057057057057e-06, | |
| "loss": 2.9809, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 7.007007007007007e-06, | |
| "loss": 2.8156, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 6.956956956956957e-06, | |
| "loss": 2.798, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 6.906906906906907e-06, | |
| "loss": 2.6703, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 6.856856856856857e-06, | |
| "loss": 2.8105, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 6.8068068068068075e-06, | |
| "loss": 2.6707, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 6.7567567567567575e-06, | |
| "loss": 2.7289, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 6.706706706706707e-06, | |
| "loss": 2.7004, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 6.656656656656657e-06, | |
| "loss": 2.8719, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 6.606606606606607e-06, | |
| "loss": 2.6273, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 6.556556556556557e-06, | |
| "loss": 2.8629, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 6.506506506506507e-06, | |
| "loss": 2.8672, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 6.456456456456457e-06, | |
| "loss": 2.725, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 6.406406406406407e-06, | |
| "loss": 2.8004, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 6.356356356356357e-06, | |
| "loss": 2.882, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 6.3063063063063065e-06, | |
| "loss": 2.8027, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 6.2562562562562565e-06, | |
| "loss": 2.8988, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 6.206206206206207e-06, | |
| "loss": 2.6898, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 6.156156156156157e-06, | |
| "loss": 2.9855, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 6.106106106106107e-06, | |
| "loss": 2.8246, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 6.056056056056057e-06, | |
| "loss": 2.8918, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 6.006006006006007e-06, | |
| "loss": 2.8988, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 5.955955955955957e-06, | |
| "loss": 2.7313, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 5.905905905905906e-06, | |
| "loss": 2.8082, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 5.855855855855856e-06, | |
| "loss": 2.8164, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 5.805805805805806e-06, | |
| "loss": 2.7496, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 5.755755755755756e-06, | |
| "loss": 2.5914, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 5.7057057057057065e-06, | |
| "loss": 2.7824, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 5.6556556556556565e-06, | |
| "loss": 2.6574, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 5.605605605605607e-06, | |
| "loss": 2.8434, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 2.8793, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 5.05, | |
| "learning_rate": 5.505505505505506e-06, | |
| "loss": 2.7293, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 5.09, | |
| "learning_rate": 5.455455455455456e-06, | |
| "loss": 2.7664, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 5.405405405405406e-06, | |
| "loss": 2.7891, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 5.18, | |
| "learning_rate": 5.355355355355356e-06, | |
| "loss": 2.868, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 5.23, | |
| "learning_rate": 5.305305305305306e-06, | |
| "loss": 2.8664, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 5.27, | |
| "learning_rate": 5.255255255255256e-06, | |
| "loss": 2.8324, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 5.205205205205206e-06, | |
| "loss": 2.7469, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 5.36, | |
| "learning_rate": 5.155155155155156e-06, | |
| "loss": 2.7973, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 5.41, | |
| "learning_rate": 5.105105105105106e-06, | |
| "loss": 2.8254, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 5.055055055055056e-06, | |
| "loss": 2.7656, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 5.5, | |
| "learning_rate": 5.005005005005006e-06, | |
| "loss": 2.766, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 5.54, | |
| "learning_rate": 4.954954954954955e-06, | |
| "loss": 2.8266, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 5.59, | |
| "learning_rate": 4.904904904904905e-06, | |
| "loss": 2.868, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 4.854854854854855e-06, | |
| "loss": 2.7523, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 5.68, | |
| "learning_rate": 4.804804804804805e-06, | |
| "loss": 2.8625, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 5.72, | |
| "learning_rate": 4.754754754754755e-06, | |
| "loss": 2.7098, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 4.704704704704705e-06, | |
| "loss": 2.584, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 4.654654654654655e-06, | |
| "loss": 2.7652, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 5.86, | |
| "learning_rate": 4.604604604604605e-06, | |
| "loss": 2.7668, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 5.9, | |
| "learning_rate": 4.554554554554555e-06, | |
| "loss": 2.9934, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 4.504504504504505e-06, | |
| "loss": 2.807, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 5.99, | |
| "learning_rate": 4.454454454454455e-06, | |
| "loss": 2.7543, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 6.04, | |
| "learning_rate": 4.404404404404405e-06, | |
| "loss": 2.7055, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 6.08, | |
| "learning_rate": 4.354354354354355e-06, | |
| "loss": 2.6199, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 6.13, | |
| "learning_rate": 4.304304304304305e-06, | |
| "loss": 2.757, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 4.254254254254255e-06, | |
| "loss": 2.7941, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 4.204204204204204e-06, | |
| "loss": 2.7836, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 6.26, | |
| "learning_rate": 4.154154154154154e-06, | |
| "loss": 2.7062, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 6.31, | |
| "learning_rate": 4.1041041041041045e-06, | |
| "loss": 2.807, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 6.35, | |
| "learning_rate": 4.0540540540540545e-06, | |
| "loss": 2.8246, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 6.4, | |
| "learning_rate": 4.004004004004005e-06, | |
| "loss": 2.8395, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 6.44, | |
| "learning_rate": 3.953953953953955e-06, | |
| "loss": 2.7781, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 6.49, | |
| "learning_rate": 3.903903903903904e-06, | |
| "loss": 2.732, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 3.853853853853854e-06, | |
| "loss": 2.7281, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 6.58, | |
| "learning_rate": 3.803803803803804e-06, | |
| "loss": 2.6957, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 6.62, | |
| "learning_rate": 3.7537537537537537e-06, | |
| "loss": 2.7516, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 3.7037037037037037e-06, | |
| "loss": 2.741, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 6.71, | |
| "learning_rate": 3.653653653653654e-06, | |
| "loss": 2.798, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 6.76, | |
| "learning_rate": 3.603603603603604e-06, | |
| "loss": 2.741, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 6.8, | |
| "learning_rate": 3.5535535535535535e-06, | |
| "loss": 2.7691, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 6.85, | |
| "learning_rate": 3.5035035035035036e-06, | |
| "loss": 2.7039, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 3.4534534534534537e-06, | |
| "loss": 2.7797, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 6.94, | |
| "learning_rate": 3.4034034034034037e-06, | |
| "loss": 2.918, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 6.98, | |
| "learning_rate": 3.3533533533533534e-06, | |
| "loss": 2.6195, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 3.3033033033033035e-06, | |
| "loss": 2.5387, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 3.2532532532532535e-06, | |
| "loss": 2.7352, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 3.2032032032032036e-06, | |
| "loss": 2.5961, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 7.16, | |
| "learning_rate": 3.1531531531531532e-06, | |
| "loss": 2.8273, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 7.21, | |
| "learning_rate": 3.1031031031031033e-06, | |
| "loss": 2.6625, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 3.0530530530530534e-06, | |
| "loss": 2.7, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 7.3, | |
| "learning_rate": 3.0030030030030034e-06, | |
| "loss": 2.7527, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 7.34, | |
| "learning_rate": 2.952952952952953e-06, | |
| "loss": 2.7172, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 2.902902902902903e-06, | |
| "loss": 2.7395, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 7.43, | |
| "learning_rate": 2.8528528528528532e-06, | |
| "loss": 2.6105, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 7.48, | |
| "learning_rate": 2.8028028028028033e-06, | |
| "loss": 2.7566, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 7.52, | |
| "learning_rate": 2.752752752752753e-06, | |
| "loss": 2.6984, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 2.702702702702703e-06, | |
| "loss": 2.8969, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 2.652652652652653e-06, | |
| "loss": 2.6301, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 7.66, | |
| "learning_rate": 2.602602602602603e-06, | |
| "loss": 2.7102, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 7.7, | |
| "learning_rate": 2.552552552552553e-06, | |
| "loss": 2.9152, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 2.502502502502503e-06, | |
| "loss": 2.6543, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 7.79, | |
| "learning_rate": 2.4524524524524525e-06, | |
| "loss": 2.6109, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 7.84, | |
| "learning_rate": 2.4024024024024026e-06, | |
| "loss": 2.5871, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 2.3523523523523527e-06, | |
| "loss": 2.752, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 7.93, | |
| "learning_rate": 2.3023023023023023e-06, | |
| "loss": 2.716, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 2.2522522522522524e-06, | |
| "loss": 2.807, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 2.2022022022022024e-06, | |
| "loss": 2.877, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 8.06, | |
| "learning_rate": 2.1521521521521525e-06, | |
| "loss": 2.716, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 2.102102102102102e-06, | |
| "loss": 2.6879, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 8.15, | |
| "learning_rate": 2.0520520520520522e-06, | |
| "loss": 2.7137, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 2.0020020020020023e-06, | |
| "loss": 2.7133, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 8.24, | |
| "learning_rate": 1.951951951951952e-06, | |
| "loss": 2.7453, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 8.29, | |
| "learning_rate": 1.901901901901902e-06, | |
| "loss": 2.8281, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 8.33, | |
| "learning_rate": 1.8518518518518519e-06, | |
| "loss": 2.857, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 8.38, | |
| "learning_rate": 1.801801801801802e-06, | |
| "loss": 2.7105, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 8.42, | |
| "learning_rate": 1.7517517517517518e-06, | |
| "loss": 2.7879, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 1.7017017017017019e-06, | |
| "loss": 2.8539, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 8.51, | |
| "learning_rate": 1.6516516516516517e-06, | |
| "loss": 2.7656, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 8.56, | |
| "learning_rate": 1.6016016016016018e-06, | |
| "loss": 2.677, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 8.6, | |
| "learning_rate": 1.5515515515515517e-06, | |
| "loss": 2.809, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 8.65, | |
| "learning_rate": 1.5015015015015017e-06, | |
| "loss": 2.5945, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 8.69, | |
| "learning_rate": 1.4514514514514516e-06, | |
| "loss": 2.8148, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 8.74, | |
| "learning_rate": 1.4014014014014016e-06, | |
| "loss": 2.7242, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 8.78, | |
| "learning_rate": 1.3513513513513515e-06, | |
| "loss": 2.9172, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 1.3013013013013016e-06, | |
| "loss": 2.7965, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 8.87, | |
| "learning_rate": 1.2512512512512514e-06, | |
| "loss": 2.8918, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 1.2012012012012013e-06, | |
| "loss": 2.6551, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 8.96, | |
| "learning_rate": 1.1511511511511512e-06, | |
| "loss": 2.8613, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 9.01, | |
| "learning_rate": 1.1011011011011012e-06, | |
| "loss": 2.7629, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 9.05, | |
| "learning_rate": 1.051051051051051e-06, | |
| "loss": 2.5543, | |
| "step": 1005 | |
| }, | |
| { | |
| "epoch": 9.1, | |
| "learning_rate": 1.0010010010010011e-06, | |
| "loss": 2.9523, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 9.14, | |
| "learning_rate": 9.50950950950951e-07, | |
| "loss": 2.7785, | |
| "step": 1015 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 9.00900900900901e-07, | |
| "loss": 2.6758, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 9.23, | |
| "learning_rate": 8.508508508508509e-07, | |
| "loss": 2.5539, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 9.28, | |
| "learning_rate": 8.008008008008009e-07, | |
| "loss": 2.6223, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 7.507507507507509e-07, | |
| "loss": 2.882, | |
| "step": 1035 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 7.007007007007008e-07, | |
| "loss": 2.9254, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 9.41, | |
| "learning_rate": 6.506506506506508e-07, | |
| "loss": 2.825, | |
| "step": 1045 | |
| }, | |
| { | |
| "epoch": 9.46, | |
| "learning_rate": 6.006006006006006e-07, | |
| "loss": 2.7879, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 9.5, | |
| "learning_rate": 5.505505505505506e-07, | |
| "loss": 2.7621, | |
| "step": 1055 | |
| }, | |
| { | |
| "epoch": 9.55, | |
| "learning_rate": 5.005005005005006e-07, | |
| "loss": 2.7711, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 9.59, | |
| "learning_rate": 4.504504504504505e-07, | |
| "loss": 2.7539, | |
| "step": 1065 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 4.0040040040040045e-07, | |
| "loss": 2.7781, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 9.68, | |
| "learning_rate": 3.503503503503504e-07, | |
| "loss": 2.7246, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 9.73, | |
| "learning_rate": 3.003003003003003e-07, | |
| "loss": 2.9125, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 9.77, | |
| "learning_rate": 2.502502502502503e-07, | |
| "loss": 2.757, | |
| "step": 1085 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "learning_rate": 2.0020020020020022e-07, | |
| "loss": 2.7824, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 9.86, | |
| "learning_rate": 1.5015015015015016e-07, | |
| "loss": 2.6824, | |
| "step": 1095 | |
| }, | |
| { | |
| "epoch": 9.91, | |
| "learning_rate": 1.0010010010010011e-07, | |
| "loss": 2.6035, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 9.95, | |
| "learning_rate": 5.0050050050050056e-08, | |
| "loss": 2.7152, | |
| "step": 1105 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "learning_rate": 0.0, | |
| "loss": 2.7191, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 1110, | |
| "total_flos": 2.1618654360131705e+21, | |
| "train_loss": 2.8070910754504506, | |
| "train_runtime": 11217.8167, | |
| "train_samples_per_second": 101.845, | |
| "train_steps_per_second": 0.099 | |
| } | |
| ], | |
| "max_steps": 1110, | |
| "num_train_epochs": 10, | |
| "total_flos": 2.1618654360131705e+21, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |