| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 185.1851851851852, | |
| "eval_steps": 500, | |
| "global_step": 3000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06172839506172839, | |
| "learning_rate": 5e-09, | |
| "loss": 8.604, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 1.2345679012345678, | |
| "learning_rate": 1e-07, | |
| "loss": 8.0873, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 2.4691358024691357, | |
| "learning_rate": 2e-07, | |
| "loss": 7.8074, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 3.7037037037037037, | |
| "learning_rate": 3e-07, | |
| "loss": 7.0763, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 4.938271604938271, | |
| "learning_rate": 4e-07, | |
| "loss": 6.2969, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 6.172839506172839, | |
| "learning_rate": 5e-07, | |
| "loss": 5.9334, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 7.407407407407407, | |
| "learning_rate": 6e-07, | |
| "loss": 5.4919, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 8.641975308641975, | |
| "learning_rate": 7e-07, | |
| "loss": 5.0659, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 9.876543209876543, | |
| "learning_rate": 8e-07, | |
| "loss": 4.9089, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 11.11111111111111, | |
| "learning_rate": 9e-07, | |
| "loss": 4.7144, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 12.345679012345679, | |
| "learning_rate": 1e-06, | |
| "loss": 4.7373, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 13.580246913580247, | |
| "learning_rate": 9.953798767967146e-07, | |
| "loss": 4.3848, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 14.814814814814815, | |
| "learning_rate": 9.90759753593429e-07, | |
| "loss": 4.3461, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 16.049382716049383, | |
| "learning_rate": 9.861396303901435e-07, | |
| "loss": 4.2226, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 17.28395061728395, | |
| "learning_rate": 9.815195071868584e-07, | |
| "loss": 4.1291, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 18.51851851851852, | |
| "learning_rate": 9.768993839835729e-07, | |
| "loss": 4.0118, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 19.753086419753085, | |
| "learning_rate": 9.722792607802873e-07, | |
| "loss": 4.0819, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 20.987654320987655, | |
| "learning_rate": 9.67659137577002e-07, | |
| "loss": 3.9431, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 22.22222222222222, | |
| "learning_rate": 9.630390143737167e-07, | |
| "loss": 3.8781, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 23.45679012345679, | |
| "learning_rate": 9.58418891170431e-07, | |
| "loss": 3.8416, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 24.691358024691358, | |
| "learning_rate": 9.537987679671456e-07, | |
| "loss": 3.744, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 25.925925925925927, | |
| "learning_rate": 9.491786447638603e-07, | |
| "loss": 3.7832, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 27.160493827160494, | |
| "learning_rate": 9.445585215605749e-07, | |
| "loss": 3.7876, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 28.395061728395063, | |
| "learning_rate": 9.399383983572896e-07, | |
| "loss": 3.6849, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 29.62962962962963, | |
| "learning_rate": 9.35318275154004e-07, | |
| "loss": 3.7099, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 30.864197530864196, | |
| "learning_rate": 9.306981519507186e-07, | |
| "loss": 3.7103, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 32.098765432098766, | |
| "learning_rate": 9.260780287474332e-07, | |
| "loss": 3.6856, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 33.333333333333336, | |
| "learning_rate": 9.214579055441478e-07, | |
| "loss": 3.625, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 34.5679012345679, | |
| "learning_rate": 9.168377823408624e-07, | |
| "loss": 3.5128, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 35.80246913580247, | |
| "learning_rate": 9.122176591375768e-07, | |
| "loss": 3.4695, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 37.03703703703704, | |
| "learning_rate": 9.075975359342916e-07, | |
| "loss": 3.5786, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 38.27160493827161, | |
| "learning_rate": 9.029774127310061e-07, | |
| "loss": 3.4751, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 39.50617283950617, | |
| "learning_rate": 8.983572895277207e-07, | |
| "loss": 3.4836, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 40.74074074074074, | |
| "learning_rate": 8.937371663244353e-07, | |
| "loss": 3.4209, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 41.97530864197531, | |
| "learning_rate": 8.891170431211498e-07, | |
| "loss": 3.3073, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 43.20987654320987, | |
| "learning_rate": 8.844969199178645e-07, | |
| "loss": 3.3712, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 44.44444444444444, | |
| "learning_rate": 8.79876796714579e-07, | |
| "loss": 3.3412, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 45.67901234567901, | |
| "learning_rate": 8.752566735112936e-07, | |
| "loss": 3.5548, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 46.91358024691358, | |
| "learning_rate": 8.706365503080082e-07, | |
| "loss": 3.5773, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 48.148148148148145, | |
| "learning_rate": 8.660164271047227e-07, | |
| "loss": 3.3696, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 49.382716049382715, | |
| "learning_rate": 8.613963039014373e-07, | |
| "loss": 3.2795, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 50.617283950617285, | |
| "learning_rate": 8.56776180698152e-07, | |
| "loss": 3.2955, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 51.851851851851855, | |
| "learning_rate": 8.521560574948665e-07, | |
| "loss": 3.2566, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 53.08641975308642, | |
| "learning_rate": 8.475359342915811e-07, | |
| "loss": 3.2109, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 54.32098765432099, | |
| "learning_rate": 8.429158110882956e-07, | |
| "loss": 3.1969, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 55.55555555555556, | |
| "learning_rate": 8.382956878850103e-07, | |
| "loss": 3.1772, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 56.79012345679013, | |
| "learning_rate": 8.336755646817249e-07, | |
| "loss": 3.175, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 58.02469135802469, | |
| "learning_rate": 8.290554414784393e-07, | |
| "loss": 3.1683, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 59.25925925925926, | |
| "learning_rate": 8.244353182751539e-07, | |
| "loss": 3.1107, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 60.49382716049383, | |
| "learning_rate": 8.198151950718686e-07, | |
| "loss": 3.1564, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 61.72839506172839, | |
| "learning_rate": 8.151950718685832e-07, | |
| "loss": 3.0278, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 62.96296296296296, | |
| "learning_rate": 8.105749486652977e-07, | |
| "loss": 3.0933, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 64.19753086419753, | |
| "learning_rate": 8.059548254620122e-07, | |
| "loss": 3.1135, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 65.4320987654321, | |
| "learning_rate": 8.013347022587269e-07, | |
| "loss": 3.1195, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 66.66666666666667, | |
| "learning_rate": 7.967145790554415e-07, | |
| "loss": 3.0757, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 67.90123456790124, | |
| "learning_rate": 7.92094455852156e-07, | |
| "loss": 2.9796, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 69.1358024691358, | |
| "learning_rate": 7.874743326488706e-07, | |
| "loss": 3.0387, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 70.37037037037037, | |
| "learning_rate": 7.828542094455851e-07, | |
| "loss": 2.9829, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 71.60493827160494, | |
| "learning_rate": 7.782340862422998e-07, | |
| "loss": 2.9699, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 72.8395061728395, | |
| "learning_rate": 7.736139630390144e-07, | |
| "loss": 3.0227, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 74.07407407407408, | |
| "learning_rate": 7.689938398357289e-07, | |
| "loss": 3.0769, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 75.30864197530865, | |
| "learning_rate": 7.643737166324435e-07, | |
| "loss": 2.9983, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 76.54320987654322, | |
| "learning_rate": 7.59753593429158e-07, | |
| "loss": 3.0172, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 77.77777777777777, | |
| "learning_rate": 7.551334702258726e-07, | |
| "loss": 2.8962, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 79.01234567901234, | |
| "learning_rate": 7.505133470225873e-07, | |
| "loss": 2.9221, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 80.24691358024691, | |
| "learning_rate": 7.458932238193018e-07, | |
| "loss": 2.9511, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 81.48148148148148, | |
| "learning_rate": 7.412731006160164e-07, | |
| "loss": 2.9408, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 82.71604938271605, | |
| "learning_rate": 7.36652977412731e-07, | |
| "loss": 2.9079, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 83.95061728395062, | |
| "learning_rate": 7.320328542094455e-07, | |
| "loss": 2.9289, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 85.18518518518519, | |
| "learning_rate": 7.274127310061602e-07, | |
| "loss": 2.9366, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 86.41975308641975, | |
| "learning_rate": 7.227926078028747e-07, | |
| "loss": 2.9584, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 87.65432098765432, | |
| "learning_rate": 7.181724845995892e-07, | |
| "loss": 2.8547, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 88.88888888888889, | |
| "learning_rate": 7.135523613963038e-07, | |
| "loss": 2.8381, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 90.12345679012346, | |
| "learning_rate": 7.089322381930185e-07, | |
| "loss": 2.8748, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 91.35802469135803, | |
| "learning_rate": 7.04312114989733e-07, | |
| "loss": 2.9128, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 92.5925925925926, | |
| "learning_rate": 6.996919917864475e-07, | |
| "loss": 2.8814, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 93.82716049382717, | |
| "learning_rate": 6.950718685831621e-07, | |
| "loss": 2.8452, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 95.06172839506173, | |
| "learning_rate": 6.904517453798768e-07, | |
| "loss": 2.8592, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 96.29629629629629, | |
| "learning_rate": 6.858316221765913e-07, | |
| "loss": 2.8163, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 97.53086419753086, | |
| "learning_rate": 6.812114989733059e-07, | |
| "loss": 2.786, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 98.76543209876543, | |
| "learning_rate": 6.765913757700205e-07, | |
| "loss": 2.8323, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 100.0, | |
| "learning_rate": 6.719712525667351e-07, | |
| "loss": 2.8326, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 101.23456790123457, | |
| "learning_rate": 6.673511293634497e-07, | |
| "loss": 2.8083, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 102.46913580246914, | |
| "learning_rate": 6.627310061601642e-07, | |
| "loss": 2.8271, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 103.70370370370371, | |
| "learning_rate": 6.581108829568788e-07, | |
| "loss": 2.8487, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 104.93827160493827, | |
| "learning_rate": 6.534907597535934e-07, | |
| "loss": 2.7786, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 106.17283950617283, | |
| "learning_rate": 6.488706365503079e-07, | |
| "loss": 2.8222, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 107.4074074074074, | |
| "learning_rate": 6.442505133470226e-07, | |
| "loss": 2.7864, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 108.64197530864197, | |
| "learning_rate": 6.39630390143737e-07, | |
| "loss": 2.7766, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 109.87654320987654, | |
| "learning_rate": 6.350102669404517e-07, | |
| "loss": 2.7741, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 111.11111111111111, | |
| "learning_rate": 6.303901437371663e-07, | |
| "loss": 2.8072, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 112.34567901234568, | |
| "learning_rate": 6.257700205338809e-07, | |
| "loss": 2.7551, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 113.58024691358025, | |
| "learning_rate": 6.211498973305955e-07, | |
| "loss": 2.7096, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 114.81481481481481, | |
| "learning_rate": 6.165297741273101e-07, | |
| "loss": 2.7817, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 116.04938271604938, | |
| "learning_rate": 6.119096509240245e-07, | |
| "loss": 2.7716, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 117.28395061728395, | |
| "learning_rate": 6.072895277207392e-07, | |
| "loss": 2.7476, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 118.51851851851852, | |
| "learning_rate": 6.026694045174538e-07, | |
| "loss": 2.758, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 119.75308641975309, | |
| "learning_rate": 5.980492813141684e-07, | |
| "loss": 2.7445, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 120.98765432098766, | |
| "learning_rate": 5.934291581108829e-07, | |
| "loss": 2.7234, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 122.22222222222223, | |
| "learning_rate": 5.888090349075973e-07, | |
| "loss": 2.7103, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 123.45679012345678, | |
| "learning_rate": 5.841889117043121e-07, | |
| "loss": 2.7609, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 124.69135802469135, | |
| "learning_rate": 5.795687885010268e-07, | |
| "loss": 2.7613, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 125.92592592592592, | |
| "learning_rate": 5.749486652977412e-07, | |
| "loss": 2.7793, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 127.1604938271605, | |
| "learning_rate": 5.703285420944558e-07, | |
| "loss": 2.6717, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 128.39506172839506, | |
| "learning_rate": 5.657084188911705e-07, | |
| "loss": 2.6988, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 129.62962962962962, | |
| "learning_rate": 5.61088295687885e-07, | |
| "loss": 2.7089, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 130.8641975308642, | |
| "learning_rate": 5.564681724845996e-07, | |
| "loss": 2.6745, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 132.09876543209876, | |
| "learning_rate": 5.518480492813141e-07, | |
| "loss": 2.6574, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 133.33333333333334, | |
| "learning_rate": 5.472279260780288e-07, | |
| "loss": 2.6682, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 134.5679012345679, | |
| "learning_rate": 5.426078028747432e-07, | |
| "loss": 2.7689, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 135.80246913580248, | |
| "learning_rate": 5.379876796714579e-07, | |
| "loss": 2.6941, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 137.03703703703704, | |
| "learning_rate": 5.333675564681725e-07, | |
| "loss": 2.692, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 138.2716049382716, | |
| "learning_rate": 5.28747433264887e-07, | |
| "loss": 2.6515, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 139.50617283950618, | |
| "learning_rate": 5.241273100616017e-07, | |
| "loss": 2.7289, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 140.74074074074073, | |
| "learning_rate": 5.195071868583161e-07, | |
| "loss": 2.6351, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 141.97530864197532, | |
| "learning_rate": 5.148870636550308e-07, | |
| "loss": 2.6695, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 143.20987654320987, | |
| "learning_rate": 5.102669404517453e-07, | |
| "loss": 2.6735, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 144.44444444444446, | |
| "learning_rate": 5.056468172484599e-07, | |
| "loss": 2.6271, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 145.679012345679, | |
| "learning_rate": 5.010266940451745e-07, | |
| "loss": 2.677, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 146.91358024691357, | |
| "learning_rate": 4.964065708418891e-07, | |
| "loss": 2.6575, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 148.14814814814815, | |
| "learning_rate": 4.917864476386037e-07, | |
| "loss": 2.5944, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 149.3827160493827, | |
| "learning_rate": 4.871663244353182e-07, | |
| "loss": 2.6355, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 150.6172839506173, | |
| "learning_rate": 4.825462012320328e-07, | |
| "loss": 2.6434, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 151.85185185185185, | |
| "learning_rate": 4.779260780287474e-07, | |
| "loss": 2.6251, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 153.08641975308643, | |
| "learning_rate": 4.73305954825462e-07, | |
| "loss": 2.6228, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 154.320987654321, | |
| "learning_rate": 4.686858316221766e-07, | |
| "loss": 2.6255, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 155.55555555555554, | |
| "learning_rate": 4.640657084188911e-07, | |
| "loss": 2.6125, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 156.79012345679013, | |
| "learning_rate": 4.594455852156057e-07, | |
| "loss": 2.6306, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 158.02469135802468, | |
| "learning_rate": 4.5482546201232026e-07, | |
| "loss": 2.6137, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 159.25925925925927, | |
| "learning_rate": 4.502053388090349e-07, | |
| "loss": 2.6427, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 160.49382716049382, | |
| "learning_rate": 4.4558521560574943e-07, | |
| "loss": 2.6137, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 161.7283950617284, | |
| "learning_rate": 4.409650924024641e-07, | |
| "loss": 2.6186, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 162.96296296296296, | |
| "learning_rate": 4.363449691991786e-07, | |
| "loss": 2.6397, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 164.19753086419752, | |
| "learning_rate": 4.3172484599589327e-07, | |
| "loss": 2.6389, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 165.4320987654321, | |
| "learning_rate": 4.271047227926078e-07, | |
| "loss": 2.5892, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 166.66666666666666, | |
| "learning_rate": 4.2248459958932234e-07, | |
| "loss": 2.6166, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 167.90123456790124, | |
| "learning_rate": 4.1786447638603695e-07, | |
| "loss": 2.5763, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 169.1358024691358, | |
| "learning_rate": 4.132443531827515e-07, | |
| "loss": 2.643, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 170.37037037037038, | |
| "learning_rate": 4.0862422997946613e-07, | |
| "loss": 2.5972, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 171.60493827160494, | |
| "learning_rate": 4.040041067761807e-07, | |
| "loss": 2.5463, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 172.8395061728395, | |
| "learning_rate": 3.993839835728953e-07, | |
| "loss": 2.6293, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 174.07407407407408, | |
| "learning_rate": 3.947638603696098e-07, | |
| "loss": 2.5722, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 175.30864197530863, | |
| "learning_rate": 3.9014373716632437e-07, | |
| "loss": 2.5601, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 176.54320987654322, | |
| "learning_rate": 3.85523613963039e-07, | |
| "loss": 2.5368, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 177.77777777777777, | |
| "learning_rate": 3.8090349075975354e-07, | |
| "loss": 2.5328, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 179.01234567901236, | |
| "learning_rate": 3.7628336755646816e-07, | |
| "loss": 2.5885, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 180.2469135802469, | |
| "learning_rate": 3.716632443531827e-07, | |
| "loss": 2.5708, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 181.4814814814815, | |
| "learning_rate": 3.6704312114989733e-07, | |
| "loss": 2.548, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 182.71604938271605, | |
| "learning_rate": 3.624229979466119e-07, | |
| "loss": 2.5854, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 183.9506172839506, | |
| "learning_rate": 3.578028747433265e-07, | |
| "loss": 2.5578, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 185.1851851851852, | |
| "learning_rate": 3.5318275154004107e-07, | |
| "loss": 2.6103, | |
| "step": 3000 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 4096, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 256, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8772602674397184.0, | |
| "train_batch_size": 5, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |