| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 100, | |
| "global_step": 753, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 4.347826086956522e-06, | |
| "loss": 4.2511, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 8.695652173913044e-06, | |
| "loss": 4.2595, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0, | |
| "learning_rate": 1.3043478260869566e-05, | |
| "loss": 3.0895, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 1.739130434782609e-05, | |
| "loss": 0.8882, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.173913043478261e-05, | |
| "loss": 0.4819, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 2.608695652173913e-05, | |
| "loss": 0.8914, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.0434782608695656e-05, | |
| "loss": 1.1554, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.478260869565218e-05, | |
| "loss": 0.3562, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 3.91304347826087e-05, | |
| "loss": 0.4085, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.347826086956522e-05, | |
| "loss": 0.2655, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.782608695652174e-05, | |
| "loss": 0.281, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.217391304347826e-05, | |
| "loss": 0.4708, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 5.652173913043478e-05, | |
| "loss": 4.1421, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.086956521739131e-05, | |
| "loss": 2.4328, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.521739130434783e-05, | |
| "loss": 1.7092, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 6.956521739130436e-05, | |
| "loss": 0.687, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.391304347826086e-05, | |
| "loss": 0.2745, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 7.82608695652174e-05, | |
| "loss": 0.3214, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.260869565217392e-05, | |
| "loss": 0.4915, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 8.695652173913044e-05, | |
| "loss": 0.2447, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.130434782608696e-05, | |
| "loss": 0.2353, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.565217391304348e-05, | |
| "loss": 0.2507, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.0001, | |
| "loss": 0.2238, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.99995369868095e-05, | |
| "loss": 0.2327, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.999814795581328e-05, | |
| "loss": 0.5584, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 9.99958329327369e-05, | |
| "loss": 1.6468, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.999259196045582e-05, | |
| "loss": 5.5723, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.998842509899456e-05, | |
| "loss": 4.1912, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.998333242552556e-05, | |
| "loss": 0.9005, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.997731403436788e-05, | |
| "loss": 0.5078, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.997037003698525e-05, | |
| "loss": 0.2464, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.996250056198417e-05, | |
| "loss": 0.258, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 9.995370575511151e-05, | |
| "loss": 0.2438, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.994398577925169e-05, | |
| "loss": 0.2159, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.993334081442381e-05, | |
| "loss": 0.1959, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.992177105777822e-05, | |
| "loss": 0.2217, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.990927672359294e-05, | |
| "loss": 0.1972, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.989585804326962e-05, | |
| "loss": 0.1947, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.988151526532929e-05, | |
| "loss": 0.2028, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.986624865540778e-05, | |
| "loss": 0.1989, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 9.985005849625076e-05, | |
| "loss": 0.1963, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.983294508770851e-05, | |
| "loss": 0.1935, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.981490874673039e-05, | |
| "loss": 0.1949, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.979594980735896e-05, | |
| "loss": 0.1979, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.977606862072377e-05, | |
| "loss": 0.1882, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.975526555503488e-05, | |
| "loss": 0.1911, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.973354099557606e-05, | |
| "loss": 0.1914, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 9.97108953446976e-05, | |
| "loss": 0.1945, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.968732902180891e-05, | |
| "loss": 0.1876, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.966284246337072e-05, | |
| "loss": 0.1893, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.963743612288701e-05, | |
| "loss": 0.1946, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.961111047089662e-05, | |
| "loss": 0.1909, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.95838659949645e-05, | |
| "loss": 0.1959, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.955570319967273e-05, | |
| "loss": 0.1922, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.952662260661115e-05, | |
| "loss": 0.195, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 9.949662475436765e-05, | |
| "loss": 0.1941, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.946571019851832e-05, | |
| "loss": 0.1941, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.943387951161703e-05, | |
| "loss": 0.189, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.940113328318488e-05, | |
| "loss": 0.193, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.936747211969932e-05, | |
| "loss": 0.1918, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.933289664458284e-05, | |
| "loss": 0.1939, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.929740749819152e-05, | |
| "loss": 0.1912, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.926100533780303e-05, | |
| "loss": 0.195, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 9.92236908376046e-05, | |
| "loss": 0.1909, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.91854646886805e-05, | |
| "loss": 0.1926, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.91463275989991e-05, | |
| "loss": 0.1911, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.910628029340003e-05, | |
| "loss": 0.1951, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.906532351358047e-05, | |
| "loss": 0.1877, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.902345801808162e-05, | |
| "loss": 0.1894, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.898068458227451e-05, | |
| "loss": 0.19, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 9.893700399834577e-05, | |
| "loss": 0.1927, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.889241707528285e-05, | |
| "loss": 0.1919, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.88469246388591e-05, | |
| "loss": 0.1906, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.880052753161842e-05, | |
| "loss": 0.1928, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.875322661285975e-05, | |
| "loss": 0.1887, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.870502275862104e-05, | |
| "loss": 0.1919, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.86559168616631e-05, | |
| "loss": 0.1892, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.860590983145307e-05, | |
| "loss": 0.1876, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 9.855500259414753e-05, | |
| "loss": 0.1946, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.850319609257535e-05, | |
| "loss": 0.1889, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.845049128622032e-05, | |
| "loss": 0.1909, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.839688915120322e-05, | |
| "loss": 0.1914, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.834239068026387e-05, | |
| "loss": 0.1923, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.828699688274275e-05, | |
| "loss": 0.1902, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.823070878456217e-05, | |
| "loss": 0.1919, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 9.817352742820744e-05, | |
| "loss": 0.1925, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.811545387270743e-05, | |
| "loss": 0.1871, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.805648919361504e-05, | |
| "loss": 0.1922, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.799663448298724e-05, | |
| "loss": 0.1953, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.793589084936483e-05, | |
| "loss": 0.192, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.787425941775197e-05, | |
| "loss": 0.19, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.781174132959529e-05, | |
| "loss": 0.1866, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.774833774276278e-05, | |
| "loss": 0.1906, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 9.768404983152229e-05, | |
| "loss": 0.1907, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.761887878651987e-05, | |
| "loss": 0.1915, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.755282581475769e-05, | |
| "loss": 0.1903, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.74858921395716e-05, | |
| "loss": 0.1917, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.741807900060858e-05, | |
| "loss": 0.1908, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.734938765380377e-05, | |
| "loss": 0.1921, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.727981937135713e-05, | |
| "loss": 0.1879, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "eval_loss": 0.18334217369556427, | |
| "eval_runtime": 12.1378, | |
| "eval_samples_per_second": 164.775, | |
| "eval_steps_per_second": 0.659, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 9.720937544170999e-05, | |
| "loss": 0.1908, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.713805716952105e-05, | |
| "loss": 0.1892, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.706586587564237e-05, | |
| "loss": 0.1902, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.699280289709478e-05, | |
| "loss": 0.1905, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.691886958704321e-05, | |
| "loss": 0.1894, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.684406731477158e-05, | |
| "loss": 0.1943, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.676839746565743e-05, | |
| "loss": 0.1883, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.669186144114625e-05, | |
| "loss": 0.1935, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 9.661446065872568e-05, | |
| "loss": 0.1914, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.653619655189897e-05, | |
| "loss": 0.1877, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.645707057015871e-05, | |
| "loss": 0.1898, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.637708417895983e-05, | |
| "loss": 0.1898, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.62962388596925e-05, | |
| "loss": 0.1922, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.621453610965467e-05, | |
| "loss": 0.1906, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.613197744202438e-05, | |
| "loss": 0.1881, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 9.604856438583174e-05, | |
| "loss": 0.1894, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.596429848593053e-05, | |
| "loss": 0.1872, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.587918130296969e-05, | |
| "loss": 0.192, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.579321441336436e-05, | |
| "loss": 0.1876, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.57063994092667e-05, | |
| "loss": 0.1914, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.561873789853637e-05, | |
| "loss": 0.1897, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.553023150471082e-05, | |
| "loss": 0.1907, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.544088186697515e-05, | |
| "loss": 0.1905, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 9.53506906401318e-05, | |
| "loss": 0.1918, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.525965949456988e-05, | |
| "loss": 0.1908, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.516779011623422e-05, | |
| "loss": 0.1898, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.50750842065942e-05, | |
| "loss": 0.1871, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.498154348261216e-05, | |
| "loss": 0.1906, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.48871696767117e-05, | |
| "loss": 0.1905, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.479196453674544e-05, | |
| "loss": 0.1906, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 9.469592982596285e-05, | |
| "loss": 0.1906, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.459906732297743e-05, | |
| "loss": 0.1929, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.450137882173384e-05, | |
| "loss": 0.1918, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.440286613147467e-05, | |
| "loss": 0.1895, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.430353107670691e-05, | |
| "loss": 0.189, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.420337549716818e-05, | |
| "loss": 0.1942, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.410240124779268e-05, | |
| "loss": 0.192, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.400061019867679e-05, | |
| "loss": 0.1908, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 9.389800423504442e-05, | |
| "loss": 0.1908, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.379458525721215e-05, | |
| "loss": 0.1892, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.369035518055404e-05, | |
| "loss": 0.1887, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.35853159354661e-05, | |
| "loss": 0.1901, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.347946946733055e-05, | |
| "loss": 0.1908, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.337281773647986e-05, | |
| "loss": 0.1958, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.326536271816031e-05, | |
| "loss": 0.1905, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 9.315710640249558e-05, | |
| "loss": 0.1932, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.304805079444971e-05, | |
| "loss": 0.1894, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.293819791379016e-05, | |
| "loss": 0.1921, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.282754979505018e-05, | |
| "loss": 0.1927, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.271610848749135e-05, | |
| "loss": 0.1971, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.260387605506547e-05, | |
| "loss": 0.1894, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.249085457637641e-05, | |
| "loss": 0.1908, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.237704614464156e-05, | |
| "loss": 0.189, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 9.226245286765316e-05, | |
| "loss": 0.1888, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.214707686773914e-05, | |
| "loss": 0.1883, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.203092028172391e-05, | |
| "loss": 0.1918, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.191398526088873e-05, | |
| "loss": 0.1936, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.179627397093185e-05, | |
| "loss": 0.1901, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.16777885919285e-05, | |
| "loss": 0.1891, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.15585313182904e-05, | |
| "loss": 0.1919, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 9.143850435872518e-05, | |
| "loss": 0.1909, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.131770993619548e-05, | |
| "loss": 0.1936, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.119615028787771e-05, | |
| "loss": 0.1867, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.107382766512071e-05, | |
| "loss": 0.1925, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.095074433340401e-05, | |
| "loss": 0.1902, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.08269025722958e-05, | |
| "loss": 0.1921, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.07023046754109e-05, | |
| "loss": 0.1869, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.057695295036805e-05, | |
| "loss": 0.1901, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 9.045084971874738e-05, | |
| "loss": 0.1903, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.032399731604723e-05, | |
| "loss": 0.1933, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.019639809164106e-05, | |
| "loss": 0.1929, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 9.006805440873383e-05, | |
| "loss": 0.1948, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.993896864431826e-05, | |
| "loss": 0.1917, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.980914318913078e-05, | |
| "loss": 0.1936, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.967858044760737e-05, | |
| "loss": 0.1902, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 8.954728283783886e-05, | |
| "loss": 0.1893, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.941525279152627e-05, | |
| "loss": 0.1925, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.928249275393572e-05, | |
| "loss": 0.1905, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.914900518385314e-05, | |
| "loss": 0.1912, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.901479255353879e-05, | |
| "loss": 0.1889, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.887985734868135e-05, | |
| "loss": 0.189, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.874420206835203e-05, | |
| "loss": 0.191, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.860782922495822e-05, | |
| "loss": 0.1889, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 8.847074134419691e-05, | |
| "loss": 0.1869, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.833294096500797e-05, | |
| "loss": 0.1857, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.819443063952713e-05, | |
| "loss": 0.1865, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.805521293303874e-05, | |
| "loss": 0.1917, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.791529042392813e-05, | |
| "loss": 0.1904, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.777466570363401e-05, | |
| "loss": 0.1861, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.763334137660042e-05, | |
| "loss": 0.1886, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.749132006022844e-05, | |
| "loss": 0.1878, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 8.734860438482777e-05, | |
| "loss": 0.1903, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.720519699356804e-05, | |
| "loss": 0.1881, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.70611005424298e-05, | |
| "loss": 0.1894, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.691631770015533e-05, | |
| "loss": 0.1933, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.677085114819928e-05, | |
| "loss": 0.1882, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.662470358067895e-05, | |
| "loss": 0.1853, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.647787770432439e-05, | |
| "loss": 0.1874, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 8.633037623842828e-05, | |
| "loss": 0.1896, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.618220191479563e-05, | |
| "loss": 0.1891, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "eval_loss": 0.18544606864452362, | |
| "eval_runtime": 11.5227, | |
| "eval_samples_per_second": 173.57, | |
| "eval_steps_per_second": 0.694, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.603335747769306e-05, | |
| "loss": 0.1901, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.58838456837981e-05, | |
| "loss": 0.1894, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.573366930214806e-05, | |
| "loss": 0.1856, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.558283111408873e-05, | |
| "loss": 0.1866, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.5431333913223e-05, | |
| "loss": 0.1885, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.52791805053589e-05, | |
| "loss": 0.1906, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 8.512637370845785e-05, | |
| "loss": 0.1869, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.497291635258235e-05, | |
| "loss": 0.1894, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.481881127984361e-05, | |
| "loss": 0.19, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.466406134434887e-05, | |
| "loss": 0.1851, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.45086694121486e-05, | |
| "loss": 0.1846, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.435263836118335e-05, | |
| "loss": 0.1905, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.419597108123054e-05, | |
| "loss": 0.1872, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 8.403867047385081e-05, | |
| "loss": 0.1867, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.388073945233445e-05, | |
| "loss": 0.1859, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.372218094164728e-05, | |
| "loss": 0.189, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.356299787837658e-05, | |
| "loss": 0.1855, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.340319321067668e-05, | |
| "loss": 0.1844, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.324276989821434e-05, | |
| "loss": 0.1915, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.308173091211391e-05, | |
| "loss": 0.1894, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.292007923490245e-05, | |
| "loss": 0.1874, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 8.275781786045427e-05, | |
| "loss": 0.1887, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.259494979393563e-05, | |
| "loss": 0.1849, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.243147805174908e-05, | |
| "loss": 0.1858, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.226740566147753e-05, | |
| "loss": 0.1859, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.210273566182818e-05, | |
| "loss": 0.1871, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.193747110257637e-05, | |
| "loss": 0.1849, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.177161504450888e-05, | |
| "loss": 0.1886, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 8.160517055936744e-05, | |
| "loss": 0.1858, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.143814072979173e-05, | |
| "loss": 0.186, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.127052864926232e-05, | |
| "loss": 0.1888, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.110233742204339e-05, | |
| "loss": 0.1852, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.093357016312517e-05, | |
| "loss": 0.1865, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.07642299981664e-05, | |
| "loss": 0.1864, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.059432006343623e-05, | |
| "loss": 0.1816, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.042384350575632e-05, | |
| "loss": 0.1834, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 8.025280348244246e-05, | |
| "loss": 0.1859, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 8.008120316124612e-05, | |
| "loss": 0.1831, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.990904572029583e-05, | |
| "loss": 0.1879, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.973633434803817e-05, | |
| "loss": 0.1839, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.956307224317891e-05, | |
| "loss": 0.1888, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.938926261462366e-05, | |
| "loss": 0.1832, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.921490868141843e-05, | |
| "loss": 0.1849, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 7.904001367269004e-05, | |
| "loss": 0.1815, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.886458082758637e-05, | |
| "loss": 0.1827, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.868861339521624e-05, | |
| "loss": 0.1835, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.851211463458936e-05, | |
| "loss": 0.1868, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.833508781455588e-05, | |
| "loss": 0.1837, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.815753621374594e-05, | |
| "loss": 0.182, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.797946312050883e-05, | |
| "loss": 0.1827, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.780087183285223e-05, | |
| "loss": 0.1911, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 7.7621765658381e-05, | |
| "loss": 0.1867, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.744214791423596e-05, | |
| "loss": 0.184, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.726202192703256e-05, | |
| "loss": 0.1837, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.708139103279908e-05, | |
| "loss": 0.1864, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.690025857691498e-05, | |
| "loss": 0.1843, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.671862791404896e-05, | |
| "loss": 0.1843, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.653650240809667e-05, | |
| "loss": 0.1816, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 7.635388543211861e-05, | |
| "loss": 0.1887, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.617078036827752e-05, | |
| "loss": 0.1826, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.59871906077758e-05, | |
| "loss": 0.1876, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.580311955079264e-05, | |
| "loss": 0.1852, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.56185706064212e-05, | |
| "loss": 0.1837, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.543354719260523e-05, | |
| "loss": 0.1813, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.524805273607603e-05, | |
| "loss": 0.1899, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.506209067228878e-05, | |
| "loss": 0.1878, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 7.487566444535903e-05, | |
| "loss": 0.1852, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.468877750799886e-05, | |
| "loss": 0.1879, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.450143332145296e-05, | |
| "loss": 0.1868, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.43136353554345e-05, | |
| "loss": 0.1851, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.412538708806093e-05, | |
| "loss": 0.1835, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.393669200578943e-05, | |
| "loss": 0.1855, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.374755360335253e-05, | |
| "loss": 0.1827, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 7.355797538369321e-05, | |
| "loss": 0.1889, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.336796085790013e-05, | |
| "loss": 0.1799, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.317751354514255e-05, | |
| "loss": 0.1863, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.298663697260522e-05, | |
| "loss": 0.1855, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.279533467542294e-05, | |
| "loss": 0.1851, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.260361019661522e-05, | |
| "loss": 0.1851, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.241146708702053e-05, | |
| "loss": 0.1856, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.221890890523068e-05, | |
| "loss": 0.187, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 7.202593921752475e-05, | |
| "loss": 0.1826, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.18325615978032e-05, | |
| "loss": 0.1866, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.163877962752157e-05, | |
| "loss": 0.1859, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.144459689562418e-05, | |
| "loss": 0.1843, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.12500169984777e-05, | |
| "loss": 0.1833, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.105504353980448e-05, | |
| "loss": 0.1841, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.085968013061584e-05, | |
| "loss": 0.185, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 7.066393038914521e-05, | |
| "loss": 0.1838, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.046779794078108e-05, | |
| "loss": 0.1852, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.027128641799986e-05, | |
| "loss": 0.1832, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 7.007439946029865e-05, | |
| "loss": 0.1899, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.98771407141278e-05, | |
| "loss": 0.1841, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.967951383282334e-05, | |
| "loss": 0.1869, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.94815224765394e-05, | |
| "loss": 0.1847, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.928317031218035e-05, | |
| "loss": 0.1848, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 6.908446101333295e-05, | |
| "loss": 0.1849, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.888539826019825e-05, | |
| "loss": 0.1829, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.868598573952345e-05, | |
| "loss": 0.1849, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.848622714453366e-05, | |
| "loss": 0.1844, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "eval_loss": 0.17592795193195343, | |
| "eval_runtime": 11.447, | |
| "eval_samples_per_second": 174.719, | |
| "eval_steps_per_second": 0.699, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.828612617486347e-05, | |
| "loss": 0.1841, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.808568653648838e-05, | |
| "loss": 0.1843, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.78849119416563e-05, | |
| "loss": 0.1872, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 6.76838061088186e-05, | |
| "loss": 0.1856, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.748237276256143e-05, | |
| "loss": 0.1849, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.728061563353667e-05, | |
| "loss": 0.1817, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.707853845839278e-05, | |
| "loss": 0.1852, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.687614497970566e-05, | |
| "loss": 0.1837, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.667343894590935e-05, | |
| "loss": 0.1832, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.647042411122652e-05, | |
| "loss": 0.1843, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.626710423559901e-05, | |
| "loss": 0.1826, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 6.606348308461823e-05, | |
| "loss": 0.1848, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.585956442945532e-05, | |
| "loss": 0.1848, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.565535204679134e-05, | |
| "loss": 0.1812, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.545084971874738e-05, | |
| "loss": 0.186, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.524606123281445e-05, | |
| "loss": 0.1813, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.504099038178338e-05, | |
| "loss": 0.1814, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.483564096367451e-05, | |
| "loss": 0.1849, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.463001678166744e-05, | |
| "loss": 0.1855, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 6.442412164403045e-05, | |
| "loss": 0.1865, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.42179593640502e-05, | |
| "loss": 0.1828, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.401153375996081e-05, | |
| "loss": 0.1834, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.380484865487347e-05, | |
| "loss": 0.1817, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.359790787670527e-05, | |
| "loss": 0.1855, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.339071525810871e-05, | |
| "loss": 0.1837, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.318327463640037e-05, | |
| "loss": 0.1845, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 6.297558985348998e-05, | |
| "loss": 0.1817, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.276766475580935e-05, | |
| "loss": 0.1869, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.255950319424098e-05, | |
| "loss": 0.1838, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.235110902404679e-05, | |
| "loss": 0.1851, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.21424861047968e-05, | |
| "loss": 0.184, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.193363830029751e-05, | |
| "loss": 0.1847, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.172456947852049e-05, | |
| "loss": 0.1855, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.151528351153061e-05, | |
| "loss": 0.1854, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 6.130578427541441e-05, | |
| "loss": 0.183, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.109607565020828e-05, | |
| "loss": 0.1794, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.08861615198266e-05, | |
| "loss": 0.1869, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.067604577198981e-05, | |
| "loss": 0.188, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.046573229815243e-05, | |
| "loss": 0.1831, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.025522499343097e-05, | |
| "loss": 0.1886, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 6.004452775653178e-05, | |
| "loss": 0.1861, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 5.9833644489678816e-05, | |
| "loss": 0.1817, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.96225790985415e-05, | |
| "loss": 0.1863, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.941133549216221e-05, | |
| "loss": 0.1807, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.919991758288401e-05, | |
| "loss": 0.1846, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.898832928627811e-05, | |
| "loss": 0.1828, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.877657452107142e-05, | |
| "loss": 0.1823, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.856465720907388e-05, | |
| "loss": 0.1843, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.835258127510596e-05, | |
| "loss": 0.183, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 5.8140350646925845e-05, | |
| "loss": 0.1842, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.7927969255156736e-05, | |
| "loss": 0.1845, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.771544103321407e-05, | |
| "loss": 0.1794, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.7502769917232635e-05, | |
| "loss": 0.1782, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.7289959845993736e-05, | |
| "loss": 0.187, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.7077014760852164e-05, | |
| "loss": 0.1829, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.686393860566324e-05, | |
| "loss": 0.1849, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 5.66507353267098e-05, | |
| "loss": 0.1848, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.643740887262905e-05, | |
| "loss": 0.1825, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.6223963194339466e-05, | |
| "loss": 0.1856, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.601040224496764e-05, | |
| "loss": 0.183, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.579672997977503e-05, | |
| "loss": 0.1833, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.5582950356084726e-05, | |
| "loss": 0.1821, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.536906733320816e-05, | |
| "loss": 0.1823, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.515508487237174e-05, | |
| "loss": 0.1845, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 5.494100693664358e-05, | |
| "loss": 0.1825, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.4726837490859964e-05, | |
| "loss": 0.181, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.4512580501552056e-05, | |
| "loss": 0.1839, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.429823993687233e-05, | |
| "loss": 0.1834, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.4083819766521135e-05, | |
| "loss": 0.1843, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.386932396167316e-05, | |
| "loss": 0.1827, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.365475649490388e-05, | |
| "loss": 0.1844, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 5.3440121340116e-05, | |
| "loss": 0.1854, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.3225422472465824e-05, | |
| "loss": 0.1843, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.3010663868289655e-05, | |
| "loss": 0.1815, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.279584950503017e-05, | |
| "loss": 0.1811, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.2580983361162696e-05, | |
| "loss": 0.1856, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.23660694161216e-05, | |
| "loss": 0.1832, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.215111165022652e-05, | |
| "loss": 0.1826, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.193611404460873e-05, | |
| "loss": 0.1826, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 5.172108058113729e-05, | |
| "loss": 0.1845, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.1506015242345415e-05, | |
| "loss": 0.1815, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.129092201135666e-05, | |
| "loss": 0.1848, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.1075804871811115e-05, | |
| "loss": 0.1845, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.086066780779174e-05, | |
| "loss": 0.184, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.064551480375046e-05, | |
| "loss": 0.1851, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.0430349844434424e-05, | |
| "loss": 0.187, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 5.021517691481221e-05, | |
| "loss": 0.1817, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 5e-05, | |
| "loss": 0.1835, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.978482308518779e-05, | |
| "loss": 0.1827, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.956965015556559e-05, | |
| "loss": 0.1839, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.9354485196249554e-05, | |
| "loss": 0.1828, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.9139332192208276e-05, | |
| "loss": 0.1843, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.892419512818889e-05, | |
| "loss": 0.187, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.870907798864337e-05, | |
| "loss": 0.1799, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.8493984757654596e-05, | |
| "loss": 0.1842, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.827891941886273e-05, | |
| "loss": 0.183, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.806388595539129e-05, | |
| "loss": 0.1813, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.784888834977347e-05, | |
| "loss": 0.1822, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.763393058387841e-05, | |
| "loss": 0.1836, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.741901663883731e-05, | |
| "loss": 0.1853, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "eval_loss": 0.1743704080581665, | |
| "eval_runtime": 12.7542, | |
| "eval_samples_per_second": 156.811, | |
| "eval_steps_per_second": 0.627, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.720415049496984e-05, | |
| "loss": 0.1851, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.698933613171035e-05, | |
| "loss": 0.1835, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.6774577527534195e-05, | |
| "loss": 0.185, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.655987865988401e-05, | |
| "loss": 0.1858, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.634524350509613e-05, | |
| "loss": 0.1825, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.613067603832685e-05, | |
| "loss": 0.1885, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.5916180233478883e-05, | |
| "loss": 0.1833, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.5701760063127686e-05, | |
| "loss": 0.1862, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.548741949844795e-05, | |
| "loss": 0.1837, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.527316250914004e-05, | |
| "loss": 0.1873, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.505899306335643e-05, | |
| "loss": 0.1787, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.484491512762827e-05, | |
| "loss": 0.1883, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.463093266679185e-05, | |
| "loss": 0.1806, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.4417049643915286e-05, | |
| "loss": 0.183, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.420327002022498e-05, | |
| "loss": 0.1831, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.398959775503238e-05, | |
| "loss": 0.183, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.377603680566054e-05, | |
| "loss": 0.1782, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.356259112737096e-05, | |
| "loss": 0.1871, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.334926467329021e-05, | |
| "loss": 0.1825, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.313606139433676e-05, | |
| "loss": 0.1838, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.292298523914785e-05, | |
| "loss": 0.1816, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.271004015400627e-05, | |
| "loss": 0.1824, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.249723008276737e-05, | |
| "loss": 0.182, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.228455896678595e-05, | |
| "loss": 0.1817, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.207203074484328e-05, | |
| "loss": 0.1808, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.185964935307417e-05, | |
| "loss": 0.1776, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.164741872489405e-05, | |
| "loss": 0.1831, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.143534279092612e-05, | |
| "loss": 0.18, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.1223425478928594e-05, | |
| "loss": 0.1816, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.1011670713721905e-05, | |
| "loss": 0.1811, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0800082417115995e-05, | |
| "loss": 0.1805, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.05886645078378e-05, | |
| "loss": 0.1814, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0377420901458506e-05, | |
| "loss": 0.1798, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.0166355510321195e-05, | |
| "loss": 0.1826, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.995547224346824e-05, | |
| "loss": 0.1818, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.9744775006569057e-05, | |
| "loss": 0.1812, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.953426770184757e-05, | |
| "loss": 0.1824, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.9323954228010196e-05, | |
| "loss": 0.181, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.911383848017341e-05, | |
| "loss": 0.1829, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 3.890392434979172e-05, | |
| "loss": 0.1849, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.8694215724585594e-05, | |
| "loss": 0.1821, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.848471648846939e-05, | |
| "loss": 0.1822, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.827543052147952e-05, | |
| "loss": 0.1866, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.8066361699702495e-05, | |
| "loss": 0.1826, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.785751389520323e-05, | |
| "loss": 0.1825, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.764889097595322e-05, | |
| "loss": 0.1813, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.744049680575905e-05, | |
| "loss": 0.1817, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 3.7232335244190656e-05, | |
| "loss": 0.1799, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.702441014651001e-05, | |
| "loss": 0.182, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.6816725363599644e-05, | |
| "loss": 0.18, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.6609284741891295e-05, | |
| "loss": 0.1793, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.640209212329473e-05, | |
| "loss": 0.1801, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.619515134512656e-05, | |
| "loss": 0.1798, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.5988466240039206e-05, | |
| "loss": 0.179, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.578204063594982e-05, | |
| "loss": 0.1827, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.5575878355969566e-05, | |
| "loss": 0.1843, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.536998321833258e-05, | |
| "loss": 0.1821, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.516435903632548e-05, | |
| "loss": 0.182, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.495900961821662e-05, | |
| "loss": 0.1792, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.475393876718555e-05, | |
| "loss": 0.1836, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.4549150281252636e-05, | |
| "loss": 0.1849, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.4344647953208675e-05, | |
| "loss": 0.1823, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.41404355705447e-05, | |
| "loss": 0.1785, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.393651691538178e-05, | |
| "loss": 0.181, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.3732895764401004e-05, | |
| "loss": 0.1817, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.35295758887735e-05, | |
| "loss": 0.1807, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.332656105409066e-05, | |
| "loss": 0.1805, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.312385502029434e-05, | |
| "loss": 0.1839, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.2921461541607225e-05, | |
| "loss": 0.1811, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.271938436646334e-05, | |
| "loss": 0.1829, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.251762723743858e-05, | |
| "loss": 0.1818, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.231619389118144e-05, | |
| "loss": 0.1798, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.2115088058343725e-05, | |
| "loss": 0.1813, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.191431346351164e-05, | |
| "loss": 0.18, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.171387382513654e-05, | |
| "loss": 0.1822, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.151377285546635e-05, | |
| "loss": 0.1823, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.1314014260476553e-05, | |
| "loss": 0.1826, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.111460173980175e-05, | |
| "loss": 0.1785, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.091553898666705e-05, | |
| "loss": 0.1844, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.0716829687819643e-05, | |
| "loss": 0.1809, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.051847752346061e-05, | |
| "loss": 0.182, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.0320486167176664e-05, | |
| "loss": 0.1792, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.0122859285872214e-05, | |
| "loss": 0.183, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.9925600539701347e-05, | |
| "loss": 0.1814, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.972871358200015e-05, | |
| "loss": 0.1817, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.9532202059218933e-05, | |
| "loss": 0.1777, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.9336069610854788e-05, | |
| "loss": 0.1837, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.914031986938417e-05, | |
| "loss": 0.1805, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8944956460195515e-05, | |
| "loss": 0.18, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8749983001522302e-05, | |
| "loss": 0.18, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8555403104375827e-05, | |
| "loss": 0.1798, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8361220372478446e-05, | |
| "loss": 0.1829, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 2.8167438402196805e-05, | |
| "loss": 0.1796, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.7974060782475258e-05, | |
| "loss": 0.1786, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.778109109476934e-05, | |
| "loss": 0.1827, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.7588532912979483e-05, | |
| "loss": 0.1789, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.739638980338479e-05, | |
| "loss": 0.1817, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.720466532457707e-05, | |
| "loss": 0.1815, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.70133630273948e-05, | |
| "loss": 0.1788, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 2.6822486454857453e-05, | |
| "loss": 0.1802, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "eval_loss": 0.1759687066078186, | |
| "eval_runtime": 12.5156, | |
| "eval_samples_per_second": 159.801, | |
| "eval_steps_per_second": 0.639, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.6632039142099896e-05, | |
| "loss": 0.1804, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.6442024616306804e-05, | |
| "loss": 0.1814, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.62524463966475e-05, | |
| "loss": 0.1806, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.6063307994210584e-05, | |
| "loss": 0.1818, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.5874612911939094e-05, | |
| "loss": 0.1796, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.5686364644565486e-05, | |
| "loss": 0.1797, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.5498566678547042e-05, | |
| "loss": 0.1803, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 2.531122249200114e-05, | |
| "loss": 0.1814, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.5124335554640967e-05, | |
| "loss": 0.1796, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.493790932771122e-05, | |
| "loss": 0.181, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4751947263923982e-05, | |
| "loss": 0.1802, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4566452807394785e-05, | |
| "loss": 0.1806, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.438142939357882e-05, | |
| "loss": 0.1833, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4196880449207366e-05, | |
| "loss": 0.1803, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 2.4012809392224227e-05, | |
| "loss": 0.1813, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.38292196317225e-05, | |
| "loss": 0.1845, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.3646114567881393e-05, | |
| "loss": 0.1793, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.3463497591903322e-05, | |
| "loss": 0.1806, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.328137208595107e-05, | |
| "loss": 0.1822, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.309974142308502e-05, | |
| "loss": 0.179, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.291860896720094e-05, | |
| "loss": 0.181, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.273797807296744e-05, | |
| "loss": 0.1832, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 2.2557852085764053e-05, | |
| "loss": 0.1801, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.237823434161902e-05, | |
| "loss": 0.1807, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.2199128167147785e-05, | |
| "loss": 0.1812, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.2020536879491167e-05, | |
| "loss": 0.1777, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.184246378625407e-05, | |
| "loss": 0.1801, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1664912185444124e-05, | |
| "loss": 0.1759, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1487885365410644e-05, | |
| "loss": 0.184, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 2.1311386604783763e-05, | |
| "loss": 0.1844, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.113541917241364e-05, | |
| "loss": 0.1787, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0959986327309968e-05, | |
| "loss": 0.1798, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0785091318581577e-05, | |
| "loss": 0.178, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.061073738537635e-05, | |
| "loss": 0.18, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0436927756821094e-05, | |
| "loss": 0.1792, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0263665651961834e-05, | |
| "loss": 0.1813, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 2.0090954279704183e-05, | |
| "loss": 0.1829, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.9918796838753863e-05, | |
| "loss": 0.1773, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.974719651755756e-05, | |
| "loss": 0.1797, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.957615649424369e-05, | |
| "loss": 0.1785, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9405679936563786e-05, | |
| "loss": 0.1825, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9235770001833607e-05, | |
| "loss": 0.1785, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.9066429836874844e-05, | |
| "loss": 0.176, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.889766257795663e-05, | |
| "loss": 0.1788, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.8729471350737693e-05, | |
| "loss": 0.1807, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.856185927020827e-05, | |
| "loss": 0.1799, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8394829440632565e-05, | |
| "loss": 0.182, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.8228384955491135e-05, | |
| "loss": 0.1825, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.806252889742364e-05, | |
| "loss": 0.1755, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.789726433817182e-05, | |
| "loss": 0.1777, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7732594338522495e-05, | |
| "loss": 0.1823, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.756852194825094e-05, | |
| "loss": 0.1793, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.7405050206064373e-05, | |
| "loss": 0.1794, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7242182139545744e-05, | |
| "loss": 0.1745, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.7079920765097563e-05, | |
| "loss": 0.1771, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.691826908788608e-05, | |
| "loss": 0.1755, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.675723010178568e-05, | |
| "loss": 0.1806, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.6596806789323315e-05, | |
| "loss": 0.1818, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.6437002121623436e-05, | |
| "loss": 0.1838, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.6277819058352727e-05, | |
| "loss": 0.1775, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.611926054766556e-05, | |
| "loss": 0.1781, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.596132952614918e-05, | |
| "loss": 0.1791, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5804028918769485e-05, | |
| "loss": 0.1794, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5647361638816655e-05, | |
| "loss": 0.1817, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.549133058785141e-05, | |
| "loss": 0.1797, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5335938655651122e-05, | |
| "loss": 0.1815, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5181188720156392e-05, | |
| "loss": 0.1824, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.5027083647417655e-05, | |
| "loss": 0.1785, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4873626291542148e-05, | |
| "loss": 0.1811, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4720819494641109e-05, | |
| "loss": 0.1794, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4568666086777021e-05, | |
| "loss": 0.183, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4417168885911276e-05, | |
| "loss": 0.1773, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4266330697851954e-05, | |
| "loss": 0.176, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.4116154316201907e-05, | |
| "loss": 0.1754, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.3966642522306956e-05, | |
| "loss": 0.1771, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.381779808520438e-05, | |
| "loss": 0.1816, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3669623761571726e-05, | |
| "loss": 0.1799, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3522122295675616e-05, | |
| "loss": 0.1761, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.337529641932107e-05, | |
| "loss": 0.1803, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.3229148851800721e-05, | |
| "loss": 0.1811, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.308368229984468e-05, | |
| "loss": 0.1794, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2938899457570208e-05, | |
| "loss": 0.1805, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.2794803006431982e-05, | |
| "loss": 0.1784, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.265139561517224e-05, | |
| "loss": 0.1807, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2508679939771583e-05, | |
| "loss": 0.1771, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2366658623399584e-05, | |
| "loss": 0.1801, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2225334296365987e-05, | |
| "loss": 0.1794, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.2084709576071884e-05, | |
| "loss": 0.1778, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1944787066961265e-05, | |
| "loss": 0.1818, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.1805569360472868e-05, | |
| "loss": 0.181, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.166705903499205e-05, | |
| "loss": 0.1812, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.1529258655803116e-05, | |
| "loss": 0.1804, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.1392170775041788e-05, | |
| "loss": 0.1799, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.1255797931647971e-05, | |
| "loss": 0.1758, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.1120142651318665e-05, | |
| "loss": 0.1787, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0985207446461221e-05, | |
| "loss": 0.1805, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0850994816146858e-05, | |
| "loss": 0.1797, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.0717507246064273e-05, | |
| "loss": 0.1815, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0584747208473739e-05, | |
| "loss": 0.1799, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0452717162161141e-05, | |
| "loss": 0.1798, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "eval_loss": 0.1740500032901764, | |
| "eval_runtime": 12.3559, | |
| "eval_samples_per_second": 161.866, | |
| "eval_steps_per_second": 0.647, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0321419552392636e-05, | |
| "loss": 0.1745, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0190856810869215e-05, | |
| "loss": 0.1753, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.0061031355681766e-05, | |
| "loss": 0.1838, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.931945591266174e-06, | |
| "loss": 0.1854, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.803601908358944e-06, | |
| "loss": 0.1819, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 9.676002683952768e-06, | |
| "loss": 0.1781, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.549150281252633e-06, | |
| "loss": 0.1789, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.423047049631956e-06, | |
| "loss": 0.1802, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.297695324589106e-06, | |
| "loss": 0.1771, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.173097427704203e-06, | |
| "loss": 0.1806, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 9.049255666596012e-06, | |
| "loss": 0.1805, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.926172334879296e-06, | |
| "loss": 0.178, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 8.803849712122292e-06, | |
| "loss": 0.1794, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.682290063804526e-06, | |
| "loss": 0.1736, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.561495641274825e-06, | |
| "loss": 0.1759, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.441468681709602e-06, | |
| "loss": 0.1752, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.322211408071512e-06, | |
| "loss": 0.181, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.203726029068148e-06, | |
| "loss": 0.1796, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 8.086014739111297e-06, | |
| "loss": 0.1765, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.969079718276096e-06, | |
| "loss": 0.1818, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 7.852923132260864e-06, | |
| "loss": 0.1747, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.73754713234684e-06, | |
| "loss": 0.1779, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.6229538553584556e-06, | |
| "loss": 0.1817, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.509145423623609e-06, | |
| "loss": 0.1778, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.3961239449345355e-06, | |
| "loss": 0.1847, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.2838915125086504e-06, | |
| "loss": 0.1789, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.172450204949821e-06, | |
| "loss": 0.1795, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 7.061802086209857e-06, | |
| "loss": 0.1791, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.951949205550284e-06, | |
| "loss": 0.1826, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.842893597504435e-06, | |
| "loss": 0.177, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.734637281839701e-06, | |
| "loss": 0.1802, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.627182263520165e-06, | |
| "loss": 0.1785, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.52053053266945e-06, | |
| "loss": 0.1793, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.41468406453391e-06, | |
| "loss": 0.1798, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.30964481944597e-06, | |
| "loss": 0.1783, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 6.205414742787852e-06, | |
| "loss": 0.1783, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 6.1019957649555985e-06, | |
| "loss": 0.1762, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.999389801323219e-06, | |
| "loss": 0.1808, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.897598752207328e-06, | |
| "loss": 0.1756, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.796624502831821e-06, | |
| "loss": 0.1791, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.696468923293108e-06, | |
| "loss": 0.1774, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.59713386852534e-06, | |
| "loss": 0.1781, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 5.498621178266167e-06, | |
| "loss": 0.1763, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.40093267702258e-06, | |
| "loss": 0.1749, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.304070174037146e-06, | |
| "loss": 0.1769, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.208035463254557e-06, | |
| "loss": 0.1763, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.112830323288314e-06, | |
| "loss": 0.1813, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 5.018456517387837e-06, | |
| "loss": 0.1764, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.924915793405799e-06, | |
| "loss": 0.1795, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.832209883765782e-06, | |
| "loss": 0.1787, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 4.74034050543013e-06, | |
| "loss": 0.1728, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.649309359868209e-06, | |
| "loss": 0.1754, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.559118133024853e-06, | |
| "loss": 0.1747, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.4697684952891895e-06, | |
| "loss": 0.1775, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.38126210146364e-06, | |
| "loss": 0.1758, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.29360059073331e-06, | |
| "loss": 0.182, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.2067855866356475e-06, | |
| "loss": 0.1763, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.12081869703031e-06, | |
| "loss": 0.1756, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.035701514069484e-06, | |
| "loss": 0.1768, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.951435614168275e-06, | |
| "loss": 0.18, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.868022557975626e-06, | |
| "loss": 0.1781, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.7854638903453367e-06, | |
| "loss": 0.178, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.70376114030751e-06, | |
| "loss": 0.1777, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.6229158210401737e-06, | |
| "loss": 0.1773, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.5429294298412853e-06, | |
| "loss": 0.1762, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.463803448101033e-06, | |
| "loss": 0.179, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.3855393412743352e-06, | |
| "loss": 0.1788, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.308138558853746e-06, | |
| "loss": 0.1796, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.231602534342587e-06, | |
| "loss": 0.1791, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.15593268522843e-06, | |
| "loss": 0.1765, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.081130412956795e-06, | |
| "loss": 0.1771, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.0071971029052348e-06, | |
| "loss": 0.1765, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 2.934134124357646e-06, | |
| "loss": 0.1746, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.86194283047897e-06, | |
| "loss": 0.177, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.7906245582900338e-06, | |
| "loss": 0.1771, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.720180628642871e-06, | |
| "loss": 0.1756, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.6506123461962408e-06, | |
| "loss": 0.1767, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.5819209993914184e-06, | |
| "loss": 0.1774, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.5141078604284105e-06, | |
| "loss": 0.1796, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.4471741852423237e-06, | |
| "loss": 0.1821, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 2.381121213480131e-06, | |
| "loss": 0.1815, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.3159501684777207e-06, | |
| "loss": 0.1796, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.2516622572372414e-06, | |
| "loss": 0.1771, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.188258670404719e-06, | |
| "loss": 0.1796, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.125740582248037e-06, | |
| "loss": 0.1805, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.0641091506351796e-06, | |
| "loss": 0.1793, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 2.003365517012773e-06, | |
| "loss": 0.1804, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 1.943510806384968e-06, | |
| "loss": 0.1777, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.884546127292569e-06, | |
| "loss": 0.176, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.8264725717925591e-06, | |
| "loss": 0.1745, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.7692912154378294e-06, | |
| "loss": 0.1754, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.7130031172572614e-06, | |
| "loss": 0.1777, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6576093197361253e-06, | |
| "loss": 0.1799, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.603110848796785e-06, | |
| "loss": 0.1751, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.549508713779696e-06, | |
| "loss": 0.1756, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.4968039074246486e-06, | |
| "loss": 0.1772, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.444997405852483e-06, | |
| "loss": 0.1752, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3940901685469298e-06, | |
| "loss": 0.181, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.3440831383369046e-06, | |
| "loss": 0.1781, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.294977241378975e-06, | |
| "loss": 0.1766, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "eval_loss": 0.1736355423927307, | |
| "eval_runtime": 12.0759, | |
| "eval_samples_per_second": 165.62, | |
| "eval_steps_per_second": 0.662, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.2467733871402655e-06, | |
| "loss": 0.1771, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.199472468381585e-06, | |
| "loss": 0.1788, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.153075361140915e-06, | |
| "loss": 0.1777, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 1.1075829247171598e-06, | |
| "loss": 0.1747, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0629960016542373e-06, | |
| "loss": 0.1772, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.0193154177254971e-06, | |
| "loss": 0.1758, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.765419819183997e-07, | |
| "loss": 0.1785, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 9.346764864195335e-07, | |
| "loss": 0.1782, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.937197065999714e-07, | |
| "loss": 0.1784, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.536724010008878e-07, | |
| "loss": 0.178, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 8.145353113195142e-07, | |
| "loss": 0.1757, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.763091623953889e-07, | |
| "loss": 0.1776, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.389946621969679e-07, | |
| "loss": 0.176, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 7.0259250180848e-07, | |
| "loss": 0.1766, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.67103355417148e-07, | |
| "loss": 0.1761, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 6.325278803006818e-07, | |
| "loss": 0.1734, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.988667168151219e-07, | |
| "loss": 0.1789, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.661204883829763e-07, | |
| "loss": 0.1748, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 5.342898014816855e-07, | |
| "loss": 0.1772, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 5.033752456323482e-07, | |
| "loss": 0.1772, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.73377393388863e-07, | |
| "loss": 0.1761, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.4429680032726517e-07, | |
| "loss": 0.1796, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.1613400503550114e-07, | |
| "loss": 0.1789, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.888895291033867e-07, | |
| "loss": 0.1789, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.6256387711299247e-07, | |
| "loss": 0.179, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.371575366292845e-07, | |
| "loss": 0.1768, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.126709781910986e-07, | |
| "loss": 0.1772, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.8910465530240793e-07, | |
| "loss": 0.1805, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.6645900442394677e-07, | |
| "loss": 0.1769, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.4473444496512233e-07, | |
| "loss": 0.1796, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.2393137927623763e-07, | |
| "loss": 0.1773, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 2.0405019264104762e-07, | |
| "loss": 0.1774, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.850912532696092e-07, | |
| "loss": 0.1786, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.6705491229149218e-07, | |
| "loss": 0.1756, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.4994150374924575e-07, | |
| "loss": 0.1771, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.337513445922256e-07, | |
| "loss": 0.1763, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.184847346707152e-07, | |
| "loss": 0.1767, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 1.041419567303914e-07, | |
| "loss": 0.173, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 9.072327640706756e-08, | |
| "loss": 0.1774, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 7.822894222178633e-08, | |
| "loss": 0.1801, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 6.665918557620665e-08, | |
| "loss": 0.1772, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 5.6014220748318303e-08, | |
| "loss": 0.1734, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.6294244888500645e-08, | |
| "loss": 0.1778, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.749943801582556e-08, | |
| "loss": 0.1755, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.9629963014760065e-08, | |
| "loss": 0.1782, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 2.2685965632135432e-08, | |
| "loss": 0.1765, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.666757447443823e-08, | |
| "loss": 0.1809, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.1574901005456662e-08, | |
| "loss": 0.1772, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 7.4080395441877834e-09, | |
| "loss": 0.175, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.167067263105562e-09, | |
| "loss": 0.176, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.8520441867231341e-09, | |
| "loss": 0.1788, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 4.630131904936885e-10, | |
| "loss": 0.1729, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.1755, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 753, | |
| "total_flos": 0.0, | |
| "train_loss": 0.10856808526107514, | |
| "train_runtime": 9051.9774, | |
| "train_samples_per_second": 85.227, | |
| "train_steps_per_second": 0.083 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 753, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 100, | |
| "total_flos": 0.0, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |