diff --git "a/trainer_state.json" "b/trainer_state.json" --- "a/trainer_state.json" +++ "b/trainer_state.json" @@ -1,4174 +1,8312 @@ { "best_metric": null, "best_model_checkpoint": null, - "epoch": 0.9999153188246253, + "epoch": 1.9996613039796782, "eval_steps": 500, - "global_step": 2952, + "global_step": 5904, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, - "grad_norm": 0.439453125, - "learning_rate": 3.378378378378379e-07, - "loss": 2.6105, + "grad_norm": 0.671875, + "learning_rate": 3.38409475465313e-07, + "loss": 2.4832, "step": 1 }, { "epoch": 0.0, - "grad_norm": 0.55078125, - "learning_rate": 1.6891891891891894e-06, - "loss": 2.5818, + "grad_norm": 0.546875, + "learning_rate": 1.6920473773265652e-06, + "loss": 2.5283, "step": 5 }, { "epoch": 0.0, - "grad_norm": 0.482421875, - "learning_rate": 3.3783783783783788e-06, - "loss": 2.5823, + "grad_norm": 0.71484375, + "learning_rate": 3.3840947546531303e-06, + "loss": 2.5147, "step": 10 }, { "epoch": 0.01, - "grad_norm": 0.546875, - "learning_rate": 5.067567567567568e-06, - "loss": 2.5736, + "grad_norm": 0.8671875, + "learning_rate": 5.076142131979695e-06, + "loss": 2.5183, "step": 15 }, { "epoch": 0.01, - "grad_norm": 0.53125, - "learning_rate": 6.7567567567567575e-06, - "loss": 2.5517, + "grad_norm": 0.88671875, + "learning_rate": 6.768189509306261e-06, + "loss": 2.5027, "step": 20 }, { "epoch": 0.01, - "grad_norm": 0.51171875, - "learning_rate": 8.445945945945946e-06, - "loss": 2.5821, + "grad_norm": 0.76953125, + "learning_rate": 8.460236886632826e-06, + "loss": 2.5038, "step": 25 }, { "epoch": 0.01, - "grad_norm": 0.6796875, - "learning_rate": 1.0135135135135136e-05, - "loss": 2.526, + "grad_norm": 0.59375, + "learning_rate": 1.015228426395939e-05, + "loss": 2.4691, "step": 30 }, { "epoch": 0.01, - "grad_norm": 0.5625, - "learning_rate": 1.1824324324324325e-05, - "loss": 2.5476, + "grad_norm": 0.5546875, + "learning_rate": 1.1844331641285957e-05, + "loss": 2.4495, "step": 35 }, { "epoch": 0.01, - "grad_norm": 0.40234375, - "learning_rate": 1.3513513513513515e-05, - "loss": 2.492, + "grad_norm": 0.470703125, + "learning_rate": 1.3536379018612521e-05, + "loss": 2.4366, "step": 40 }, { "epoch": 0.02, - "grad_norm": 0.345703125, - "learning_rate": 1.5202702702702704e-05, - "loss": 2.5255, + "grad_norm": 0.349609375, + "learning_rate": 1.5228426395939088e-05, + "loss": 2.4702, "step": 45 }, { "epoch": 0.02, - "grad_norm": 0.283203125, - "learning_rate": 1.6891891891891892e-05, - "loss": 2.4882, + "grad_norm": 0.259765625, + "learning_rate": 1.6920473773265652e-05, + "loss": 2.4238, "step": 50 }, { "epoch": 0.02, - "grad_norm": 0.2412109375, - "learning_rate": 1.8581081081081082e-05, - "loss": 2.4763, + "grad_norm": 0.2265625, + "learning_rate": 1.8612521150592218e-05, + "loss": 2.402, "step": 55 }, { "epoch": 0.02, - "grad_norm": 0.2138671875, - "learning_rate": 2.0270270270270273e-05, - "loss": 2.4817, + "grad_norm": 0.19140625, + "learning_rate": 2.030456852791878e-05, + "loss": 2.3977, "step": 60 }, { "epoch": 0.02, - "grad_norm": 0.193359375, - "learning_rate": 2.195945945945946e-05, - "loss": 2.491, + "grad_norm": 0.1708984375, + "learning_rate": 2.199661590524535e-05, + "loss": 2.3896, "step": 65 }, { "epoch": 0.02, - "grad_norm": 0.1943359375, - "learning_rate": 2.364864864864865e-05, - "loss": 2.4636, + "grad_norm": 0.17578125, + "learning_rate": 2.3688663282571914e-05, + "loss": 2.4104, "step": 70 }, { "epoch": 0.03, - "grad_norm": 0.216796875, - "learning_rate": 2.533783783783784e-05, - "loss": 2.4359, + "grad_norm": 0.1552734375, + "learning_rate": 2.5380710659898476e-05, + "loss": 2.3876, "step": 75 }, { "epoch": 0.03, - "grad_norm": 0.2099609375, - "learning_rate": 2.702702702702703e-05, - "loss": 2.4172, + "grad_norm": 0.1630859375, + "learning_rate": 2.7072758037225043e-05, + "loss": 2.4002, "step": 80 }, { "epoch": 0.03, - "grad_norm": 0.177734375, - "learning_rate": 2.8716216216216217e-05, - "loss": 2.395, + "grad_norm": 0.1650390625, + "learning_rate": 2.876480541455161e-05, + "loss": 2.3435, "step": 85 }, { "epoch": 0.03, - "grad_norm": 0.177734375, - "learning_rate": 3.0405405405405407e-05, - "loss": 2.3687, + "grad_norm": 0.1806640625, + "learning_rate": 3.0456852791878175e-05, + "loss": 2.3393, "step": 90 }, { "epoch": 0.03, - "grad_norm": 0.1650390625, - "learning_rate": 3.20945945945946e-05, - "loss": 2.3712, + "grad_norm": 0.1611328125, + "learning_rate": 3.214890016920474e-05, + "loss": 2.3178, "step": 95 }, { "epoch": 0.03, - "grad_norm": 0.1630859375, - "learning_rate": 3.3783783783783784e-05, - "loss": 2.3344, + "grad_norm": 0.150390625, + "learning_rate": 3.3840947546531304e-05, + "loss": 2.3124, "step": 100 }, { "epoch": 0.04, - "grad_norm": 0.1630859375, - "learning_rate": 3.547297297297297e-05, - "loss": 2.3406, + "grad_norm": 0.1552734375, + "learning_rate": 3.553299492385787e-05, + "loss": 2.305, "step": 105 }, { "epoch": 0.04, - "grad_norm": 0.18359375, - "learning_rate": 3.7162162162162165e-05, - "loss": 2.286, + "grad_norm": 0.138671875, + "learning_rate": 3.7225042301184437e-05, + "loss": 2.3204, "step": 110 }, { "epoch": 0.04, - "grad_norm": 0.1591796875, - "learning_rate": 3.885135135135135e-05, - "loss": 2.2872, + "grad_norm": 0.1416015625, + "learning_rate": 3.8917089678510996e-05, + "loss": 2.293, "step": 115 }, { "epoch": 0.04, - "grad_norm": 0.1572265625, - "learning_rate": 4.0540540540540545e-05, - "loss": 2.3243, + "grad_norm": 0.1435546875, + "learning_rate": 4.060913705583756e-05, + "loss": 2.3071, "step": 120 }, { "epoch": 0.04, - "grad_norm": 0.1357421875, - "learning_rate": 4.222972972972973e-05, - "loss": 2.2734, + "grad_norm": 0.138671875, + "learning_rate": 4.230118443316413e-05, + "loss": 2.273, "step": 125 }, { "epoch": 0.04, - "grad_norm": 0.13671875, - "learning_rate": 4.391891891891892e-05, - "loss": 2.2828, + "grad_norm": 0.134765625, + "learning_rate": 4.39932318104907e-05, + "loss": 2.2847, "step": 130 }, { "epoch": 0.05, - "grad_norm": 0.1376953125, - "learning_rate": 4.560810810810811e-05, - "loss": 2.2821, + "grad_norm": 0.134765625, + "learning_rate": 4.568527918781726e-05, + "loss": 2.2559, "step": 135 }, { "epoch": 0.05, "grad_norm": 0.1328125, - "learning_rate": 4.72972972972973e-05, - "loss": 2.2867, + "learning_rate": 4.737732656514383e-05, + "loss": 2.2525, "step": 140 }, { "epoch": 0.05, - "grad_norm": 0.1337890625, - "learning_rate": 4.8986486486486486e-05, - "loss": 2.2866, + "grad_norm": 0.1298828125, + "learning_rate": 4.906937394247039e-05, + "loss": 2.2716, "step": 145 }, { "epoch": 0.05, - "grad_norm": 0.1337890625, - "learning_rate": 5.067567567567568e-05, - "loss": 2.2754, + "grad_norm": 0.1318359375, + "learning_rate": 5.076142131979695e-05, + "loss": 2.2682, "step": 150 }, { "epoch": 0.05, - "grad_norm": 0.1357421875, - "learning_rate": 5.2364864864864873e-05, - "loss": 2.2497, + "grad_norm": 0.1279296875, + "learning_rate": 5.245346869712352e-05, + "loss": 2.2575, "step": 155 }, { "epoch": 0.05, - "grad_norm": 0.13671875, - "learning_rate": 5.405405405405406e-05, - "loss": 2.2668, + "grad_norm": 0.130859375, + "learning_rate": 5.4145516074450085e-05, + "loss": 2.2448, "step": 160 }, { "epoch": 0.06, "grad_norm": 0.1318359375, - "learning_rate": 5.574324324324325e-05, - "loss": 2.2712, + "learning_rate": 5.583756345177665e-05, + "loss": 2.211, "step": 165 }, { "epoch": 0.06, - "grad_norm": 0.1337890625, - "learning_rate": 5.7432432432432434e-05, - "loss": 2.2803, + "grad_norm": 0.1396484375, + "learning_rate": 5.752961082910322e-05, + "loss": 2.2194, "step": 170 }, { "epoch": 0.06, - "grad_norm": 0.1337890625, - "learning_rate": 5.912162162162163e-05, - "loss": 2.275, + "grad_norm": 0.13671875, + "learning_rate": 5.9221658206429784e-05, + "loss": 2.2511, "step": 175 }, { "epoch": 0.06, - "grad_norm": 0.138671875, - "learning_rate": 6.0810810810810814e-05, - "loss": 2.2391, + "grad_norm": 0.1357421875, + "learning_rate": 6.091370558375635e-05, + "loss": 2.2315, "step": 180 }, { "epoch": 0.06, - "grad_norm": 0.13671875, - "learning_rate": 6.25e-05, - "loss": 2.2509, + "grad_norm": 0.142578125, + "learning_rate": 6.26057529610829e-05, + "loss": 2.2351, "step": 185 }, { "epoch": 0.06, - "grad_norm": 0.1416015625, - "learning_rate": 6.41891891891892e-05, - "loss": 2.2538, + "grad_norm": 0.140625, + "learning_rate": 6.429780033840948e-05, + "loss": 2.2085, "step": 190 }, { "epoch": 0.07, - "grad_norm": 0.150390625, - "learning_rate": 6.587837837837837e-05, - "loss": 2.2511, + "grad_norm": 0.140625, + "learning_rate": 6.598984771573604e-05, + "loss": 2.2269, "step": 195 }, { "epoch": 0.07, - "grad_norm": 0.1455078125, - "learning_rate": 6.756756756756757e-05, - "loss": 2.2711, + "grad_norm": 0.1416015625, + "learning_rate": 6.768189509306261e-05, + "loss": 2.2031, "step": 200 }, { "epoch": 0.07, - "grad_norm": 0.13671875, - "learning_rate": 6.925675675675676e-05, - "loss": 2.2703, + "grad_norm": 0.14453125, + "learning_rate": 6.937394247038918e-05, + "loss": 2.212, "step": 205 }, { "epoch": 0.07, - "grad_norm": 0.1328125, - "learning_rate": 7.094594594594594e-05, - "loss": 2.2641, + "grad_norm": 0.1376953125, + "learning_rate": 7.106598984771574e-05, + "loss": 2.2063, "step": 210 }, { "epoch": 0.07, - "grad_norm": 0.1484375, - "learning_rate": 7.263513513513514e-05, - "loss": 2.2312, + "grad_norm": 0.1396484375, + "learning_rate": 7.275803722504231e-05, + "loss": 2.2315, "step": 215 }, { "epoch": 0.07, - "grad_norm": 0.1484375, - "learning_rate": 7.432432432432433e-05, - "loss": 2.2609, + "grad_norm": 0.142578125, + "learning_rate": 7.445008460236887e-05, + "loss": 2.2089, "step": 220 }, { "epoch": 0.08, "grad_norm": 0.1435546875, - "learning_rate": 7.601351351351351e-05, - "loss": 2.2292, + "learning_rate": 7.614213197969543e-05, + "loss": 2.2178, "step": 225 }, { "epoch": 0.08, - "grad_norm": 0.1416015625, - "learning_rate": 7.77027027027027e-05, - "loss": 2.2355, + "grad_norm": 0.1455078125, + "learning_rate": 7.783417935702199e-05, + "loss": 2.2294, "step": 230 }, { "epoch": 0.08, - "grad_norm": 0.15234375, - "learning_rate": 7.93918918918919e-05, - "loss": 2.2207, + "grad_norm": 0.150390625, + "learning_rate": 7.952622673434857e-05, + "loss": 2.2275, "step": 235 }, { "epoch": 0.08, - "grad_norm": 0.1474609375, - "learning_rate": 8.108108108108109e-05, - "loss": 2.2336, + "grad_norm": 0.150390625, + "learning_rate": 8.121827411167512e-05, + "loss": 2.1849, "step": 240 }, { "epoch": 0.08, - "grad_norm": 0.150390625, - "learning_rate": 8.277027027027028e-05, - "loss": 2.2604, + "grad_norm": 0.15234375, + "learning_rate": 8.29103214890017e-05, + "loss": 2.1791, "step": 245 }, { "epoch": 0.08, - "grad_norm": 0.150390625, - "learning_rate": 8.445945945945946e-05, - "loss": 2.2512, + "grad_norm": 0.1494140625, + "learning_rate": 8.460236886632826e-05, + "loss": 2.2004, "step": 250 }, { "epoch": 0.09, - "grad_norm": 0.1552734375, - "learning_rate": 8.614864864864866e-05, - "loss": 2.226, + "grad_norm": 0.1494140625, + "learning_rate": 8.629441624365483e-05, + "loss": 2.23, "step": 255 }, { "epoch": 0.09, - "grad_norm": 0.1728515625, - "learning_rate": 8.783783783783784e-05, - "loss": 2.2042, + "grad_norm": 0.1435546875, + "learning_rate": 8.79864636209814e-05, + "loss": 2.2296, "step": 260 }, { "epoch": 0.09, - "grad_norm": 0.1669921875, - "learning_rate": 8.952702702702703e-05, - "loss": 2.224, + "grad_norm": 0.1513671875, + "learning_rate": 8.967851099830795e-05, + "loss": 2.2362, "step": 265 }, { "epoch": 0.09, - "grad_norm": 0.1611328125, - "learning_rate": 9.121621621621623e-05, - "loss": 2.224, + "grad_norm": 0.1494140625, + "learning_rate": 9.137055837563452e-05, + "loss": 2.2202, "step": 270 }, { "epoch": 0.09, - "grad_norm": 0.1611328125, - "learning_rate": 9.29054054054054e-05, - "loss": 2.2425, + "grad_norm": 0.15234375, + "learning_rate": 9.306260575296108e-05, + "loss": 2.1897, "step": 275 }, { "epoch": 0.09, - "grad_norm": 0.1513671875, - "learning_rate": 9.45945945945946e-05, - "loss": 2.231, + "grad_norm": 0.1474609375, + "learning_rate": 9.475465313028765e-05, + "loss": 2.1788, "step": 280 }, { "epoch": 0.1, - "grad_norm": 0.150390625, - "learning_rate": 9.628378378378379e-05, - "loss": 2.2299, + "grad_norm": 0.1552734375, + "learning_rate": 9.644670050761421e-05, + "loss": 2.1784, "step": 285 }, { "epoch": 0.1, - "grad_norm": 0.154296875, - "learning_rate": 9.797297297297297e-05, - "loss": 2.2135, + "grad_norm": 0.150390625, + "learning_rate": 9.813874788494079e-05, + "loss": 2.1903, "step": 290 }, { "epoch": 0.1, - "grad_norm": 0.1591796875, - "learning_rate": 9.966216216216217e-05, - "loss": 2.2346, + "grad_norm": 0.1474609375, + "learning_rate": 9.983079526226735e-05, + "loss": 2.1736, "step": 295 }, { "epoch": 0.1, - "grad_norm": 0.1611328125, - "learning_rate": 9.999944036768366e-05, - "loss": 2.2271, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001015228426395939, + "loss": 2.1937, "step": 300 }, { "epoch": 0.1, - "grad_norm": 0.1533203125, - "learning_rate": 9.999716688286903e-05, - "loss": 2.2334, + "grad_norm": 0.150390625, + "learning_rate": 0.00010321489001692048, + "loss": 2.1986, "step": 305 }, { - "epoch": 0.11, - "grad_norm": 0.150390625, - "learning_rate": 9.999314464799477e-05, - "loss": 2.2153, + "epoch": 0.1, + "grad_norm": 0.1591796875, + "learning_rate": 0.00010490693739424704, + "loss": 2.1945, "step": 310 }, { "epoch": 0.11, - "grad_norm": 0.1650390625, - "learning_rate": 9.998737380374656e-05, - "loss": 2.2373, + "grad_norm": 0.1591796875, + "learning_rate": 0.00010659898477157362, + "loss": 2.2217, "step": 315 }, { "epoch": 0.11, - "grad_norm": 0.1669921875, - "learning_rate": 9.997985455197114e-05, - "loss": 2.2456, + "grad_norm": 0.1572265625, + "learning_rate": 0.00010829103214890017, + "loss": 2.2081, "step": 320 }, { "epoch": 0.11, - "grad_norm": 0.1640625, - "learning_rate": 9.99705871556692e-05, - "loss": 2.2055, + "grad_norm": 0.1572265625, + "learning_rate": 0.00010998307952622673, + "loss": 2.1782, "step": 325 }, { "epoch": 0.11, - "grad_norm": 0.1591796875, - "learning_rate": 9.995957193898633e-05, - "loss": 2.1951, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001116751269035533, + "loss": 2.1853, "step": 330 }, { "epoch": 0.11, - "grad_norm": 0.1591796875, - "learning_rate": 9.994680928720159e-05, - "loss": 2.2664, + "grad_norm": 0.158203125, + "learning_rate": 0.00011336717428087986, + "loss": 2.1776, "step": 335 }, { "epoch": 0.12, - "grad_norm": 0.1533203125, - "learning_rate": 9.993229964671401e-05, - "loss": 2.2517, + "grad_norm": 0.158203125, + "learning_rate": 0.00011505922165820644, + "loss": 2.1847, "step": 340 }, { "epoch": 0.12, - "grad_norm": 0.158203125, - "learning_rate": 9.991604352502706e-05, - "loss": 2.2064, + "grad_norm": 0.1611328125, + "learning_rate": 0.000116751269035533, + "loss": 2.2162, "step": 345 }, { "epoch": 0.12, - "grad_norm": 0.1552734375, - "learning_rate": 9.989804149073081e-05, - "loss": 2.2151, + "grad_norm": 0.15625, + "learning_rate": 0.00011844331641285957, + "loss": 2.1783, "step": 350 }, { "epoch": 0.12, - "grad_norm": 0.1552734375, - "learning_rate": 9.987829417348213e-05, - "loss": 2.23, + "grad_norm": 0.1640625, + "learning_rate": 0.00012013536379018613, + "loss": 2.1957, "step": 355 }, { "epoch": 0.12, - "grad_norm": 0.15234375, - "learning_rate": 9.985680226398261e-05, - "loss": 2.2422, + "grad_norm": 0.1552734375, + "learning_rate": 0.0001218274111675127, + "loss": 2.1976, "step": 360 }, { "epoch": 0.12, - "grad_norm": 0.1591796875, - "learning_rate": 9.983356651395436e-05, - "loss": 2.2123, + "grad_norm": 0.1611328125, + "learning_rate": 0.00012351945854483927, + "loss": 2.2132, "step": 365 }, { "epoch": 0.13, - "grad_norm": 0.1669921875, - "learning_rate": 9.980858773611388e-05, - "loss": 2.2226, + "grad_norm": 0.1650390625, + "learning_rate": 0.0001252115059221658, + "loss": 2.1745, "step": 370 }, { "epoch": 0.13, - "grad_norm": 0.154296875, - "learning_rate": 9.97818668041434e-05, - "loss": 2.2272, + "grad_norm": 0.1494140625, + "learning_rate": 0.0001269035532994924, + "loss": 2.1744, "step": 375 }, { "epoch": 0.13, - "grad_norm": 0.1630859375, - "learning_rate": 9.975340465266053e-05, - "loss": 2.2006, + "grad_norm": 0.158203125, + "learning_rate": 0.00012859560067681895, + "loss": 2.1901, "step": 380 }, { "epoch": 0.13, "grad_norm": 0.158203125, - "learning_rate": 9.972320227718546e-05, - "loss": 2.2035, + "learning_rate": 0.00013028764805414554, + "loss": 2.2184, "step": 385 }, { "epoch": 0.13, - "grad_norm": 0.154296875, - "learning_rate": 9.969126073410617e-05, - "loss": 2.2141, + "grad_norm": 0.162109375, + "learning_rate": 0.00013197969543147207, + "loss": 2.1992, "step": 390 }, { "epoch": 0.13, - "grad_norm": 0.15625, - "learning_rate": 9.965758114064147e-05, - "loss": 2.204, + "grad_norm": 0.16015625, + "learning_rate": 0.00013367174280879866, + "loss": 2.1905, "step": 395 }, { "epoch": 0.14, "grad_norm": 0.1572265625, - "learning_rate": 9.962216467480193e-05, - "loss": 2.2051, + "learning_rate": 0.00013536379018612522, + "loss": 2.1837, "step": 400 }, { "epoch": 0.14, - "grad_norm": 0.1572265625, - "learning_rate": 9.958501257534866e-05, - "loss": 2.2387, + "grad_norm": 0.15625, + "learning_rate": 0.00013705583756345178, + "loss": 2.2044, "step": 405 }, { "epoch": 0.14, "grad_norm": 0.158203125, - "learning_rate": 9.954612614175003e-05, - "loss": 2.2191, + "learning_rate": 0.00013874788494077836, + "loss": 2.2061, "step": 410 }, { "epoch": 0.14, - "grad_norm": 0.158203125, - "learning_rate": 9.950550673413617e-05, - "loss": 2.2016, + "grad_norm": 0.150390625, + "learning_rate": 0.0001404399323181049, + "loss": 2.1866, "step": 415 }, { "epoch": 0.14, - "grad_norm": 0.1611328125, - "learning_rate": 9.94631557732514e-05, - "loss": 2.2261, + "grad_norm": 0.1552734375, + "learning_rate": 0.00014213197969543148, + "loss": 2.1918, "step": 420 }, { "epoch": 0.14, - "grad_norm": 0.1611328125, - "learning_rate": 9.941907474040458e-05, - "loss": 2.1897, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014382402707275804, + "loss": 2.1879, "step": 425 }, { "epoch": 0.15, - "grad_norm": 0.1611328125, - "learning_rate": 9.937326517741724e-05, - "loss": 2.2203, + "grad_norm": 0.15234375, + "learning_rate": 0.00014551607445008463, + "loss": 2.1818, "step": 430 }, { "epoch": 0.15, - "grad_norm": 0.16015625, - "learning_rate": 9.932572868656969e-05, - "loss": 2.194, + "grad_norm": 0.1611328125, + "learning_rate": 0.00014720812182741116, + "loss": 2.164, "step": 435 }, { "epoch": 0.15, - "grad_norm": 0.1611328125, - "learning_rate": 9.927646693054496e-05, - "loss": 2.2442, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014890016920473775, + "loss": 2.1927, "step": 440 }, { "epoch": 0.15, - "grad_norm": 0.1591796875, - "learning_rate": 9.922548163237066e-05, - "loss": 2.1735, + "grad_norm": 0.15625, + "learning_rate": 0.0001505922165820643, + "loss": 2.1951, "step": 445 }, { "epoch": 0.15, - "grad_norm": 0.1640625, - "learning_rate": 9.917277457535872e-05, - "loss": 2.1946, + "grad_norm": 0.166015625, + "learning_rate": 0.00015228426395939087, + "loss": 2.1963, "step": 450 }, { "epoch": 0.15, - "grad_norm": 0.1533203125, - "learning_rate": 9.911834760304294e-05, - "loss": 2.2291, + "grad_norm": 0.1484375, + "learning_rate": 0.00015397631133671742, + "loss": 2.1982, "step": 455 }, { "epoch": 0.16, - "grad_norm": 0.1630859375, - "learning_rate": 9.906220261911465e-05, - "loss": 2.1984, + "grad_norm": 0.1474609375, + "learning_rate": 0.00015566835871404398, + "loss": 2.1704, "step": 460 }, { "epoch": 0.16, - "grad_norm": 0.1591796875, - "learning_rate": 9.900434158735598e-05, - "loss": 2.2231, + "grad_norm": 0.15234375, + "learning_rate": 0.00015736040609137057, + "loss": 2.1999, "step": 465 }, { "epoch": 0.16, - "grad_norm": 0.162109375, - "learning_rate": 9.894476653157126e-05, - "loss": 2.1912, + "grad_norm": 0.15234375, + "learning_rate": 0.00015905245346869713, + "loss": 2.1524, "step": 470 }, { "epoch": 0.16, - "grad_norm": 0.1689453125, - "learning_rate": 9.88834795355162e-05, - "loss": 2.2118, + "grad_norm": 0.146484375, + "learning_rate": 0.00016074450084602372, + "loss": 2.1904, "step": 475 }, { "epoch": 0.16, - "grad_norm": 0.1630859375, - "learning_rate": 9.882048274282505e-05, - "loss": 2.2189, + "grad_norm": 0.1494140625, + "learning_rate": 0.00016243654822335025, + "loss": 2.1741, "step": 480 }, { "epoch": 0.16, - "grad_norm": 0.1689453125, - "learning_rate": 9.875577835693554e-05, - "loss": 2.2098, + "grad_norm": 0.1552734375, + "learning_rate": 0.00016412859560067684, + "loss": 2.2073, "step": 485 }, { "epoch": 0.17, - "grad_norm": 0.1572265625, - "learning_rate": 9.868936864101188e-05, - "loss": 2.2567, + "grad_norm": 0.150390625, + "learning_rate": 0.0001658206429780034, + "loss": 2.1577, "step": 490 }, { "epoch": 0.17, - "grad_norm": 0.1611328125, - "learning_rate": 9.86212559178656e-05, - "loss": 2.2096, + "grad_norm": 0.1552734375, + "learning_rate": 0.00016751269035532995, + "loss": 2.1828, "step": 495 }, { "epoch": 0.17, - "grad_norm": 0.166015625, - "learning_rate": 9.855144256987423e-05, - "loss": 2.2169, + "grad_norm": 0.15234375, + "learning_rate": 0.00016920473773265651, + "loss": 2.1915, "step": 500 }, { "epoch": 0.17, - "grad_norm": 0.1533203125, - "learning_rate": 9.84799310388981e-05, - "loss": 2.206, + "grad_norm": 0.16015625, + "learning_rate": 0.00017089678510998307, + "loss": 2.1873, "step": 505 }, { "epoch": 0.17, - "grad_norm": 0.15625, - "learning_rate": 9.840672382619478e-05, - "loss": 2.2387, + "grad_norm": 0.154296875, + "learning_rate": 0.00017258883248730966, + "loss": 2.1937, "step": 510 }, { "epoch": 0.17, - "grad_norm": 0.16015625, - "learning_rate": 9.833182349233174e-05, - "loss": 2.1945, + "grad_norm": 0.15625, + "learning_rate": 0.00017428087986463622, + "loss": 2.1888, "step": 515 }, { "epoch": 0.18, - "grad_norm": 0.1572265625, - "learning_rate": 9.825523265709666e-05, - "loss": 2.2079, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001759729272419628, + "loss": 2.2072, "step": 520 }, { "epoch": 0.18, - "grad_norm": 0.1630859375, - "learning_rate": 9.817695399940592e-05, - "loss": 2.2095, + "grad_norm": 0.1533203125, + "learning_rate": 0.00017766497461928934, + "loss": 2.2001, "step": 525 }, { "epoch": 0.18, - "grad_norm": 0.1572265625, - "learning_rate": 9.80969902572108e-05, - "loss": 2.2254, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001793570219966159, + "loss": 2.193, "step": 530 }, { "epoch": 0.18, - "grad_norm": 0.1630859375, - "learning_rate": 9.801534422740173e-05, - "loss": 2.2193, + "grad_norm": 0.154296875, + "learning_rate": 0.00018104906937394248, + "loss": 2.1653, "step": 535 }, { "epoch": 0.18, - "grad_norm": 0.1640625, - "learning_rate": 9.793201876571053e-05, - "loss": 2.2287, + "grad_norm": 0.1513671875, + "learning_rate": 0.00018274111675126904, + "loss": 2.191, "step": 540 }, { "epoch": 0.18, - "grad_norm": 0.15625, - "learning_rate": 9.784701678661045e-05, - "loss": 2.2142, + "grad_norm": 0.154296875, + "learning_rate": 0.0001844331641285956, + "loss": 2.2017, "step": 545 }, { "epoch": 0.19, - "grad_norm": 0.16015625, - "learning_rate": 9.77603412632143e-05, - "loss": 2.1981, + "grad_norm": 0.1533203125, + "learning_rate": 0.00018612521150592216, + "loss": 2.1921, "step": 550 }, { "epoch": 0.19, - "grad_norm": 0.1630859375, - "learning_rate": 9.767199522717036e-05, - "loss": 2.1763, + "grad_norm": 0.150390625, + "learning_rate": 0.00018781725888324875, + "loss": 2.1688, "step": 555 }, { "epoch": 0.19, - "grad_norm": 0.166015625, - "learning_rate": 9.758198176855648e-05, - "loss": 2.1829, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001895093062605753, + "loss": 2.1696, "step": 560 }, { "epoch": 0.19, - "grad_norm": 0.1572265625, - "learning_rate": 9.749030403577184e-05, - "loss": 2.1872, + "grad_norm": 0.1494140625, + "learning_rate": 0.00019120135363790187, + "loss": 2.1755, "step": 565 }, { "epoch": 0.19, - "grad_norm": 0.1572265625, - "learning_rate": 9.739696523542696e-05, - "loss": 2.2023, + "grad_norm": 0.1552734375, + "learning_rate": 0.00019289340101522843, + "loss": 2.2083, "step": 570 }, { "epoch": 0.19, - "grad_norm": 0.1640625, - "learning_rate": 9.73019686322315e-05, - "loss": 2.1966, + "grad_norm": 0.1533203125, + "learning_rate": 0.000194585448392555, + "loss": 2.1612, "step": 575 }, { "epoch": 0.2, - "grad_norm": 0.1650390625, - "learning_rate": 9.720531754888e-05, - "loss": 2.2187, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019627749576988157, + "loss": 2.1808, "step": 580 }, { "epoch": 0.2, - "grad_norm": 0.16015625, - "learning_rate": 9.710701536593581e-05, - "loss": 2.2016, + "grad_norm": 0.15234375, + "learning_rate": 0.00019796954314720813, + "loss": 2.1557, "step": 585 }, { "epoch": 0.2, - "grad_norm": 0.1572265625, - "learning_rate": 9.700706552171268e-05, - "loss": 2.2113, + "grad_norm": 0.14453125, + "learning_rate": 0.0001996615905245347, + "loss": 2.1605, "step": 590 }, { "epoch": 0.2, - "grad_norm": 0.162109375, - "learning_rate": 9.690547151215463e-05, - "loss": 2.1862, + "grad_norm": 0.15234375, + "learning_rate": 0.00019999972028877317, + "loss": 2.1787, "step": 595 }, { "epoch": 0.2, - "grad_norm": 0.1630859375, - "learning_rate": 9.680223689071364e-05, - "loss": 2.1791, + "grad_norm": 0.1494140625, + "learning_rate": 0.00019999858396459598, + "loss": 2.1948, "step": 600 }, { "epoch": 0.2, - "grad_norm": 0.15625, - "learning_rate": 9.669736526822528e-05, - "loss": 2.2076, + "grad_norm": 0.14453125, + "learning_rate": 0.00019999657355544167, + "loss": 2.1524, "step": 605 }, { "epoch": 0.21, - "grad_norm": 0.162109375, - "learning_rate": 9.659086031278254e-05, - "loss": 2.2267, + "grad_norm": 0.142578125, + "learning_rate": 0.00019999368907888313, + "loss": 2.1582, "step": 610 }, { "epoch": 0.21, - "grad_norm": 0.1630859375, - "learning_rate": 9.648272574960744e-05, - "loss": 2.1674, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001999899305601336, + "loss": 2.2172, "step": 615 }, { "epoch": 0.21, - "grad_norm": 0.158203125, - "learning_rate": 9.637296536092075e-05, - "loss": 2.1736, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001999852980320461, + "loss": 2.1962, "step": 620 }, { "epoch": 0.21, - "grad_norm": 0.1591796875, - "learning_rate": 9.626158298580973e-05, - "loss": 2.196, + "grad_norm": 0.142578125, + "learning_rate": 0.0001999797915351135, + "loss": 2.1669, "step": 625 }, { "epoch": 0.21, - "grad_norm": 0.1572265625, - "learning_rate": 9.614858252009385e-05, - "loss": 2.1903, + "grad_norm": 0.142578125, + "learning_rate": 0.00019997341111746791, + "loss": 2.1439, "step": 630 }, { "epoch": 0.22, - "grad_norm": 0.1572265625, - "learning_rate": 9.603396791618844e-05, - "loss": 2.1976, + "grad_norm": 0.146484375, + "learning_rate": 0.00019996615683488039, + "loss": 2.1757, "step": 635 }, { "epoch": 0.22, - "grad_norm": 0.1611328125, - "learning_rate": 9.591774318296661e-05, - "loss": 2.1985, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019995802875076042, + "loss": 2.1692, "step": 640 }, { "epoch": 0.22, - "grad_norm": 0.154296875, - "learning_rate": 9.579991238561887e-05, - "loss": 2.1934, + "grad_norm": 0.14453125, + "learning_rate": 0.0001999490269361554, + "loss": 2.1709, "step": 645 }, { "epoch": 0.22, - "grad_norm": 0.162109375, - "learning_rate": 9.568047964551102e-05, - "loss": 2.1879, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019993915146974996, + "loss": 2.1642, "step": 650 }, { "epoch": 0.22, - "grad_norm": 0.1591796875, - "learning_rate": 9.555944914003998e-05, - "loss": 2.1918, + "grad_norm": 0.142578125, + "learning_rate": 0.00019992840243786525, + "loss": 2.1699, "step": 655 }, { "epoch": 0.22, - "grad_norm": 0.1611328125, - "learning_rate": 9.54368251024877e-05, - "loss": 2.2127, + "grad_norm": 0.1494140625, + "learning_rate": 0.0001999167799344583, + "loss": 2.1677, "step": 660 }, { "epoch": 0.23, - "grad_norm": 0.15625, - "learning_rate": 9.531261182187308e-05, - "loss": 2.1711, + "grad_norm": 0.14453125, + "learning_rate": 0.0001999042840611211, + "loss": 2.195, "step": 665 }, { "epoch": 0.23, - "grad_norm": 0.1591796875, - "learning_rate": 9.51868136428019e-05, - "loss": 2.2108, + "grad_norm": 0.142578125, + "learning_rate": 0.00019989091492707975, + "loss": 2.1986, "step": 670 }, { "epoch": 0.23, "grad_norm": 0.150390625, - "learning_rate": 9.505943496531496e-05, - "loss": 2.2295, + "learning_rate": 0.0001998766726491935, + "loss": 2.1512, "step": 675 }, { "epoch": 0.23, - "grad_norm": 0.1591796875, - "learning_rate": 9.493048024473412e-05, - "loss": 2.1847, + "grad_norm": 0.142578125, + "learning_rate": 0.00019986155735195372, + "loss": 2.1755, "step": 680 }, { "epoch": 0.23, - "grad_norm": 0.162109375, - "learning_rate": 9.479995399150644e-05, - "loss": 2.1936, + "grad_norm": 0.1484375, + "learning_rate": 0.0001998455691674828, + "loss": 2.1832, "step": 685 }, { "epoch": 0.23, - "grad_norm": 0.1640625, - "learning_rate": 9.466786077104646e-05, - "loss": 2.1942, + "grad_norm": 0.1396484375, + "learning_rate": 0.00019982870823553308, + "loss": 2.1831, "step": 690 }, { "epoch": 0.24, - "grad_norm": 0.1591796875, - "learning_rate": 9.453420520357652e-05, - "loss": 2.1972, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019981097470348548, + "loss": 2.1756, "step": 695 }, { "epoch": 0.24, - "grad_norm": 0.15234375, - "learning_rate": 9.439899196396515e-05, - "loss": 2.1822, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019979236872634838, + "loss": 2.1821, "step": 700 }, { "epoch": 0.24, - "grad_norm": 0.1572265625, - "learning_rate": 9.426222578156356e-05, - "loss": 2.2155, + "grad_norm": 0.14453125, + "learning_rate": 0.0001997728904667561, + "loss": 2.193, "step": 705 }, { "epoch": 0.24, - "grad_norm": 0.15625, - "learning_rate": 9.412391144004017e-05, - "loss": 2.1891, + "grad_norm": 0.140625, + "learning_rate": 0.00019975254009496762, + "loss": 2.1561, "step": 710 }, { "epoch": 0.24, - "grad_norm": 0.158203125, - "learning_rate": 9.398405377721338e-05, - "loss": 2.1806, + "grad_norm": 0.140625, + "learning_rate": 0.00019973131778886497, + "loss": 2.1596, "step": 715 }, { "epoch": 0.24, - "grad_norm": 0.1630859375, - "learning_rate": 9.384265768488225e-05, - "loss": 2.1999, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019970922373395176, + "loss": 2.1796, "step": 720 }, { "epoch": 0.25, - "grad_norm": 0.1611328125, - "learning_rate": 9.369972810865557e-05, - "loss": 2.224, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019968625812335158, + "loss": 2.1946, "step": 725 }, { "epoch": 0.25, - "grad_norm": 0.16015625, - "learning_rate": 9.355527004777868e-05, - "loss": 2.2033, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019966242115780617, + "loss": 2.1836, "step": 730 }, { "epoch": 0.25, - "grad_norm": 0.1611328125, - "learning_rate": 9.340928855495872e-05, - "loss": 2.1892, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019963771304567387, + "loss": 2.1734, "step": 735 }, { "epoch": 0.25, - "grad_norm": 0.1630859375, - "learning_rate": 9.32617887361879e-05, - "loss": 2.2023, + "grad_norm": 0.146484375, + "learning_rate": 0.00019961213400292762, + "loss": 2.1611, "step": 740 }, { "epoch": 0.25, - "grad_norm": 0.1650390625, - "learning_rate": 9.311277575056489e-05, - "loss": 2.2089, + "grad_norm": 0.1513671875, + "learning_rate": 0.00019958568425315314, + "loss": 2.188, "step": 745 }, { "epoch": 0.25, - "grad_norm": 0.1611328125, - "learning_rate": 9.296225481011436e-05, - "loss": 2.1764, + "grad_norm": 0.146484375, + "learning_rate": 0.000199558364027547, + "loss": 2.1382, "step": 750 }, { "epoch": 0.26, - "grad_norm": 0.1611328125, - "learning_rate": 9.281023117960468e-05, - "loss": 2.2349, + "grad_norm": 0.1533203125, + "learning_rate": 0.00019953017356491457, + "loss": 2.1972, "step": 755 }, { "epoch": 0.26, - "grad_norm": 0.162109375, - "learning_rate": 9.265671017636383e-05, - "loss": 2.2008, + "grad_norm": 0.150390625, + "learning_rate": 0.0001995011131116679, + "loss": 2.1808, "step": 760 }, { "epoch": 0.26, - "grad_norm": 0.1630859375, - "learning_rate": 9.250169717009334e-05, - "loss": 2.1711, + "grad_norm": 0.1494140625, + "learning_rate": 0.00019947118292182377, + "loss": 2.1648, "step": 765 }, { "epoch": 0.26, - "grad_norm": 0.1669921875, - "learning_rate": 9.234519758268049e-05, - "loss": 2.2002, + "grad_norm": 0.146484375, + "learning_rate": 0.00019944038325700103, + "loss": 2.171, "step": 770 }, { "epoch": 0.26, - "grad_norm": 0.158203125, - "learning_rate": 9.218721688800868e-05, - "loss": 2.184, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019940871438641882, + "loss": 2.1709, "step": 775 }, { "epoch": 0.26, - "grad_norm": 0.1611328125, - "learning_rate": 9.202776061176605e-05, - "loss": 2.1849, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019937617658689384, + "loss": 2.1942, "step": 780 }, { "epoch": 0.27, - "grad_norm": 0.1689453125, - "learning_rate": 9.186683433125203e-05, - "loss": 2.1952, + "grad_norm": 0.1396484375, + "learning_rate": 0.0001993427701428382, + "loss": 2.1225, "step": 785 }, { "epoch": 0.27, - "grad_norm": 0.1611328125, - "learning_rate": 9.170444367518241e-05, - "loss": 2.2033, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001993084953462567, + "loss": 2.1786, "step": 790 }, { "epoch": 0.27, - "grad_norm": 0.166015625, - "learning_rate": 9.154059432349245e-05, - "loss": 2.191, + "grad_norm": 0.142578125, + "learning_rate": 0.00019927335249674447, + "loss": 2.1774, "step": 795 }, { "epoch": 0.27, - "grad_norm": 0.158203125, - "learning_rate": 9.13752920071381e-05, - "loss": 2.1984, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019923734190148418, + "loss": 2.1731, "step": 800 }, { "epoch": 0.27, - "grad_norm": 0.1591796875, - "learning_rate": 9.120854250789573e-05, - "loss": 2.2054, + "grad_norm": 0.14453125, + "learning_rate": 0.0001992004638752435, + "loss": 2.1626, "step": 805 }, { "epoch": 0.27, - "grad_norm": 0.1669921875, - "learning_rate": 9.104035165815971e-05, - "loss": 2.2218, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001991627187403723, + "loss": 2.1665, "step": 810 }, { "epoch": 0.28, - "grad_norm": 0.16015625, - "learning_rate": 9.087072534073859e-05, - "loss": 2.2102, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001991241068267998, + "loss": 2.1873, "step": 815 }, { "epoch": 0.28, - "grad_norm": 0.1669921875, - "learning_rate": 9.069966948864916e-05, - "loss": 2.2294, + "grad_norm": 0.146484375, + "learning_rate": 0.00019908462847203172, + "loss": 2.1638, "step": 820 }, { "epoch": 0.28, - "grad_norm": 0.1630859375, - "learning_rate": 9.052719008490909e-05, - "loss": 2.203, + "grad_norm": 0.1416015625, + "learning_rate": 0.0001990442840211473, + "loss": 2.1676, "step": 825 }, { "epoch": 0.28, - "grad_norm": 0.1611328125, - "learning_rate": 9.035329316232755e-05, - "loss": 2.1868, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019900307382679638, + "loss": 2.1379, "step": 830 }, { "epoch": 0.28, - "grad_norm": 0.1650390625, - "learning_rate": 9.017798480329427e-05, - "loss": 2.1755, + "grad_norm": 0.1376953125, + "learning_rate": 0.00019896099824919604, + "loss": 2.1727, "step": 835 }, { "epoch": 0.28, - "grad_norm": 0.1669921875, - "learning_rate": 9.000127113956674e-05, - "loss": 2.2099, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019891805765612794, + "loss": 2.1587, "step": 840 }, { "epoch": 0.29, - "grad_norm": 0.1552734375, - "learning_rate": 8.982315835205578e-05, - "loss": 2.2155, + "grad_norm": 0.14453125, + "learning_rate": 0.0001988742524229346, + "loss": 2.1581, "step": 845 }, { "epoch": 0.29, - "grad_norm": 0.16796875, - "learning_rate": 8.964365267060935e-05, - "loss": 2.2156, + "grad_norm": 0.142578125, + "learning_rate": 0.00019882958293251636, + "loss": 2.1801, "step": 850 }, { "epoch": 0.29, - "grad_norm": 0.1640625, - "learning_rate": 8.946276037379467e-05, - "loss": 2.1751, + "grad_norm": 0.138671875, + "learning_rate": 0.00019878404957532814, + "loss": 2.1448, "step": 855 }, { "epoch": 0.29, - "grad_norm": 0.1689453125, - "learning_rate": 8.928048778867848e-05, - "loss": 2.1603, + "grad_norm": 0.142578125, + "learning_rate": 0.00019873765274937578, + "loss": 2.1585, "step": 860 }, { "epoch": 0.29, - "grad_norm": 0.166015625, - "learning_rate": 8.909684129060593e-05, - "loss": 2.179, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019869039286021271, + "loss": 2.1749, "step": 865 }, { "epoch": 0.29, - "grad_norm": 0.16015625, - "learning_rate": 8.89118273029775e-05, - "loss": 2.1944, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019864227032093633, + "loss": 2.1607, "step": 870 }, { "epoch": 0.3, - "grad_norm": 0.1650390625, - "learning_rate": 8.872545229702426e-05, - "loss": 2.214, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019859328555218455, + "loss": 2.1621, "step": 875 }, { "epoch": 0.3, - "grad_norm": 0.1611328125, - "learning_rate": 8.853772279158166e-05, - "loss": 2.1941, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001985434389821319, + "loss": 2.1608, "step": 880 }, { "epoch": 0.3, - "grad_norm": 0.16796875, - "learning_rate": 8.834864535286143e-05, - "loss": 2.168, + "grad_norm": 0.142578125, + "learning_rate": 0.00019849273104648592, + "loss": 2.1601, "step": 885 }, { "epoch": 0.3, - "grad_norm": 0.1640625, - "learning_rate": 8.815822659422195e-05, - "loss": 2.206, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019844116218848334, + "loss": 2.1498, "step": 890 }, { "epoch": 0.3, - "grad_norm": 0.1611328125, - "learning_rate": 8.796647317593691e-05, - "loss": 2.2074, + "grad_norm": 0.142578125, + "learning_rate": 0.0001983887328588862, + "loss": 2.1895, "step": 895 }, { "epoch": 0.3, - "grad_norm": 0.1669921875, - "learning_rate": 8.777339180496238e-05, - "loss": 2.19, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019833544351597788, + "loss": 2.1391, "step": 900 }, { "epoch": 0.31, - "grad_norm": 0.1630859375, - "learning_rate": 8.757898923470218e-05, - "loss": 2.2011, + "grad_norm": 0.1484375, + "learning_rate": 0.0001982812946255591, + "loss": 2.1489, "step": 905 }, { "epoch": 0.31, - "grad_norm": 0.162109375, - "learning_rate": 8.738327226477176e-05, - "loss": 2.1937, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001982262866609439, + "loss": 2.1337, "step": 910 }, { "epoch": 0.31, - "grad_norm": 0.158203125, - "learning_rate": 8.718624774076023e-05, - "loss": 2.1904, + "grad_norm": 0.14453125, + "learning_rate": 0.00019817042010295544, + "loss": 2.1387, "step": 915 }, { "epoch": 0.31, - "grad_norm": 0.1630859375, - "learning_rate": 8.698792255399104e-05, - "loss": 2.1914, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019811369543992183, + "loss": 2.1662, "step": 920 }, { "epoch": 0.31, - "grad_norm": 0.1630859375, - "learning_rate": 8.67883036412809e-05, - "loss": 2.1953, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001980561131676718, + "loss": 2.1467, "step": 925 }, { - "epoch": 0.32, - "grad_norm": 0.1630859375, - "learning_rate": 8.658739798469712e-05, - "loss": 2.1928, + "epoch": 0.31, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001979976737895305, + "loss": 2.1823, "step": 930 }, { "epoch": 0.32, - "grad_norm": 0.162109375, - "learning_rate": 8.638521261131349e-05, - "loss": 2.2242, + "grad_norm": 0.14453125, + "learning_rate": 0.00019793837781631506, + "loss": 2.1764, "step": 935 }, { "epoch": 0.32, - "grad_norm": 0.166015625, - "learning_rate": 8.618175459296433e-05, - "loss": 2.1607, + "grad_norm": 0.142578125, + "learning_rate": 0.0001978782257663299, + "loss": 2.1521, "step": 940 }, { "epoch": 0.32, - "grad_norm": 0.1611328125, - "learning_rate": 8.597703104599736e-05, - "loss": 2.2007, + "grad_norm": 0.146484375, + "learning_rate": 0.00019781721816536257, + "loss": 2.1694, "step": 945 }, { "epoch": 0.32, - "grad_norm": 0.1689453125, - "learning_rate": 8.577104913102458e-05, - "loss": 2.1936, + "grad_norm": 0.14453125, + "learning_rate": 0.00019775535554667886, + "loss": 2.1781, "step": 950 }, { "epoch": 0.32, - "grad_norm": 0.1669921875, - "learning_rate": 8.556381605267196e-05, - "loss": 2.1839, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019769263845101828, + "loss": 2.1894, "step": 955 }, { "epoch": 0.33, - "grad_norm": 0.1748046875, - "learning_rate": 8.535533905932738e-05, - "loss": 2.1786, + "grad_norm": 0.1396484375, + "learning_rate": 0.00019762906742658935, + "loss": 2.1465, "step": 960 }, { "epoch": 0.33, - "grad_norm": 0.1611328125, - "learning_rate": 8.51456254428871e-05, - "loss": 2.1859, + "grad_norm": 0.1484375, + "learning_rate": 0.00019756464302906465, + "loss": 2.1602, "step": 965 }, { "epoch": 0.33, - "grad_norm": 0.162109375, - "learning_rate": 8.49346825385007e-05, - "loss": 2.2096, + "grad_norm": 0.1494140625, + "learning_rate": 0.0001974993658215762, + "loss": 2.161, "step": 970 }, { "epoch": 0.33, - "grad_norm": 0.162109375, - "learning_rate": 8.472251772431461e-05, - "loss": 2.2057, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019743323637471028, + "loss": 2.1644, "step": 975 }, { "epoch": 0.33, - "grad_norm": 0.1708984375, - "learning_rate": 8.450913842121396e-05, - "loss": 2.1645, + "grad_norm": 0.140625, + "learning_rate": 0.00019736625526650269, + "loss": 2.1671, "step": 980 }, { "epoch": 0.33, - "grad_norm": 0.171875, - "learning_rate": 8.429455209256297e-05, - "loss": 2.1972, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001972984230824335, + "loss": 2.2012, "step": 985 }, { "epoch": 0.34, - "grad_norm": 0.162109375, - "learning_rate": 8.407876624394406e-05, - "loss": 2.194, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019722974041542203, + "loss": 2.1562, "step": 990 }, { "epoch": 0.34, - "grad_norm": 0.162109375, - "learning_rate": 8.38617884228952e-05, - "loss": 2.1823, + "grad_norm": 0.1396484375, + "learning_rate": 0.00019716020786582169, + "loss": 2.1351, "step": 995 }, { "epoch": 0.34, - "grad_norm": 0.1591796875, - "learning_rate": 8.364362621864595e-05, - "loss": 2.2122, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001970898260414146, + "loss": 2.1662, "step": 1000 }, { "epoch": 0.34, - "grad_norm": 0.1650390625, - "learning_rate": 8.342428726185205e-05, - "loss": 2.2262, + "grad_norm": 0.1572265625, + "learning_rate": 0.00019701859555740648, + "loss": 2.1601, "step": 1005 }, { "epoch": 0.34, - "grad_norm": 0.1669921875, - "learning_rate": 8.320377922432848e-05, - "loss": 2.1803, + "grad_norm": 0.1513671875, + "learning_rate": 0.00019694651703642104, + "loss": 2.1325, "step": 1010 }, { "epoch": 0.34, - "grad_norm": 0.1650390625, - "learning_rate": 8.298210981878112e-05, - "loss": 2.182, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001968735911084948, + "loss": 2.1698, "step": 1015 }, { "epoch": 0.35, - "grad_norm": 0.1630859375, - "learning_rate": 8.275928679853703e-05, - "loss": 2.1982, + "grad_norm": 0.142578125, + "learning_rate": 0.0001967998184110713, + "loss": 2.1379, "step": 1020 }, { "epoch": 0.35, - "grad_norm": 0.169921875, - "learning_rate": 8.253531795727319e-05, - "loss": 2.1612, + "grad_norm": 0.1484375, + "learning_rate": 0.00019672519958899583, + "loss": 2.1447, "step": 1025 }, { "epoch": 0.35, - "grad_norm": 0.1640625, - "learning_rate": 8.231021112874402e-05, - "loss": 2.1983, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019664973529450946, + "loss": 2.1684, "step": 1030 }, { "epoch": 0.35, - "grad_norm": 0.1689453125, - "learning_rate": 8.20839741865072e-05, - "loss": 2.1974, + "grad_norm": 0.146484375, + "learning_rate": 0.00019657342618724358, + "loss": 2.1496, "step": 1035 }, { "epoch": 0.35, - "grad_norm": 0.1650390625, - "learning_rate": 8.185661504364844e-05, - "loss": 2.1986, + "grad_norm": 0.146484375, + "learning_rate": 0.00019649627293421413, + "loss": 2.1179, "step": 1040 }, { "epoch": 0.35, - "grad_norm": 0.1591796875, - "learning_rate": 8.162814165250464e-05, - "loss": 2.1974, + "grad_norm": 0.142578125, + "learning_rate": 0.00019641827620981564, + "loss": 2.1262, "step": 1045 }, { "epoch": 0.36, - "grad_norm": 0.166015625, - "learning_rate": 8.139856200438575e-05, - "loss": 2.1886, + "grad_norm": 0.15234375, + "learning_rate": 0.0001963394366958154, + "loss": 2.1477, "step": 1050 }, { "epoch": 0.36, - "grad_norm": 0.1748046875, - "learning_rate": 8.11678841292952e-05, - "loss": 2.2085, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019625975508134755, + "loss": 2.1596, "step": 1055 }, { "epoch": 0.36, - "grad_norm": 0.171875, - "learning_rate": 8.093611609564913e-05, - "loss": 2.2038, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019617923206290692, + "loss": 2.165, "step": 1060 }, { "epoch": 0.36, - "grad_norm": 0.16015625, - "learning_rate": 8.070326600999416e-05, - "loss": 2.2125, + "grad_norm": 0.150390625, + "learning_rate": 0.00019609786834434313, + "loss": 2.1671, "step": 1065 }, { "epoch": 0.36, - "grad_norm": 0.1611328125, - "learning_rate": 8.046934201672376e-05, - "loss": 2.1787, + "grad_norm": 0.146484375, + "learning_rate": 0.00019601566463685425, + "loss": 2.1638, "step": 1070 }, { "epoch": 0.36, - "grad_norm": 0.1669921875, - "learning_rate": 8.023435229779351e-05, - "loss": 2.2398, + "grad_norm": 0.14453125, + "learning_rate": 0.00019593262165898076, + "loss": 2.1579, "step": 1075 }, { "epoch": 0.37, - "grad_norm": 0.1591796875, - "learning_rate": 7.999830507243478e-05, - "loss": 2.1843, + "grad_norm": 0.1484375, + "learning_rate": 0.0001958487401365991, + "loss": 2.1628, "step": 1080 }, { "epoch": 0.37, - "grad_norm": 0.1650390625, - "learning_rate": 7.976120859686744e-05, - "loss": 2.1823, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019576402080291545, + "loss": 2.1963, "step": 1085 }, { "epoch": 0.37, - "grad_norm": 0.1689453125, - "learning_rate": 7.952307116401086e-05, - "loss": 2.1992, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019567846439845927, + "loss": 2.1792, "step": 1090 }, { "epoch": 0.37, - "grad_norm": 0.16796875, - "learning_rate": 7.9283901103194e-05, - "loss": 2.1805, + "grad_norm": 0.146484375, + "learning_rate": 0.00019559207167107684, + "loss": 2.1701, "step": 1095 }, { "epoch": 0.37, - "grad_norm": 0.16796875, - "learning_rate": 7.904370677986404e-05, - "loss": 2.2111, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019550484337592464, + "loss": 2.1511, "step": 1100 }, { "epoch": 0.37, - "grad_norm": 0.16015625, - "learning_rate": 7.880249659529376e-05, - "loss": 2.1732, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019541678027546296, + "loss": 2.1653, "step": 1105 }, { "epoch": 0.38, - "grad_norm": 0.1630859375, - "learning_rate": 7.85602789862877e-05, - "loss": 2.1761, + "grad_norm": 0.140625, + "learning_rate": 0.00019532788313944904, + "loss": 2.1728, "step": 1110 }, { "epoch": 0.38, - "grad_norm": 0.1708984375, - "learning_rate": 7.831706242488708e-05, - "loss": 2.1856, + "grad_norm": 0.1494140625, + "learning_rate": 0.00019523815274493031, + "loss": 2.1703, "step": 1115 }, { "epoch": 0.38, - "grad_norm": 0.1630859375, - "learning_rate": 7.80728554180734e-05, - "loss": 2.1798, + "grad_norm": 0.15234375, + "learning_rate": 0.00019514758987623784, + "loss": 2.1723, "step": 1120 }, { "epoch": 0.38, - "grad_norm": 0.166015625, - "learning_rate": 7.782766650747108e-05, - "loss": 2.1974, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019505619532497926, + "loss": 2.174, "step": 1125 }, { "epoch": 0.38, - "grad_norm": 0.1630859375, - "learning_rate": 7.758150426904845e-05, - "loss": 2.1742, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019496396989003193, + "loss": 2.1431, "step": 1130 }, { "epoch": 0.38, - "grad_norm": 0.162109375, - "learning_rate": 7.733437731281797e-05, - "loss": 2.1869, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019487091437753594, + "loss": 2.158, "step": 1135 }, { "epoch": 0.39, - "grad_norm": 0.162109375, - "learning_rate": 7.708629428253497e-05, - "loss": 2.1808, + "grad_norm": 0.14453125, + "learning_rate": 0.00019477702960088702, + "loss": 2.1763, "step": 1140 }, { "epoch": 0.39, - "grad_norm": 0.1640625, - "learning_rate": 7.683726385539544e-05, - "loss": 2.2019, + "grad_norm": 0.146484375, + "learning_rate": 0.0001946823163807296, + "loss": 2.1655, "step": 1145 }, { "epoch": 0.39, - "grad_norm": 0.1708984375, - "learning_rate": 7.658729474173241e-05, - "loss": 2.2051, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019458677554494932, + "loss": 2.1646, "step": 1150 }, { "epoch": 0.39, - "grad_norm": 0.16015625, - "learning_rate": 7.63363956847113e-05, - "loss": 2.1951, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001944904079286662, + "loss": 2.1502, "step": 1155 }, { "epoch": 0.39, - "grad_norm": 0.16796875, - "learning_rate": 7.608457546002424e-05, - "loss": 2.2108, + "grad_norm": 0.14453125, + "learning_rate": 0.00019439321437422695, + "loss": 2.1562, "step": 1160 }, { "epoch": 0.39, - "grad_norm": 0.1640625, - "learning_rate": 7.58318428755829e-05, - "loss": 2.1563, + "grad_norm": 0.146484375, + "learning_rate": 0.00019429519573119794, + "loss": 2.1559, "step": 1165 }, { "epoch": 0.4, - "grad_norm": 0.166015625, - "learning_rate": 7.557820677121067e-05, - "loss": 2.1778, + "grad_norm": 0.150390625, + "learning_rate": 0.00019419635285635746, + "loss": 2.1707, "step": 1170 }, { "epoch": 0.4, - "grad_norm": 0.16015625, - "learning_rate": 7.532367601833321e-05, - "loss": 2.2198, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001940966866136885, + "loss": 2.1393, "step": 1175 }, { "epoch": 0.4, - "grad_norm": 0.16796875, - "learning_rate": 7.506825951966843e-05, - "loss": 2.1818, + "grad_norm": 0.1494140625, + "learning_rate": 0.00019399619787437104, + "loss": 2.1421, "step": 1180 }, { "epoch": 0.4, - "grad_norm": 0.1708984375, - "learning_rate": 7.481196620891482e-05, - "loss": 2.1774, + "grad_norm": 0.146484375, + "learning_rate": 0.0001938948875167745, + "loss": 2.1468, "step": 1185 }, { "epoch": 0.4, - "grad_norm": 0.166015625, - "learning_rate": 7.45548050504392e-05, - "loss": 2.167, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019379275642645002, + "loss": 2.1617, "step": 1190 }, { "epoch": 0.4, - "grad_norm": 0.1728515625, - "learning_rate": 7.429678503896304e-05, - "loss": 2.153, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001936898054961228, + "loss": 2.1828, "step": 1195 }, { "epoch": 0.41, - "grad_norm": 0.171875, - "learning_rate": 7.403791519924794e-05, - "loss": 2.1921, + "grad_norm": 0.1484375, + "learning_rate": 0.00019358603562568416, + "loss": 2.1237, "step": 1200 }, { "epoch": 0.41, - "grad_norm": 0.1669921875, - "learning_rate": 7.377820458577987e-05, - "loss": 2.1914, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001934814477221838, + "loss": 2.1469, "step": 1205 }, { "epoch": 0.41, - "grad_norm": 0.166015625, - "learning_rate": 7.351766228245259e-05, - "loss": 2.1962, + "grad_norm": 0.140625, + "learning_rate": 0.0001933760426998218, + "loss": 2.1826, "step": 1210 }, { "epoch": 0.41, - "grad_norm": 0.1650390625, - "learning_rate": 7.325629740224979e-05, - "loss": 2.216, + "grad_norm": 0.14453125, + "learning_rate": 0.0001932698214799407, + "loss": 2.1484, "step": 1215 }, { "epoch": 0.41, - "grad_norm": 0.162109375, - "learning_rate": 7.299411908692649e-05, - "loss": 2.1387, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001931627849910174, + "loss": 2.1428, "step": 1220 }, { "epoch": 0.41, - "grad_norm": 0.1669921875, - "learning_rate": 7.273113650668919e-05, - "loss": 2.1843, + "grad_norm": 0.1435546875, + "learning_rate": 0.00019305493416865493, + "loss": 2.1281, "step": 1225 }, { "epoch": 0.42, - "grad_norm": 0.1669921875, - "learning_rate": 7.246735885987515e-05, - "loss": 2.1779, + "grad_norm": 0.14453125, + "learning_rate": 0.00019294626995557457, + "loss": 2.161, "step": 1230 }, { "epoch": 0.42, - "grad_norm": 0.1669921875, - "learning_rate": 7.220279537263063e-05, - "loss": 2.2041, + "grad_norm": 0.1494140625, + "learning_rate": 0.00019283679330160726, + "loss": 2.1455, "step": 1235 }, { "epoch": 0.42, - "grad_norm": 0.171875, - "learning_rate": 7.193745529858826e-05, - "loss": 2.1866, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001927265051636856, + "loss": 2.1554, "step": 1240 }, { "epoch": 0.42, - "grad_norm": 0.169921875, - "learning_rate": 7.167134791854333e-05, - "loss": 2.1654, + "grad_norm": 0.14453125, + "learning_rate": 0.00019261540650583522, + "loss": 2.1855, "step": 1245 }, { "epoch": 0.42, - "grad_norm": 0.1689453125, - "learning_rate": 7.140448254012912e-05, - "loss": 2.1932, + "grad_norm": 0.142578125, + "learning_rate": 0.00019250349829916661, + "loss": 2.1403, "step": 1250 }, { "epoch": 0.43, - "grad_norm": 0.166015625, - "learning_rate": 7.113686849749153e-05, - "loss": 2.1708, + "grad_norm": 0.146484375, + "learning_rate": 0.0001923907815218664, + "loss": 2.1797, "step": 1255 }, { "epoch": 0.43, - "grad_norm": 0.169921875, - "learning_rate": 7.086851515096233e-05, - "loss": 2.1748, + "grad_norm": 0.1416015625, + "learning_rate": 0.00019227725715918897, + "loss": 2.1369, "step": 1260 }, { "epoch": 0.43, - "grad_norm": 0.1689453125, - "learning_rate": 7.0599431886732e-05, - "loss": 2.1981, + "grad_norm": 0.14453125, + "learning_rate": 0.00019216292620344777, + "loss": 2.1737, "step": 1265 }, { "epoch": 0.43, - "grad_norm": 0.1650390625, - "learning_rate": 7.032962811652133e-05, - "loss": 2.1723, + "grad_norm": 0.1533203125, + "learning_rate": 0.00019204778965400667, + "loss": 2.1471, "step": 1270 }, { "epoch": 0.43, - "grad_norm": 0.16796875, - "learning_rate": 7.005911327725222e-05, - "loss": 2.1849, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001919318485172712, + "loss": 2.1757, "step": 1275 }, { "epoch": 0.43, - "grad_norm": 0.1728515625, - "learning_rate": 6.97878968307176e-05, - "loss": 2.1794, + "grad_norm": 0.15234375, + "learning_rate": 0.00019181510380667977, + "loss": 2.1649, "step": 1280 }, { "epoch": 0.44, - "grad_norm": 0.1669921875, - "learning_rate": 6.951598826325056e-05, - "loss": 2.1861, + "grad_norm": 0.14453125, + "learning_rate": 0.0001916975565426948, + "loss": 2.1699, "step": 1285 }, { "epoch": 0.44, - "grad_norm": 0.166015625, - "learning_rate": 6.924339708539244e-05, - "loss": 2.1937, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019157920775279383, + "loss": 2.1468, "step": 1290 }, { "epoch": 0.44, - "grad_norm": 0.1591796875, - "learning_rate": 6.897013283156026e-05, - "loss": 2.1785, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001914600584714605, + "loss": 2.1455, "step": 1295 }, { "epoch": 0.44, - "grad_norm": 0.1689453125, - "learning_rate": 6.869620505971321e-05, - "loss": 2.1728, + "grad_norm": 0.154296875, + "learning_rate": 0.0001913401097401755, + "loss": 2.1362, "step": 1300 }, { "epoch": 0.44, - "grad_norm": 0.166015625, - "learning_rate": 6.842162335101829e-05, - "loss": 2.2097, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019121936260740752, + "loss": 2.176, "step": 1305 }, { "epoch": 0.44, - "grad_norm": 0.1748046875, - "learning_rate": 6.814639730951532e-05, - "loss": 2.166, + "grad_norm": 0.1435546875, + "learning_rate": 0.0001910978181286041, + "loss": 2.1609, "step": 1310 }, { "epoch": 0.45, - "grad_norm": 0.1669921875, - "learning_rate": 6.787053656178087e-05, - "loss": 2.1783, + "grad_norm": 0.1474609375, + "learning_rate": 0.00019097547736618228, + "loss": 2.1566, "step": 1315 }, { "epoch": 0.45, - "grad_norm": 0.16796875, - "learning_rate": 6.759405075659166e-05, - "loss": 2.2017, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001908523413895194, + "loss": 2.1392, "step": 1320 }, { "epoch": 0.45, - "grad_norm": 0.1650390625, - "learning_rate": 6.731694956458701e-05, - "loss": 2.1601, + "grad_norm": 0.14453125, + "learning_rate": 0.0001907284112749438, + "loss": 2.1392, "step": 1325 }, { "epoch": 0.45, - "grad_norm": 0.173828125, - "learning_rate": 6.703924267793061e-05, - "loss": 2.1962, + "grad_norm": 0.1455078125, + "learning_rate": 0.00019060368810572539, + "loss": 2.1563, "step": 1330 }, { "epoch": 0.45, - "grad_norm": 0.162109375, - "learning_rate": 6.676093980997155e-05, - "loss": 2.1764, + "grad_norm": 0.150390625, + "learning_rate": 0.00019047817297206598, + "loss": 2.1203, "step": 1335 }, { "epoch": 0.45, - "grad_norm": 0.1689453125, - "learning_rate": 6.648205069490451e-05, - "loss": 2.2019, + "grad_norm": 0.1513671875, + "learning_rate": 0.00019035186697109011, + "loss": 2.1533, "step": 1340 }, { "epoch": 0.46, - "grad_norm": 0.1708984375, - "learning_rate": 6.620258508742935e-05, - "loss": 2.205, + "grad_norm": 0.14453125, + "learning_rate": 0.0001902247712068352, + "loss": 2.1503, "step": 1345 }, { "epoch": 0.46, - "grad_norm": 0.16796875, - "learning_rate": 6.592255276240994e-05, - "loss": 2.162, + "grad_norm": 0.142578125, + "learning_rate": 0.0001900968867902419, + "loss": 2.1798, "step": 1350 }, { "epoch": 0.46, - "grad_norm": 0.16796875, - "learning_rate": 6.564196351453209e-05, - "loss": 2.1992, + "grad_norm": 0.150390625, + "learning_rate": 0.0001899682148391446, + "loss": 2.1541, "step": 1355 }, { "epoch": 0.46, - "grad_norm": 0.1669921875, - "learning_rate": 6.536082715796125e-05, - "loss": 2.1835, + "grad_norm": 0.1513671875, + "learning_rate": 0.00018983875647826136, + "loss": 2.1286, "step": 1360 }, { "epoch": 0.46, - "grad_norm": 0.16796875, - "learning_rate": 6.507915352599895e-05, - "loss": 2.1851, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018970851283918428, + "loss": 2.1471, "step": 1365 }, { "epoch": 0.46, - "grad_norm": 0.1650390625, - "learning_rate": 6.479695247073907e-05, - "loss": 2.2148, + "grad_norm": 0.1435546875, + "learning_rate": 0.00018957748506036957, + "loss": 2.137, "step": 1370 }, { "epoch": 0.47, - "grad_norm": 0.1630859375, - "learning_rate": 6.451423386272312e-05, - "loss": 2.1823, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018944567428712765, + "loss": 2.1514, "step": 1375 }, { "epoch": 0.47, - "grad_norm": 0.1689453125, - "learning_rate": 6.423100759059509e-05, - "loss": 2.1676, + "grad_norm": 0.1484375, + "learning_rate": 0.0001893130816716129, + "loss": 2.1667, "step": 1380 }, { "epoch": 0.47, - "grad_norm": 0.1669921875, - "learning_rate": 6.394728356075551e-05, - "loss": 2.1841, + "grad_norm": 0.1455078125, + "learning_rate": 0.00018917970837281392, + "loss": 2.1724, "step": 1385 }, { "epoch": 0.47, - "grad_norm": 0.1728515625, - "learning_rate": 6.366307169701496e-05, - "loss": 2.1638, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018904555555654317, + "loss": 2.1847, "step": 1390 }, { "epoch": 0.47, - "grad_norm": 0.162109375, - "learning_rate": 6.337838194024697e-05, - "loss": 2.1984, + "grad_norm": 0.150390625, + "learning_rate": 0.0001889106243954269, + "loss": 2.157, "step": 1395 }, { "epoch": 0.47, - "grad_norm": 0.1650390625, - "learning_rate": 6.309322424804034e-05, - "loss": 2.1869, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018877491606889476, + "loss": 2.1312, "step": 1400 }, { "epoch": 0.48, - "grad_norm": 0.166015625, - "learning_rate": 6.280760859435087e-05, - "loss": 2.1684, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001886384317631697, + "loss": 2.1694, "step": 1405 }, { "epoch": 0.48, - "grad_norm": 0.1689453125, - "learning_rate": 6.252154496915244e-05, - "loss": 2.1879, + "grad_norm": 0.146484375, + "learning_rate": 0.00018850117267125738, + "loss": 2.1772, "step": 1410 }, { "epoch": 0.48, - "grad_norm": 0.1708984375, - "learning_rate": 6.223504337808761e-05, - "loss": 2.1625, + "grad_norm": 0.14453125, + "learning_rate": 0.00018836313999293593, + "loss": 2.1599, "step": 1415 }, { "epoch": 0.48, - "grad_norm": 0.1640625, - "learning_rate": 6.194811384211768e-05, - "loss": 2.1731, + "grad_norm": 0.1435546875, + "learning_rate": 0.00018822433493474532, + "loss": 2.1282, "step": 1420 }, { "epoch": 0.48, - "grad_norm": 0.1630859375, - "learning_rate": 6.166076639717218e-05, - "loss": 2.1673, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001880847587099769, + "loss": 2.1316, "step": 1425 }, { "epoch": 0.48, - "grad_norm": 0.1708984375, - "learning_rate": 6.137301109379783e-05, - "loss": 2.1944, + "grad_norm": 0.146484375, + "learning_rate": 0.00018794441253866274, + "loss": 2.17, "step": 1430 }, { "epoch": 0.49, - "grad_norm": 0.169921875, - "learning_rate": 6.108485799680701e-05, - "loss": 2.1989, + "grad_norm": 0.150390625, + "learning_rate": 0.00018780329764756505, + "loss": 2.1501, "step": 1435 }, { "epoch": 0.49, - "grad_norm": 0.166015625, - "learning_rate": 6.079631718492569e-05, - "loss": 2.163, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018766141527016533, + "loss": 2.1335, "step": 1440 }, { "epoch": 0.49, - "grad_norm": 0.1708984375, - "learning_rate": 6.050739875044098e-05, - "loss": 2.1824, + "grad_norm": 0.14453125, + "learning_rate": 0.00018751876664665367, + "loss": 2.1353, "step": 1445 }, { "epoch": 0.49, - "grad_norm": 0.1650390625, - "learning_rate": 6.021811279884807e-05, - "loss": 2.193, + "grad_norm": 0.1513671875, + "learning_rate": 0.00018737535302391795, + "loss": 2.1778, "step": 1450 }, { "epoch": 0.49, - "grad_norm": 0.169921875, - "learning_rate": 5.992846944849679e-05, - "loss": 2.196, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018723117565553284, + "loss": 2.1515, "step": 1455 }, { "epoch": 0.49, - "grad_norm": 0.16796875, - "learning_rate": 5.96384788302377e-05, - "loss": 2.1575, + "grad_norm": 0.150390625, + "learning_rate": 0.00018708623580174889, + "loss": 2.1433, "step": 1460 }, { "epoch": 0.5, - "grad_norm": 0.17578125, - "learning_rate": 5.934815108706775e-05, - "loss": 2.2159, + "grad_norm": 0.1484375, + "learning_rate": 0.00018694053472948156, + "loss": 2.1386, "step": 1465 }, { "epoch": 0.5, - "grad_norm": 0.169921875, - "learning_rate": 5.905749637377549e-05, - "loss": 2.1838, + "grad_norm": 0.1484375, + "learning_rate": 0.00018679407371230002, + "loss": 2.1572, "step": 1470 }, { "epoch": 0.5, - "grad_norm": 0.1689453125, - "learning_rate": 5.87665248565859e-05, - "loss": 2.2138, + "grad_norm": 0.1484375, + "learning_rate": 0.00018664685403041619, + "loss": 2.1624, "step": 1475 }, { "epoch": 0.5, - "grad_norm": 0.1708984375, - "learning_rate": 5.847524671280484e-05, - "loss": 2.1744, + "grad_norm": 0.142578125, + "learning_rate": 0.0001864988769706734, + "loss": 2.162, "step": 1480 }, { "epoch": 0.5, - "grad_norm": 0.1630859375, - "learning_rate": 5.818367213046298e-05, - "loss": 2.1914, + "grad_norm": 0.146484375, + "learning_rate": 0.0001863501438265352, + "loss": 2.1513, "step": 1485 }, { "epoch": 0.5, - "grad_norm": 0.1669921875, - "learning_rate": 5.7891811307959574e-05, - "loss": 2.1712, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018620065589807413, + "loss": 2.165, "step": 1490 }, { "epoch": 0.51, - "grad_norm": 0.16796875, - "learning_rate": 5.759967445370564e-05, - "loss": 2.161, + "grad_norm": 0.1484375, + "learning_rate": 0.00018605041449196012, + "loss": 2.18, "step": 1495 }, { "epoch": 0.51, - "grad_norm": 0.1689453125, - "learning_rate": 5.7307271785767034e-05, - "loss": 2.1811, + "grad_norm": 0.14453125, + "learning_rate": 0.00018589942092144942, + "loss": 2.1805, "step": 1500 }, { "epoch": 0.51, - "grad_norm": 0.1728515625, - "learning_rate": 5.701461353150687e-05, - "loss": 2.1932, + "grad_norm": 0.1552734375, + "learning_rate": 0.00018574767650637278, + "loss": 2.114, "step": 1505 }, { "epoch": 0.51, - "grad_norm": 0.1650390625, - "learning_rate": 5.6721709927227974e-05, - "loss": 2.1521, + "grad_norm": 0.1484375, + "learning_rate": 0.0001855951825731241, + "loss": 2.1331, "step": 1510 }, { "epoch": 0.51, - "grad_norm": 0.1640625, - "learning_rate": 5.642857121781475e-05, - "loss": 2.1954, + "grad_norm": 0.1435546875, + "learning_rate": 0.00018544194045464886, + "loss": 2.1368, "step": 1515 }, { "epoch": 0.51, - "grad_norm": 0.1708984375, - "learning_rate": 5.613520765637489e-05, - "loss": 2.169, + "grad_norm": 0.146484375, + "learning_rate": 0.00018528795149043236, + "loss": 2.1579, "step": 1520 }, { "epoch": 0.52, - "grad_norm": 0.16796875, - "learning_rate": 5.58416295038807e-05, - "loss": 2.2079, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018513321702648807, + "loss": 2.1433, "step": 1525 }, { "epoch": 0.52, - "grad_norm": 0.16796875, - "learning_rate": 5.554784702881025e-05, - "loss": 2.1636, + "grad_norm": 0.1484375, + "learning_rate": 0.0001849777384153458, + "loss": 2.1542, "step": 1530 }, { "epoch": 0.52, - "grad_norm": 0.1669921875, - "learning_rate": 5.525387050678819e-05, - "loss": 2.1689, + "grad_norm": 0.1455078125, + "learning_rate": 0.00018482151701604003, + "loss": 2.163, "step": 1535 }, { "epoch": 0.52, - "grad_norm": 0.1728515625, - "learning_rate": 5.495971022022638e-05, - "loss": 2.179, + "grad_norm": 0.1455078125, + "learning_rate": 0.00018466455419409786, + "loss": 2.1541, "step": 1540 }, { "epoch": 0.52, - "grad_norm": 0.166015625, - "learning_rate": 5.466537645796416e-05, - "loss": 2.195, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001845068513215271, + "loss": 2.143, "step": 1545 }, { - "epoch": 0.53, - "grad_norm": 0.169921875, - "learning_rate": 5.437087951490856e-05, - "loss": 2.1827, + "epoch": 0.52, + "grad_norm": 0.146484375, + "learning_rate": 0.00018434840977680453, + "loss": 2.1377, "step": 1550 }, { "epoch": 0.53, - "grad_norm": 0.1787109375, - "learning_rate": 5.4076229691674164e-05, - "loss": 2.1747, + "grad_norm": 0.146484375, + "learning_rate": 0.00018418923094486338, + "loss": 2.1539, "step": 1555 }, { "epoch": 0.53, - "grad_norm": 0.1689453125, - "learning_rate": 5.3781437294222845e-05, - "loss": 2.179, + "grad_norm": 0.146484375, + "learning_rate": 0.00018402931621708165, + "loss": 2.161, "step": 1560 }, { "epoch": 0.53, - "grad_norm": 0.169921875, - "learning_rate": 5.3486512633503303e-05, - "loss": 2.219, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018386866699126973, + "loss": 2.1789, "step": 1565 }, { "epoch": 0.53, - "grad_norm": 0.1767578125, - "learning_rate": 5.319146602509042e-05, - "loss": 2.198, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018370728467165828, + "loss": 2.1419, "step": 1570 }, { "epoch": 0.53, - "grad_norm": 0.1728515625, - "learning_rate": 5.289630778882442e-05, - "loss": 2.1706, + "grad_norm": 0.146484375, + "learning_rate": 0.0001835451706688859, + "loss": 2.1748, "step": 1575 }, { "epoch": 0.54, - "grad_norm": 0.171875, - "learning_rate": 5.260104824844989e-05, - "loss": 2.1725, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001833823263999867, + "loss": 2.127, "step": 1580 }, { "epoch": 0.54, - "grad_norm": 0.171875, - "learning_rate": 5.230569773125484e-05, - "loss": 2.1811, + "grad_norm": 0.150390625, + "learning_rate": 0.00018321875328837828, + "loss": 2.132, "step": 1585 }, { "epoch": 0.54, - "grad_norm": 0.1728515625, - "learning_rate": 5.201026656770926e-05, - "loss": 2.1745, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018305445276384875, + "loss": 2.1438, "step": 1590 }, { "epoch": 0.54, - "grad_norm": 0.171875, - "learning_rate": 5.1714765091104003e-05, - "loss": 2.1982, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018288942626254473, + "loss": 2.1572, "step": 1595 }, { "epoch": 0.54, - "grad_norm": 0.1640625, - "learning_rate": 5.141920363718916e-05, - "loss": 2.1957, + "grad_norm": 0.150390625, + "learning_rate": 0.00018272367522695844, + "loss": 2.1464, "step": 1600 }, { "epoch": 0.54, - "grad_norm": 0.173828125, - "learning_rate": 5.1123592543812734e-05, - "loss": 2.1758, + "grad_norm": 0.1455078125, + "learning_rate": 0.00018255720110591533, + "loss": 2.157, "step": 1605 }, { "epoch": 0.55, - "grad_norm": 0.16796875, - "learning_rate": 5.082794215055894e-05, - "loss": 2.1576, + "grad_norm": 0.150390625, + "learning_rate": 0.0001823900053545613, + "loss": 2.1392, "step": 1610 }, { "epoch": 0.55, - "grad_norm": 0.16796875, - "learning_rate": 5.0532262798386544e-05, - "loss": 2.1923, + "grad_norm": 0.154296875, + "learning_rate": 0.00018222208943434999, + "loss": 2.1497, "step": 1615 }, { "epoch": 0.55, - "grad_norm": 0.173828125, - "learning_rate": 5.023656482926727e-05, - "loss": 2.1893, + "grad_norm": 0.150390625, + "learning_rate": 0.00018205345481302998, + "loss": 2.1474, "step": 1620 }, { "epoch": 0.55, - "grad_norm": 0.166015625, - "learning_rate": 4.994085858582397e-05, - "loss": 2.163, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001818841029646321, + "loss": 2.153, "step": 1625 }, { "epoch": 0.55, - "grad_norm": 0.169921875, - "learning_rate": 4.964515441096889e-05, - "loss": 2.1753, + "grad_norm": 0.1513671875, + "learning_rate": 0.00018171403536945628, + "loss": 2.1696, "step": 1630 }, { "epoch": 0.55, - "grad_norm": 0.16796875, - "learning_rate": 4.934946264754199e-05, - "loss": 2.1831, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018154325351405897, + "loss": 2.1849, "step": 1635 }, { "epoch": 0.56, - "grad_norm": 0.16796875, - "learning_rate": 4.9053793637949067e-05, - "loss": 2.1525, + "grad_norm": 0.1513671875, + "learning_rate": 0.00018137175889123978, + "loss": 2.1445, "step": 1640 }, { "epoch": 0.56, - "grad_norm": 0.1728515625, - "learning_rate": 4.875815772380002e-05, - "loss": 2.1687, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001811995530000287, + "loss": 2.1402, "step": 1645 }, { "epoch": 0.56, - "grad_norm": 0.1640625, - "learning_rate": 4.846256524554725e-05, - "loss": 2.1775, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018102663734567283, + "loss": 2.1413, "step": 1650 }, { "epoch": 0.56, - "grad_norm": 0.1689453125, - "learning_rate": 4.8167026542123874e-05, - "loss": 2.1846, + "grad_norm": 0.1494140625, + "learning_rate": 0.0001808530134396234, + "loss": 2.1554, "step": 1655 }, { "epoch": 0.56, - "grad_norm": 0.16796875, - "learning_rate": 4.78715519505821e-05, - "loss": 2.1507, + "grad_norm": 0.1474609375, + "learning_rate": 0.00018067868279952236, + "loss": 2.1374, "step": 1660 }, { "epoch": 0.56, - "grad_norm": 0.173828125, - "learning_rate": 4.7576151805731695e-05, - "loss": 2.2013, + "grad_norm": 0.1494140625, + "learning_rate": 0.0001805036469491892, + "loss": 2.167, "step": 1665 }, { "epoch": 0.57, - "grad_norm": 0.1669921875, - "learning_rate": 4.728083643977855e-05, - "loss": 2.1392, + "grad_norm": 0.1494140625, + "learning_rate": 0.00018032790741860763, + "loss": 2.1369, "step": 1670 }, { "epoch": 0.57, - "grad_norm": 0.1708984375, - "learning_rate": 4.698561618196323e-05, - "loss": 2.184, + "grad_norm": 0.150390625, + "learning_rate": 0.00018015146574391233, + "loss": 2.1283, "step": 1675 }, { "epoch": 0.57, - "grad_norm": 0.162109375, - "learning_rate": 4.669050135819966e-05, - "loss": 2.1987, + "grad_norm": 0.1484375, + "learning_rate": 0.00017997432346737524, + "loss": 2.1505, "step": 1680 }, { "epoch": 0.57, - "grad_norm": 0.16796875, - "learning_rate": 4.639550229071407e-05, - "loss": 2.1599, + "grad_norm": 0.1474609375, + "learning_rate": 0.00017979648213739232, + "loss": 2.1338, "step": 1685 }, { "epoch": 0.57, - "grad_norm": 0.171875, - "learning_rate": 4.610062929768383e-05, - "loss": 2.1722, + "grad_norm": 0.1474609375, + "learning_rate": 0.00017961794330846994, + "loss": 2.1555, "step": 1690 }, { "epoch": 0.57, - "grad_norm": 0.1689453125, - "learning_rate": 4.580589269287661e-05, - "loss": 2.1756, + "grad_norm": 0.150390625, + "learning_rate": 0.00017943870854121124, + "loss": 2.1543, "step": 1695 }, { "epoch": 0.58, - "grad_norm": 0.1689453125, - "learning_rate": 4.5511302785289685e-05, - "loss": 2.1929, + "grad_norm": 0.15234375, + "learning_rate": 0.0001792587794023027, + "loss": 2.1429, "step": 1700 }, { "epoch": 0.58, - "grad_norm": 0.171875, - "learning_rate": 4.521686987878925e-05, - "loss": 2.1731, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017907815746450004, + "loss": 2.1174, "step": 1705 }, { "epoch": 0.58, - "grad_norm": 0.16796875, - "learning_rate": 4.492260427175007e-05, - "loss": 2.182, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001788968443066149, + "loss": 2.1252, "step": 1710 }, { "epoch": 0.58, - "grad_norm": 0.1728515625, - "learning_rate": 4.4628516256695305e-05, - "loss": 2.1768, + "grad_norm": 0.1474609375, + "learning_rate": 0.00017871484151350077, + "loss": 2.1074, "step": 1715 }, { "epoch": 0.58, - "grad_norm": 0.171875, - "learning_rate": 4.433461611993651e-05, - "loss": 2.162, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017853215067603926, + "loss": 2.1919, "step": 1720 }, { "epoch": 0.58, - "grad_norm": 0.1728515625, - "learning_rate": 4.4040914141213774e-05, - "loss": 2.1815, + "grad_norm": 0.1474609375, + "learning_rate": 0.00017834877339112612, + "loss": 2.126, "step": 1725 }, { "epoch": 0.59, - "grad_norm": 0.177734375, - "learning_rate": 4.374742059333621e-05, - "loss": 2.1603, + "grad_norm": 0.150390625, + "learning_rate": 0.0001781647112616574, + "loss": 2.1564, "step": 1730 }, { "epoch": 0.59, - "grad_norm": 0.1650390625, - "learning_rate": 4.345414574182272e-05, - "loss": 2.1844, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001779799658965153, + "loss": 2.137, "step": 1735 }, { "epoch": 0.59, - "grad_norm": 0.1708984375, - "learning_rate": 4.316109984454278e-05, - "loss": 2.1923, + "grad_norm": 0.146484375, + "learning_rate": 0.00017779453891055412, + "loss": 2.1714, "step": 1740 }, { "epoch": 0.59, - "grad_norm": 0.1669921875, - "learning_rate": 4.2868293151357806e-05, - "loss": 2.2025, + "grad_norm": 0.1533203125, + "learning_rate": 0.00017760843192458626, + "loss": 2.1456, "step": 1745 }, { "epoch": 0.59, - "grad_norm": 0.1767578125, - "learning_rate": 4.2575735903762513e-05, - "loss": 2.1585, + "grad_norm": 0.150390625, + "learning_rate": 0.00017742164656536798, + "loss": 2.1576, "step": 1750 }, { "epoch": 0.59, - "grad_norm": 0.1669921875, - "learning_rate": 4.228343833452684e-05, - "loss": 2.1745, + "grad_norm": 0.1552734375, + "learning_rate": 0.00017723418446558516, + "loss": 2.1461, "step": 1755 }, { "epoch": 0.6, - "grad_norm": 0.16796875, - "learning_rate": 4.1991410667337896e-05, - "loss": 2.2175, + "grad_norm": 0.1484375, + "learning_rate": 0.00017704604726383904, + "loss": 2.1383, "step": 1760 }, { "epoch": 0.6, - "grad_norm": 0.169921875, - "learning_rate": 4.1699663116442434e-05, - "loss": 2.1835, + "grad_norm": 0.1474609375, + "learning_rate": 0.00017685723660463193, + "loss": 2.13, "step": 1765 }, { "epoch": 0.6, - "grad_norm": 0.1650390625, - "learning_rate": 4.140820588628964e-05, - "loss": 2.1859, + "grad_norm": 0.1484375, + "learning_rate": 0.00017666775413835282, + "loss": 2.1463, "step": 1770 }, { "epoch": 0.6, - "grad_norm": 0.173828125, - "learning_rate": 4.1117049171174104e-05, - "loss": 2.1912, + "grad_norm": 0.150390625, + "learning_rate": 0.0001764776015212629, + "loss": 2.1076, "step": 1775 }, { "epoch": 0.6, - "grad_norm": 0.173828125, - "learning_rate": 4.082620315487931e-05, - "loss": 2.1961, + "grad_norm": 0.1484375, + "learning_rate": 0.0001762867804154812, + "loss": 2.1439, "step": 1780 }, { "epoch": 0.6, - "grad_norm": 0.1728515625, - "learning_rate": 4.053567801032144e-05, - "loss": 2.1623, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017609529248896997, + "loss": 2.1802, "step": 1785 }, { "epoch": 0.61, - "grad_norm": 0.1728515625, - "learning_rate": 4.0245483899193595e-05, - "loss": 2.1723, + "grad_norm": 0.1513671875, + "learning_rate": 0.00017590313941552002, + "loss": 2.1747, "step": 1790 }, { "epoch": 0.61, - "grad_norm": 0.16796875, - "learning_rate": 3.995563097161026e-05, - "loss": 2.1777, + "grad_norm": 0.1513671875, + "learning_rate": 0.00017571032287473642, + "loss": 2.1198, "step": 1795 }, { "epoch": 0.61, - "grad_norm": 0.169921875, - "learning_rate": 3.966612936575235e-05, - "loss": 2.1937, + "grad_norm": 0.15234375, + "learning_rate": 0.00017551684455202336, + "loss": 2.1279, "step": 1800 }, { "epoch": 0.61, - "grad_norm": 0.171875, - "learning_rate": 3.937698920751268e-05, - "loss": 2.184, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017532270613856976, + "loss": 2.1531, "step": 1805 }, { "epoch": 0.61, - "grad_norm": 0.1748046875, - "learning_rate": 3.9088220610141655e-05, - "loss": 2.181, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017512790933133437, + "loss": 2.1273, "step": 1810 }, { "epoch": 0.61, - "grad_norm": 0.166015625, - "learning_rate": 3.87998336738936e-05, - "loss": 2.1893, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001749324558330309, + "loss": 2.1526, "step": 1815 }, { "epoch": 0.62, - "grad_norm": 0.17578125, - "learning_rate": 3.851183848567351e-05, - "loss": 2.175, + "grad_norm": 0.146484375, + "learning_rate": 0.0001747363473521132, + "loss": 2.1484, "step": 1820 }, { "epoch": 0.62, - "grad_norm": 0.16796875, - "learning_rate": 3.822424511868421e-05, - "loss": 2.1897, + "grad_norm": 0.15234375, + "learning_rate": 0.00017453958560276038, + "loss": 2.1475, "step": 1825 }, { "epoch": 0.62, - "grad_norm": 0.171875, - "learning_rate": 3.7937063632074036e-05, - "loss": 2.16, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017434217230486164, + "loss": 2.1606, "step": 1830 }, { "epoch": 0.62, - "grad_norm": 0.169921875, - "learning_rate": 3.7650304070584955e-05, - "loss": 2.1899, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001741441091840014, + "loss": 2.1659, "step": 1835 }, { "epoch": 0.62, - "grad_norm": 0.1708984375, - "learning_rate": 3.736397646420135e-05, - "loss": 2.1879, + "grad_norm": 0.1533203125, + "learning_rate": 0.00017394539797144413, + "loss": 2.1168, "step": 1840 }, { "epoch": 0.62, - "grad_norm": 0.171875, - "learning_rate": 3.7078090827799e-05, - "loss": 2.1832, + "grad_norm": 0.150390625, + "learning_rate": 0.00017374604040411935, + "loss": 2.1317, "step": 1845 }, { "epoch": 0.63, - "grad_norm": 0.1708984375, - "learning_rate": 3.679265716079501e-05, - "loss": 2.1894, + "grad_norm": 0.150390625, + "learning_rate": 0.00017354603822460621, + "loss": 2.1322, "step": 1850 }, { "epoch": 0.63, - "grad_norm": 0.166015625, - "learning_rate": 3.650768544679788e-05, - "loss": 2.1831, + "grad_norm": 0.154296875, + "learning_rate": 0.00017334539318111856, + "loss": 2.1417, "step": 1855 }, { "epoch": 0.63, - "grad_norm": 0.1708984375, - "learning_rate": 3.622318565325847e-05, - "loss": 2.1538, + "grad_norm": 0.150390625, + "learning_rate": 0.00017314410702748932, + "loss": 2.1343, "step": 1860 }, { "epoch": 0.63, - "grad_norm": 0.173828125, - "learning_rate": 3.593916773112122e-05, - "loss": 2.1717, + "grad_norm": 0.1533203125, + "learning_rate": 0.00017294218152315546, + "loss": 2.1471, "step": 1865 }, { "epoch": 0.63, - "grad_norm": 0.1767578125, - "learning_rate": 3.565564161447617e-05, - "loss": 2.187, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017273961843314252, + "loss": 2.147, "step": 1870 }, { "epoch": 0.64, - "grad_norm": 0.173828125, - "learning_rate": 3.5372617220211525e-05, - "loss": 2.1609, + "grad_norm": 0.1552734375, + "learning_rate": 0.0001725364195280491, + "loss": 2.1529, "step": 1875 }, { "epoch": 0.64, - "grad_norm": 0.1767578125, - "learning_rate": 3.509010444766674e-05, - "loss": 2.1721, + "grad_norm": 0.1533203125, + "learning_rate": 0.00017233258658403138, + "loss": 2.1462, "step": 1880 }, { "epoch": 0.64, - "grad_norm": 0.1650390625, - "learning_rate": 3.480811317828625e-05, - "loss": 2.162, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001721281213827878, + "loss": 2.1401, "step": 1885 }, { "epoch": 0.64, - "grad_norm": 0.1708984375, - "learning_rate": 3.452665327527391e-05, - "loss": 2.202, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017192302571154331, + "loss": 2.1614, "step": 1890 }, { "epoch": 0.64, - "grad_norm": 0.16796875, - "learning_rate": 3.4245734583248e-05, - "loss": 2.1657, + "grad_norm": 0.15234375, + "learning_rate": 0.00017171730136303364, + "loss": 2.1488, "step": 1895 }, { "epoch": 0.64, - "grad_norm": 0.1669921875, - "learning_rate": 3.3965366927896864e-05, - "loss": 2.1612, + "grad_norm": 0.1474609375, + "learning_rate": 0.00017151095013548994, + "loss": 2.1445, "step": 1900 }, { "epoch": 0.65, - "grad_norm": 0.1708984375, - "learning_rate": 3.3685560115635195e-05, - "loss": 2.1608, + "grad_norm": 0.1484375, + "learning_rate": 0.00017130397383262284, + "loss": 2.125, "step": 1905 }, { "epoch": 0.65, - "grad_norm": 0.1669921875, - "learning_rate": 3.340632393326118e-05, - "loss": 2.1767, + "grad_norm": 0.1455078125, + "learning_rate": 0.0001710963742636067, + "loss": 2.1351, "step": 1910 }, { "epoch": 0.65, - "grad_norm": 0.1748046875, - "learning_rate": 3.3127668147614e-05, - "loss": 2.1851, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017088815324306392, + "loss": 2.1285, "step": 1915 }, { "epoch": 0.65, - "grad_norm": 0.1728515625, - "learning_rate": 3.284960250523237e-05, - "loss": 2.1778, + "grad_norm": 0.154296875, + "learning_rate": 0.00017067931259104885, + "loss": 2.1599, "step": 1920 }, { "epoch": 0.65, - "grad_norm": 0.1689453125, - "learning_rate": 3.2572136732013555e-05, - "loss": 2.1723, + "grad_norm": 0.1494140625, + "learning_rate": 0.00017046985413303215, + "loss": 2.122, "step": 1925 }, { "epoch": 0.65, - "grad_norm": 0.1728515625, - "learning_rate": 3.2295280532873226e-05, - "loss": 2.1986, + "grad_norm": 0.1484375, + "learning_rate": 0.00017025977969988465, + "loss": 2.1373, "step": 1930 }, { "epoch": 0.66, - "grad_norm": 0.1669921875, - "learning_rate": 3.2019043591405936e-05, - "loss": 2.1746, + "grad_norm": 0.1513671875, + "learning_rate": 0.00017004909112786144, + "loss": 2.1298, "step": 1935 }, { "epoch": 0.66, - "grad_norm": 0.1748046875, - "learning_rate": 3.174343556954652e-05, - "loss": 2.1711, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001698377902585857, + "loss": 2.1379, "step": 1940 }, { "epoch": 0.66, - "grad_norm": 0.169921875, - "learning_rate": 3.146846610723212e-05, - "loss": 2.1924, + "grad_norm": 0.1494140625, + "learning_rate": 0.00016962587893903276, + "loss": 2.1501, "step": 1945 }, { "epoch": 0.66, - "grad_norm": 0.171875, - "learning_rate": 3.1194144822064944e-05, - "loss": 2.2043, + "grad_norm": 0.150390625, + "learning_rate": 0.0001694133590215139, + "loss": 2.1611, "step": 1950 }, { "epoch": 0.66, - "grad_norm": 0.1630859375, - "learning_rate": 3.0920481308975926e-05, - "loss": 2.1798, + "grad_norm": 0.1474609375, + "learning_rate": 0.00016920023236366002, + "loss": 2.1135, "step": 1955 }, { "epoch": 0.66, - "grad_norm": 0.173828125, - "learning_rate": 3.0647485139889145e-05, - "loss": 2.1784, + "grad_norm": 0.146484375, + "learning_rate": 0.00016898650082840572, + "loss": 2.1337, "step": 1960 }, { "epoch": 0.67, - "grad_norm": 0.1826171875, - "learning_rate": 3.037516586338699e-05, - "loss": 2.2003, + "grad_norm": 0.1474609375, + "learning_rate": 0.00016877216628397257, + "loss": 2.1258, "step": 1965 }, { "epoch": 0.67, - "grad_norm": 0.1689453125, - "learning_rate": 3.0103533004376183e-05, - "loss": 2.2178, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001685572306038532, + "loss": 2.1085, "step": 1970 }, { "epoch": 0.67, - "grad_norm": 0.173828125, - "learning_rate": 2.9832596063754613e-05, - "loss": 2.1773, + "grad_norm": 0.1572265625, + "learning_rate": 0.0001683416956667947, + "loss": 2.1549, "step": 1975 }, { "epoch": 0.67, - "grad_norm": 0.166015625, - "learning_rate": 2.9562364518079105e-05, - "loss": 2.2017, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001681255633567823, + "loss": 2.1572, "step": 1980 }, { "epoch": 0.67, - "grad_norm": 0.1748046875, - "learning_rate": 2.929284781923382e-05, - "loss": 2.1396, + "grad_norm": 0.150390625, + "learning_rate": 0.00016790883556302272, + "loss": 2.152, "step": 1985 }, { "epoch": 0.67, - "grad_norm": 0.171875, - "learning_rate": 2.902405539409978e-05, - "loss": 2.1828, + "grad_norm": 0.1572265625, + "learning_rate": 0.00016769151417992791, + "loss": 2.1427, "step": 1990 }, { "epoch": 0.68, - "grad_norm": 0.1748046875, - "learning_rate": 2.8755996644225097e-05, - "loss": 2.1618, + "grad_norm": 0.1513671875, + "learning_rate": 0.00016747360110709838, + "loss": 2.1405, "step": 1995 }, { "epoch": 0.68, - "grad_norm": 0.1728515625, - "learning_rate": 2.848868094549615e-05, - "loss": 2.1648, + "grad_norm": 0.154296875, + "learning_rate": 0.00016725509824930645, + "loss": 2.1496, "step": 2000 }, { "epoch": 0.68, - "grad_norm": 0.1708984375, - "learning_rate": 2.8222117647809553e-05, - "loss": 2.1717, + "grad_norm": 0.154296875, + "learning_rate": 0.0001670360075164799, + "loss": 2.1588, "step": 2005 }, { "epoch": 0.68, - "grad_norm": 0.16796875, - "learning_rate": 2.7956316074745293e-05, - "loss": 2.1584, + "grad_norm": 0.150390625, + "learning_rate": 0.00016681633082368498, + "loss": 2.1569, "step": 2010 }, { "epoch": 0.68, - "grad_norm": 0.16796875, - "learning_rate": 2.7691285523240474e-05, - "loss": 2.1671, + "grad_norm": 0.1513671875, + "learning_rate": 0.00016659607009110984, + "loss": 2.1429, "step": 2015 }, { "epoch": 0.68, - "grad_norm": 0.1728515625, - "learning_rate": 2.7427035263264222e-05, - "loss": 2.1939, + "grad_norm": 0.15234375, + "learning_rate": 0.00016637522724404774, + "loss": 2.1548, "step": 2020 }, { "epoch": 0.69, - "grad_norm": 0.171875, - "learning_rate": 2.7163574537493407e-05, - "loss": 2.158, + "grad_norm": 0.1494140625, + "learning_rate": 0.00016615380421288018, + "loss": 2.1186, "step": 2025 }, { "epoch": 0.69, - "grad_norm": 0.1748046875, - "learning_rate": 2.690091256098936e-05, - "loss": 2.1482, + "grad_norm": 0.1474609375, + "learning_rate": 0.00016593180293306001, + "loss": 2.1406, "step": 2030 }, { "epoch": 0.69, - "grad_norm": 0.1767578125, - "learning_rate": 2.6639058520875615e-05, - "loss": 2.1577, + "grad_norm": 0.15234375, + "learning_rate": 0.0001657092253450945, + "loss": 2.1324, "step": 2035 }, { "epoch": 0.69, - "grad_norm": 0.1689453125, - "learning_rate": 2.6378021576016466e-05, - "loss": 2.1594, + "grad_norm": 0.1484375, + "learning_rate": 0.00016548607339452853, + "loss": 2.1107, "step": 2040 }, { "epoch": 0.69, - "grad_norm": 0.169921875, - "learning_rate": 2.6117810856696702e-05, - "loss": 2.194, + "grad_norm": 0.150390625, + "learning_rate": 0.00016526234903192733, + "loss": 2.1533, "step": 2045 }, { "epoch": 0.69, - "grad_norm": 0.181640625, - "learning_rate": 2.5858435464302315e-05, - "loss": 2.1481, + "grad_norm": 0.1484375, + "learning_rate": 0.00016503805421285968, + "loss": 2.1516, "step": 2050 }, { "epoch": 0.7, - "grad_norm": 0.1748046875, - "learning_rate": 2.559990447100195e-05, - "loss": 2.1835, + "grad_norm": 0.15234375, + "learning_rate": 0.00016481319089788063, + "loss": 2.1283, "step": 2055 }, { "epoch": 0.7, - "grad_norm": 0.169921875, - "learning_rate": 2.5342226919429806e-05, - "loss": 2.1606, + "grad_norm": 0.1494140625, + "learning_rate": 0.00016458776105251447, + "loss": 2.1586, "step": 2060 }, { "epoch": 0.7, - "grad_norm": 0.1689453125, - "learning_rate": 2.5085411822369244e-05, - "loss": 2.1763, + "grad_norm": 0.1494140625, + "learning_rate": 0.0001643617666472376, + "loss": 2.1291, "step": 2065 }, { "epoch": 0.7, - "grad_norm": 0.1708984375, - "learning_rate": 2.4829468162437554e-05, - "loss": 2.1724, + "grad_norm": 0.1513671875, + "learning_rate": 0.00016413520965746097, + "loss": 2.1661, "step": 2070 }, { "epoch": 0.7, - "grad_norm": 0.1689453125, - "learning_rate": 2.4574404891771826e-05, - "loss": 2.1815, + "grad_norm": 0.1484375, + "learning_rate": 0.0001639080920635134, + "loss": 2.1415, "step": 2075 }, { "epoch": 0.7, - "grad_norm": 0.166015625, - "learning_rate": 2.43202309317157e-05, - "loss": 2.1717, + "grad_norm": 0.1591796875, + "learning_rate": 0.0001636804158506237, + "loss": 2.1533, "step": 2080 }, { "epoch": 0.71, - "grad_norm": 0.1728515625, - "learning_rate": 2.406695517250753e-05, - "loss": 2.1995, + "grad_norm": 0.15234375, + "learning_rate": 0.00016345218300890357, + "loss": 2.1246, "step": 2085 }, { "epoch": 0.71, - "grad_norm": 0.173828125, - "learning_rate": 2.381458647296925e-05, - "loss": 2.1933, + "grad_norm": 0.1533203125, + "learning_rate": 0.00016322339553333034, + "loss": 2.1427, "step": 2090 }, { "epoch": 0.71, - "grad_norm": 0.17578125, - "learning_rate": 2.3563133660196556e-05, - "loss": 2.1821, + "grad_norm": 0.1484375, + "learning_rate": 0.00016299405542372924, + "loss": 2.1526, "step": 2095 }, { "epoch": 0.71, - "grad_norm": 0.1689453125, - "learning_rate": 2.3312605529250276e-05, - "loss": 2.1763, + "grad_norm": 0.15625, + "learning_rate": 0.00016276416468475607, + "loss": 2.1265, "step": 2100 }, { "epoch": 0.71, - "grad_norm": 0.169921875, - "learning_rate": 2.3063010842848564e-05, - "loss": 2.1838, + "grad_norm": 0.1484375, + "learning_rate": 0.00016253372532587976, + "loss": 2.1505, "step": 2105 }, { "epoch": 0.71, - "grad_norm": 0.16796875, - "learning_rate": 2.2814358331060532e-05, - "loss": 2.1895, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001623027393613646, + "loss": 2.1563, "step": 2110 }, { "epoch": 0.72, - "grad_norm": 0.1708984375, - "learning_rate": 2.2566656691000932e-05, - "loss": 2.1926, + "grad_norm": 0.1552734375, + "learning_rate": 0.00016207120881025282, + "loss": 2.1499, "step": 2115 }, { "epoch": 0.72, - "grad_norm": 0.171875, - "learning_rate": 2.2319914586525777e-05, - "loss": 2.19, + "grad_norm": 0.1474609375, + "learning_rate": 0.0001618391356963468, + "loss": 2.1095, "step": 2120 }, { "epoch": 0.72, - "grad_norm": 0.1669921875, - "learning_rate": 2.2074140647929503e-05, - "loss": 2.1611, + "grad_norm": 0.150390625, + "learning_rate": 0.00016160652204819158, + "loss": 2.144, "step": 2125 }, { "epoch": 0.72, - "grad_norm": 0.171875, - "learning_rate": 2.1829343471642994e-05, - "loss": 2.1936, + "grad_norm": 0.154296875, + "learning_rate": 0.0001613733698990568, + "loss": 2.1447, "step": 2130 }, { "epoch": 0.72, - "grad_norm": 0.169921875, - "learning_rate": 2.158553161993294e-05, - "loss": 2.1815, + "grad_norm": 0.150390625, + "learning_rate": 0.00016113968128691933, + "loss": 2.1404, "step": 2135 }, { "epoch": 0.72, - "grad_norm": 0.16796875, - "learning_rate": 2.134271362060238e-05, - "loss": 2.1871, + "grad_norm": 0.15234375, + "learning_rate": 0.00016090545825444506, + "loss": 2.162, "step": 2140 }, { "epoch": 0.73, - "grad_norm": 0.1728515625, - "learning_rate": 2.1100897966692297e-05, - "loss": 2.1642, + "grad_norm": 0.1494140625, + "learning_rate": 0.00016067070284897137, + "loss": 2.1605, "step": 2145 }, { "epoch": 0.73, - "grad_norm": 0.1728515625, - "learning_rate": 2.0860093116184797e-05, - "loss": 2.1861, + "grad_norm": 0.1552734375, + "learning_rate": 0.000160435417122489, + "loss": 2.1262, "step": 2150 }, { "epoch": 0.73, - "grad_norm": 0.1689453125, - "learning_rate": 2.0620307491707012e-05, - "loss": 2.1904, + "grad_norm": 0.1513671875, + "learning_rate": 0.00016019960313162434, + "loss": 2.1194, "step": 2155 }, { "epoch": 0.73, - "grad_norm": 0.1796875, - "learning_rate": 2.0381549480236685e-05, - "loss": 2.1855, + "grad_norm": 0.15234375, + "learning_rate": 0.0001599632629376212, + "loss": 2.1531, "step": 2160 }, { "epoch": 0.73, - "grad_norm": 0.1650390625, - "learning_rate": 2.0143827432808743e-05, - "loss": 2.1886, + "grad_norm": 0.1484375, + "learning_rate": 0.00015972639860632292, + "loss": 2.1829, "step": 2165 }, { - "epoch": 0.74, - "grad_norm": 0.1669921875, - "learning_rate": 1.990714966422321e-05, - "loss": 2.1681, + "epoch": 0.73, + "grad_norm": 0.1552734375, + "learning_rate": 0.00015948901220815445, + "loss": 2.164, "step": 2170 }, { "epoch": 0.74, - "grad_norm": 0.1767578125, - "learning_rate": 1.9671524452754393e-05, - "loss": 2.1596, + "grad_norm": 0.1533203125, + "learning_rate": 0.00015925110581810394, + "loss": 2.1432, "step": 2175 }, { "epoch": 0.74, - "grad_norm": 0.1708984375, - "learning_rate": 1.9436960039861324e-05, - "loss": 2.1709, + "grad_norm": 0.1533203125, + "learning_rate": 0.00015901268151570491, + "loss": 2.1862, "step": 2180 }, { "epoch": 0.74, - "grad_norm": 0.17578125, - "learning_rate": 1.9203464629899502e-05, - "loss": 2.1794, + "grad_norm": 0.1484375, + "learning_rate": 0.0001587737413850178, + "loss": 2.1313, "step": 2185 }, { "epoch": 0.74, - "grad_norm": 0.1748046875, - "learning_rate": 1.8971046389833952e-05, - "loss": 2.1867, + "grad_norm": 0.15625, + "learning_rate": 0.00015853428751461202, + "loss": 2.1317, "step": 2190 }, { "epoch": 0.74, - "grad_norm": 0.173828125, - "learning_rate": 1.873971344895347e-05, - "loss": 2.1681, + "grad_norm": 0.15234375, + "learning_rate": 0.00015829432199754756, + "loss": 2.1345, "step": 2195 }, { "epoch": 0.75, - "grad_norm": 0.173828125, - "learning_rate": 1.850947389858643e-05, - "loss": 2.1587, + "grad_norm": 0.15234375, + "learning_rate": 0.0001580538469313566, + "loss": 2.1354, "step": 2200 }, { "epoch": 0.75, - "grad_norm": 0.1689453125, - "learning_rate": 1.8280335791817733e-05, - "loss": 2.1862, + "grad_norm": 0.1484375, + "learning_rate": 0.00015781286441802534, + "loss": 2.135, "step": 2205 }, { "epoch": 0.75, - "grad_norm": 0.1767578125, - "learning_rate": 1.805230714320701e-05, - "loss": 2.1756, + "grad_norm": 0.1552734375, + "learning_rate": 0.00015757137656397557, + "loss": 2.1041, "step": 2210 }, { "epoch": 0.75, - "grad_norm": 0.169921875, - "learning_rate": 1.7825395928508447e-05, - "loss": 2.1792, + "grad_norm": 0.15234375, + "learning_rate": 0.0001573293854800462, + "loss": 2.1074, "step": 2215 }, { "epoch": 0.75, - "grad_norm": 0.171875, - "learning_rate": 1.7599610084391784e-05, - "loss": 2.1885, + "grad_norm": 0.1533203125, + "learning_rate": 0.00015708689328147493, + "loss": 2.1311, "step": 2220 }, { "epoch": 0.75, - "grad_norm": 0.17578125, - "learning_rate": 1.737495750816464e-05, - "loss": 2.1602, + "grad_norm": 0.1513671875, + "learning_rate": 0.00015684390208787962, + "loss": 2.143, "step": 2225 }, { "epoch": 0.76, - "grad_norm": 0.173828125, - "learning_rate": 1.7151446057496406e-05, - "loss": 2.1823, + "grad_norm": 0.15234375, + "learning_rate": 0.0001566004140232399, + "loss": 2.1528, "step": 2230 }, { "epoch": 0.76, - "grad_norm": 0.1748046875, - "learning_rate": 1.6929083550143255e-05, - "loss": 2.1806, + "grad_norm": 0.1474609375, + "learning_rate": 0.00015635643121587848, + "loss": 2.1641, "step": 2235 }, { "epoch": 0.76, - "grad_norm": 0.17578125, - "learning_rate": 1.670787776367489e-05, - "loss": 2.1781, + "grad_norm": 0.1552734375, + "learning_rate": 0.00015611195579844265, + "loss": 2.1366, "step": 2240 }, { "epoch": 0.76, - "grad_norm": 0.173828125, - "learning_rate": 1.6487836435202357e-05, - "loss": 2.1545, + "grad_norm": 0.15625, + "learning_rate": 0.00015586698990788554, + "loss": 2.1539, "step": 2245 }, { "epoch": 0.76, - "grad_norm": 0.17578125, - "learning_rate": 1.6268967261107426e-05, - "loss": 2.1559, + "grad_norm": 0.1484375, + "learning_rate": 0.00015562153568544752, + "loss": 2.1306, "step": 2250 }, { "epoch": 0.76, - "grad_norm": 0.1689453125, - "learning_rate": 1.6051277896773565e-05, - "loss": 2.1705, + "grad_norm": 0.15234375, + "learning_rate": 0.00015537559527663744, + "loss": 2.1832, "step": 2255 }, { "epoch": 0.77, - "grad_norm": 0.177734375, - "learning_rate": 1.583477595631794e-05, - "loss": 2.1662, + "grad_norm": 0.15625, + "learning_rate": 0.00015512917083121397, + "loss": 2.1523, "step": 2260 }, { "epoch": 0.77, - "grad_norm": 0.16796875, - "learning_rate": 1.5619469012325255e-05, - "loss": 2.171, + "grad_norm": 0.15234375, + "learning_rate": 0.00015488226450316664, + "loss": 2.1115, "step": 2265 }, { "epoch": 0.77, - "grad_norm": 0.177734375, - "learning_rate": 1.540536459558286e-05, - "loss": 2.1957, + "grad_norm": 0.1533203125, + "learning_rate": 0.00015463487845069707, + "loss": 2.1527, "step": 2270 }, { "epoch": 0.77, - "grad_norm": 0.169921875, - "learning_rate": 1.519247019481731e-05, - "loss": 2.1696, + "grad_norm": 0.1572265625, + "learning_rate": 0.00015438701483620027, + "loss": 2.1486, "step": 2275 }, { "epoch": 0.77, - "grad_norm": 0.16796875, - "learning_rate": 1.4980793256432474e-05, - "loss": 2.1624, + "grad_norm": 0.15234375, + "learning_rate": 0.00015413867582624553, + "loss": 2.1297, "step": 2280 }, { "epoch": 0.77, - "grad_norm": 0.173828125, - "learning_rate": 1.4770341184248997e-05, - "loss": 2.162, + "grad_norm": 0.15234375, + "learning_rate": 0.00015388986359155758, + "loss": 2.1407, "step": 2285 }, { "epoch": 0.78, - "grad_norm": 0.169921875, - "learning_rate": 1.4561121339245487e-05, - "loss": 2.2188, + "grad_norm": 0.15625, + "learning_rate": 0.0001536405803069975, + "loss": 2.1033, "step": 2290 }, { "epoch": 0.78, - "grad_norm": 0.1689453125, - "learning_rate": 1.4353141039300921e-05, - "loss": 2.1537, + "grad_norm": 0.1513671875, + "learning_rate": 0.00015339082815154394, + "loss": 2.1388, "step": 2295 }, { "epoch": 0.78, - "grad_norm": 0.1728515625, - "learning_rate": 1.4146407558938695e-05, - "loss": 2.1682, + "grad_norm": 0.1572265625, + "learning_rate": 0.00015314060930827393, + "loss": 2.1755, "step": 2300 }, { "epoch": 0.78, - "grad_norm": 0.1708984375, - "learning_rate": 1.3940928129072279e-05, - "loss": 2.1791, + "grad_norm": 0.1552734375, + "learning_rate": 0.0001528899259643437, + "loss": 2.1607, "step": 2305 }, { "epoch": 0.78, - "grad_norm": 0.171875, - "learning_rate": 1.3736709936752196e-05, - "loss": 2.2001, + "grad_norm": 0.1513671875, + "learning_rate": 0.00015263878031096975, + "loss": 2.1443, "step": 2310 }, { "epoch": 0.78, - "grad_norm": 0.1728515625, - "learning_rate": 1.3533760124914713e-05, - "loss": 2.1718, + "grad_norm": 0.15625, + "learning_rate": 0.00015238717454340957, + "loss": 2.1384, "step": 2315 }, { "epoch": 0.79, - "grad_norm": 0.1728515625, - "learning_rate": 1.3332085792131966e-05, - "loss": 2.1829, + "grad_norm": 0.1533203125, + "learning_rate": 0.00015213511086094254, + "loss": 2.1205, "step": 2320 }, { "epoch": 0.79, - "grad_norm": 0.171875, - "learning_rate": 1.3131693992363664e-05, - "loss": 2.17, + "grad_norm": 0.15234375, + "learning_rate": 0.00015188259146685064, + "loss": 2.1265, "step": 2325 }, { "epoch": 0.79, - "grad_norm": 0.171875, - "learning_rate": 1.293259173471041e-05, - "loss": 2.1583, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001516296185683992, + "loss": 2.1564, "step": 2330 }, { "epoch": 0.79, - "grad_norm": 0.1689453125, - "learning_rate": 1.2734785983168485e-05, - "loss": 2.1606, + "grad_norm": 0.15625, + "learning_rate": 0.00015137619437681767, + "loss": 2.1287, "step": 2335 }, { "epoch": 0.79, - "grad_norm": 0.1748046875, - "learning_rate": 1.2538283656386319e-05, - "loss": 2.1793, + "grad_norm": 0.162109375, + "learning_rate": 0.00015112232110728015, + "loss": 2.1516, "step": 2340 }, { "epoch": 0.79, - "grad_norm": 0.171875, - "learning_rate": 1.2343091627422487e-05, - "loss": 2.1768, + "grad_norm": 0.15234375, + "learning_rate": 0.00015086800097888624, + "loss": 2.1425, "step": 2345 }, { "epoch": 0.8, - "grad_norm": 0.1669921875, - "learning_rate": 1.2149216723505246e-05, - "loss": 2.1973, + "grad_norm": 0.15625, + "learning_rate": 0.00015061323621464134, + "loss": 2.1345, "step": 2350 }, { "epoch": 0.8, - "grad_norm": 0.1708984375, - "learning_rate": 1.1956665725793831e-05, - "loss": 2.1827, + "grad_norm": 0.154296875, + "learning_rate": 0.00015035802904143762, + "loss": 2.1628, "step": 2355 }, { "epoch": 0.8, - "grad_norm": 0.171875, - "learning_rate": 1.1765445369141276e-05, - "loss": 2.174, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001501023816900342, + "loss": 2.1679, "step": 2360 }, { "epoch": 0.8, - "grad_norm": 0.1728515625, - "learning_rate": 1.1575562341858709e-05, - "loss": 2.1979, + "grad_norm": 0.15625, + "learning_rate": 0.00014984629639503785, + "loss": 2.1477, "step": 2365 }, { "epoch": 0.8, - "grad_norm": 0.18359375, - "learning_rate": 1.1387023285481575e-05, - "loss": 2.1689, + "grad_norm": 0.150390625, + "learning_rate": 0.0001495897753948833, + "loss": 2.1486, "step": 2370 }, { "epoch": 0.8, - "grad_norm": 0.177734375, - "learning_rate": 1.1199834794537263e-05, - "loss": 2.2038, + "grad_norm": 0.1533203125, + "learning_rate": 0.00014933282093181383, + "loss": 2.1509, "step": 2375 }, { "epoch": 0.81, - "grad_norm": 0.1689453125, - "learning_rate": 1.1014003416314439e-05, - "loss": 2.1569, + "grad_norm": 0.15625, + "learning_rate": 0.00014907543525186166, + "loss": 2.1244, "step": 2380 }, { "epoch": 0.81, - "grad_norm": 0.1767578125, - "learning_rate": 1.0829535650634104e-05, - "loss": 2.1832, + "grad_norm": 0.1533203125, + "learning_rate": 0.00014881762060482814, + "loss": 2.123, "step": 2385 }, { "epoch": 0.81, - "grad_norm": 0.1748046875, - "learning_rate": 1.0646437949622118e-05, - "loss": 2.2037, + "grad_norm": 0.150390625, + "learning_rate": 0.00014855937924426434, + "loss": 2.1334, "step": 2390 }, { "epoch": 0.81, - "grad_norm": 0.1748046875, - "learning_rate": 1.0464716717483736e-05, - "loss": 2.1605, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014830071342745112, + "loss": 2.1473, "step": 2395 }, { "epoch": 0.81, - "grad_norm": 0.1728515625, - "learning_rate": 1.0284378310279369e-05, - "loss": 2.1733, + "grad_norm": 0.15234375, + "learning_rate": 0.00014804162541537955, + "loss": 2.1544, "step": 2400 }, { "epoch": 0.81, - "grad_norm": 0.171875, - "learning_rate": 1.0105429035702441e-05, - "loss": 2.1641, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014778211747273114, + "loss": 2.1675, "step": 2405 }, { "epoch": 0.82, - "grad_norm": 0.1748046875, - "learning_rate": 9.927875152858729e-06, - "loss": 2.1786, + "grad_norm": 0.146484375, + "learning_rate": 0.00014752219186785784, + "loss": 2.1268, "step": 2410 }, { "epoch": 0.82, - "grad_norm": 0.1728515625, - "learning_rate": 9.751722872047353e-06, - "loss": 2.1697, + "grad_norm": 0.154296875, + "learning_rate": 0.0001472618508727626, + "loss": 2.1661, "step": 2415 }, { "epoch": 0.82, - "grad_norm": 0.169921875, - "learning_rate": 9.57697835454367e-06, - "loss": 2.1725, + "grad_norm": 0.154296875, + "learning_rate": 0.00014700109676307914, + "loss": 2.1348, "step": 2420 }, { "epoch": 0.82, - "grad_norm": 0.16796875, - "learning_rate": 9.403647712383712e-06, - "loss": 2.1869, + "grad_norm": 0.15625, + "learning_rate": 0.0001467399318180522, + "loss": 2.1568, "step": 2425 }, { "epoch": 0.82, - "grad_norm": 0.173828125, - "learning_rate": 9.231737008150415e-06, - "loss": 2.1992, + "grad_norm": 0.154296875, + "learning_rate": 0.0001464783583205177, + "loss": 2.1437, "step": 2430 }, { "epoch": 0.82, - "grad_norm": 0.171875, - "learning_rate": 9.061252254761576e-06, - "loss": 2.1953, + "grad_norm": 0.1513671875, + "learning_rate": 0.0001462163785568826, + "loss": 2.1418, "step": 2435 }, { "epoch": 0.83, - "grad_norm": 0.177734375, - "learning_rate": 8.8921994152595e-06, - "loss": 2.196, + "grad_norm": 0.1552734375, + "learning_rate": 0.00014595399481710515, + "loss": 2.1295, "step": 2440 }, { "epoch": 0.83, - "grad_norm": 0.1787109375, - "learning_rate": 8.724584402602521e-06, - "loss": 2.1629, + "grad_norm": 0.1572265625, + "learning_rate": 0.00014569120939467465, + "loss": 2.1444, "step": 2445 }, { "epoch": 0.83, - "grad_norm": 0.1748046875, - "learning_rate": 8.558413079458106e-06, - "loss": 2.159, + "grad_norm": 0.15625, + "learning_rate": 0.00014542802458659152, + "loss": 2.1483, "step": 2450 }, { "epoch": 0.83, - "grad_norm": 0.1767578125, - "learning_rate": 8.393691257997782e-06, - "loss": 2.1791, + "grad_norm": 0.1552734375, + "learning_rate": 0.0001451644426933472, + "loss": 2.1281, "step": 2455 }, { "epoch": 0.83, - "grad_norm": 0.1728515625, - "learning_rate": 8.230424699693923e-06, - "loss": 2.1723, + "grad_norm": 0.1552734375, + "learning_rate": 0.00014490046601890405, + "loss": 2.1445, "step": 2460 }, { "epoch": 0.83, - "grad_norm": 0.171875, - "learning_rate": 8.068619115118176e-06, - "loss": 2.1735, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014463609687067526, + "loss": 2.1529, "step": 2465 }, { "epoch": 0.84, - "grad_norm": 0.173828125, - "learning_rate": 7.908280163741732e-06, - "loss": 2.1856, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014437133755950448, + "loss": 2.154, "step": 2470 }, { "epoch": 0.84, - "grad_norm": 0.1728515625, - "learning_rate": 7.749413453737375e-06, - "loss": 2.1971, + "grad_norm": 0.150390625, + "learning_rate": 0.00014410619039964586, + "loss": 2.1333, "step": 2475 }, { "epoch": 0.84, - "grad_norm": 0.169921875, - "learning_rate": 7.592024541783343e-06, - "loss": 2.1577, + "grad_norm": 0.1552734375, + "learning_rate": 0.00014384065770874373, + "loss": 2.1382, "step": 2480 }, { "epoch": 0.84, - "grad_norm": 0.173828125, - "learning_rate": 7.43611893286893e-06, - "loss": 2.1727, + "grad_norm": 0.15625, + "learning_rate": 0.00014357474180781232, + "loss": 2.1293, "step": 2485 }, { "epoch": 0.84, - "grad_norm": 0.1708984375, - "learning_rate": 7.281702080102004e-06, - "loss": 2.1791, + "grad_norm": 0.15625, + "learning_rate": 0.00014330844502121547, + "loss": 2.1312, "step": 2490 }, { "epoch": 0.85, - "grad_norm": 0.1708984375, - "learning_rate": 7.128779384518164e-06, - "loss": 2.139, + "grad_norm": 0.1533203125, + "learning_rate": 0.00014304176967664637, + "loss": 2.1036, "step": 2495 }, { "epoch": 0.85, - "grad_norm": 0.1708984375, - "learning_rate": 6.977356194891998e-06, - "loss": 2.1698, + "grad_norm": 0.154296875, + "learning_rate": 0.0001427747181051071, + "loss": 2.127, "step": 2500 }, { "epoch": 0.85, - "grad_norm": 0.173828125, - "learning_rate": 6.827437807549814e-06, - "loss": 2.1472, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014250729264088843, + "loss": 2.1564, "step": 2505 }, { "epoch": 0.85, - "grad_norm": 0.1767578125, - "learning_rate": 6.679029466184506e-06, - "loss": 2.198, + "grad_norm": 0.154296875, + "learning_rate": 0.00014223949562154929, + "loss": 2.1459, "step": 2510 }, { "epoch": 0.85, - "grad_norm": 0.1728515625, - "learning_rate": 6.5321363616721306e-06, - "loss": 2.1632, + "grad_norm": 0.1533203125, + "learning_rate": 0.00014197132938789629, + "loss": 2.1222, "step": 2515 }, { "epoch": 0.85, - "grad_norm": 0.17578125, - "learning_rate": 6.386763631890313e-06, - "loss": 2.1649, + "grad_norm": 0.1591796875, + "learning_rate": 0.0001417027962839634, + "loss": 2.1501, "step": 2520 }, { "epoch": 0.86, - "grad_norm": 0.1728515625, - "learning_rate": 6.242916361538559e-06, - "loss": 2.1872, + "grad_norm": 0.1513671875, + "learning_rate": 0.00014143389865699132, + "loss": 2.1567, "step": 2525 }, { "epoch": 0.86, - "grad_norm": 0.16796875, - "learning_rate": 6.100599581960415e-06, - "loss": 2.1755, + "grad_norm": 0.154296875, + "learning_rate": 0.00014116463885740723, + "loss": 2.1289, "step": 2530 }, { "epoch": 0.86, - "grad_norm": 0.1708984375, - "learning_rate": 5.9598182709674655e-06, - "loss": 2.1657, + "grad_norm": 0.154296875, + "learning_rate": 0.00014089501923880384, + "loss": 2.1546, "step": 2535 }, { "epoch": 0.86, - "grad_norm": 0.171875, - "learning_rate": 5.820577352665252e-06, - "loss": 2.1639, + "grad_norm": 0.1533203125, + "learning_rate": 0.00014062504215791905, + "loss": 2.1307, "step": 2540 }, { "epoch": 0.86, - "grad_norm": 0.1689453125, - "learning_rate": 5.682881697280984e-06, - "loss": 2.185, + "grad_norm": 0.1552734375, + "learning_rate": 0.00014035470997461548, + "loss": 2.1206, "step": 2545 }, { "epoch": 0.86, - "grad_norm": 0.1669921875, - "learning_rate": 5.546736120993318e-06, - "loss": 2.1609, + "grad_norm": 0.15625, + "learning_rate": 0.00014008402505185952, + "loss": 2.1112, "step": 2550 }, { "epoch": 0.87, - "grad_norm": 0.16796875, - "learning_rate": 5.412145385763728e-06, - "loss": 2.1895, + "grad_norm": 0.1533203125, + "learning_rate": 0.000139812989755701, + "loss": 2.1233, "step": 2555 }, { "epoch": 0.87, - "grad_norm": 0.1708984375, - "learning_rate": 5.279114199170093e-06, - "loss": 2.1446, + "grad_norm": 0.1552734375, + "learning_rate": 0.00013954160645525217, + "loss": 2.1254, "step": 2560 }, { "epoch": 0.87, - "grad_norm": 0.173828125, - "learning_rate": 5.1476472142419965e-06, - "loss": 2.156, + "grad_norm": 0.158203125, + "learning_rate": 0.00013926987752266735, + "loss": 2.1362, "step": 2565 }, { "epoch": 0.87, - "grad_norm": 0.171875, - "learning_rate": 5.017749029297919e-06, - "loss": 2.1997, + "grad_norm": 0.1572265625, + "learning_rate": 0.0001389978053331219, + "loss": 2.1494, "step": 2570 }, { "epoch": 0.87, - "grad_norm": 0.1708984375, - "learning_rate": 4.889424187784486e-06, - "loss": 2.1674, + "grad_norm": 0.15234375, + "learning_rate": 0.00013872539226479172, + "loss": 2.139, "step": 2575 }, { "epoch": 0.87, - "grad_norm": 0.1748046875, - "learning_rate": 4.762677178117503e-06, - "loss": 2.1792, + "grad_norm": 0.16015625, + "learning_rate": 0.00013845264069883216, + "loss": 2.1307, "step": 2580 }, { "epoch": 0.88, - "grad_norm": 0.17578125, - "learning_rate": 4.637512433524987e-06, - "loss": 2.1648, + "grad_norm": 0.15625, + "learning_rate": 0.00013817955301935743, + "loss": 2.1316, "step": 2585 }, { "epoch": 0.88, - "grad_norm": 0.1650390625, - "learning_rate": 4.5139343318920945e-06, - "loss": 2.1833, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001379061316134198, + "loss": 2.1408, "step": 2590 }, { "epoch": 0.88, - "grad_norm": 0.166015625, - "learning_rate": 4.391947195607965e-06, - "loss": 2.1809, + "grad_norm": 0.162109375, + "learning_rate": 0.00013763237887098843, + "loss": 2.1299, "step": 2595 }, { "epoch": 0.88, - "grad_norm": 0.169921875, - "learning_rate": 4.271555291414636e-06, - "loss": 2.2085, + "grad_norm": 0.154296875, + "learning_rate": 0.0001373582971849289, + "loss": 2.1311, "step": 2600 }, { "epoch": 0.88, - "grad_norm": 0.166015625, - "learning_rate": 4.152762830257689e-06, - "loss": 2.153, + "grad_norm": 0.1552734375, + "learning_rate": 0.00013708388895098192, + "loss": 2.1357, "step": 2605 }, { "epoch": 0.88, - "grad_norm": 0.17578125, - "learning_rate": 4.035573967139023e-06, - "loss": 2.1964, + "grad_norm": 0.15625, + "learning_rate": 0.00013680915656774265, + "loss": 2.1451, "step": 2610 }, { "epoch": 0.89, - "grad_norm": 0.1748046875, - "learning_rate": 3.919992800971517e-06, - "loss": 2.1967, + "grad_norm": 0.1572265625, + "learning_rate": 0.00013653410243663952, + "loss": 2.1327, "step": 2615 }, { "epoch": 0.89, - "grad_norm": 0.1767578125, - "learning_rate": 3.8060233744356633e-06, - "loss": 2.1448, + "grad_norm": 0.154296875, + "learning_rate": 0.00013625872896191345, + "loss": 2.1629, "step": 2620 }, { "epoch": 0.89, - "grad_norm": 0.169921875, - "learning_rate": 3.6936696738381737e-06, - "loss": 2.151, + "grad_norm": 0.16015625, + "learning_rate": 0.0001359830385505967, + "loss": 2.1211, "step": 2625 }, { "epoch": 0.89, - "grad_norm": 0.1767578125, - "learning_rate": 3.5829356289725223e-06, - "loss": 2.1674, + "grad_norm": 0.1552734375, + "learning_rate": 0.00013570703361249188, + "loss": 2.1541, "step": 2630 }, { "epoch": 0.89, - "grad_norm": 0.171875, - "learning_rate": 3.473825112981527e-06, - "loss": 2.1734, + "grad_norm": 0.158203125, + "learning_rate": 0.00013543071656015084, + "loss": 2.122, "step": 2635 }, { "epoch": 0.89, - "grad_norm": 0.1748046875, - "learning_rate": 3.3663419422218677e-06, - "loss": 2.2007, + "grad_norm": 0.158203125, + "learning_rate": 0.0001351540898088536, + "loss": 2.1544, "step": 2640 }, { "epoch": 0.9, - "grad_norm": 0.1708984375, - "learning_rate": 3.260489876130568e-06, - "loss": 2.1475, + "grad_norm": 0.1552734375, + "learning_rate": 0.00013487715577658726, + "loss": 2.1587, "step": 2645 }, { "epoch": 0.9, - "grad_norm": 0.173828125, - "learning_rate": 3.156272617093553e-06, - "loss": 2.1635, + "grad_norm": 0.150390625, + "learning_rate": 0.00013459991688402492, + "loss": 2.1158, "step": 2650 }, { "epoch": 0.9, - "grad_norm": 0.1748046875, - "learning_rate": 3.0536938103161494e-06, - "loss": 2.212, + "grad_norm": 0.154296875, + "learning_rate": 0.00013432237555450444, + "loss": 2.1499, "step": 2655 }, { "epoch": 0.9, - "grad_norm": 0.1728515625, - "learning_rate": 2.9527570436955255e-06, - "loss": 2.1406, + "grad_norm": 0.154296875, + "learning_rate": 0.00013404453421400714, + "loss": 2.152, "step": 2660 }, { "epoch": 0.9, - "grad_norm": 0.1669921875, - "learning_rate": 2.8534658476952635e-06, - "loss": 2.2019, + "grad_norm": 0.15625, + "learning_rate": 0.00013376639529113688, + "loss": 2.1234, "step": 2665 }, { "epoch": 0.9, - "grad_norm": 0.1689453125, - "learning_rate": 2.7558236952218485e-06, - "loss": 2.2006, + "grad_norm": 0.15234375, + "learning_rate": 0.00013348796121709862, + "loss": 2.1329, "step": 2670 }, { "epoch": 0.91, - "grad_norm": 0.171875, - "learning_rate": 2.659834001503186e-06, - "loss": 2.1873, + "grad_norm": 0.1572265625, + "learning_rate": 0.00013320923442567727, + "loss": 2.1402, "step": 2675 }, { "epoch": 0.91, - "grad_norm": 0.1708984375, - "learning_rate": 2.5655001239691835e-06, - "loss": 2.1746, + "grad_norm": 0.15234375, + "learning_rate": 0.00013293021735321628, + "loss": 2.1526, "step": 2680 }, { "epoch": 0.91, - "grad_norm": 0.169921875, - "learning_rate": 2.4728253621342566e-06, - "loss": 2.1732, + "grad_norm": 0.1591796875, + "learning_rate": 0.00013265091243859652, + "loss": 2.1237, "step": 2685 }, { "epoch": 0.91, - "grad_norm": 0.1748046875, - "learning_rate": 2.381812957481999e-06, - "loss": 2.1399, + "grad_norm": 0.15625, + "learning_rate": 0.00013237132212321487, + "loss": 2.1443, "step": 2690 }, { "epoch": 0.91, - "grad_norm": 0.1728515625, - "learning_rate": 2.292466093351747e-06, - "loss": 2.18, + "grad_norm": 0.158203125, + "learning_rate": 0.0001320914488509629, + "loss": 2.1558, "step": 2695 }, { "epoch": 0.91, - "grad_norm": 0.1669921875, - "learning_rate": 2.2047878948272373e-06, - "loss": 2.1758, + "grad_norm": 0.1640625, + "learning_rate": 0.00013181129506820545, + "loss": 2.1194, "step": 2700 }, { "epoch": 0.92, - "grad_norm": 0.1796875, - "learning_rate": 2.1187814286273646e-06, - "loss": 2.1574, + "grad_norm": 0.158203125, + "learning_rate": 0.0001315308632237593, + "loss": 2.1201, "step": 2705 }, { "epoch": 0.92, - "grad_norm": 0.17578125, - "learning_rate": 2.0344497029988086e-06, - "loss": 2.163, + "grad_norm": 0.1533203125, + "learning_rate": 0.00013125015576887186, + "loss": 2.1523, "step": 2710 }, { "epoch": 0.92, - "grad_norm": 0.1748046875, - "learning_rate": 1.951795667610928e-06, - "loss": 2.163, + "grad_norm": 0.15625, + "learning_rate": 0.00013096917515719952, + "loss": 2.1443, "step": 2715 }, { "epoch": 0.92, - "grad_norm": 0.171875, - "learning_rate": 1.8708222134525167e-06, - "loss": 2.2011, + "grad_norm": 0.16015625, + "learning_rate": 0.00013068792384478636, + "loss": 2.1487, "step": 2720 }, { "epoch": 0.92, - "grad_norm": 0.173828125, - "learning_rate": 1.7915321727307088e-06, - "loss": 2.182, + "grad_norm": 0.1552734375, + "learning_rate": 0.00013040640429004267, + "loss": 2.1623, "step": 2725 }, { "epoch": 0.92, - "grad_norm": 0.171875, - "learning_rate": 1.7139283187719124e-06, - "loss": 2.1744, + "grad_norm": 0.1513671875, + "learning_rate": 0.00013012461895372344, + "loss": 2.1395, "step": 2730 }, { "epoch": 0.93, - "grad_norm": 0.1708984375, - "learning_rate": 1.6380133659248176e-06, - "loss": 2.1674, + "grad_norm": 0.15625, + "learning_rate": 0.00012984257029890683, + "loss": 2.1418, "step": 2735 }, { "epoch": 0.93, - "grad_norm": 0.1689453125, - "learning_rate": 1.5637899694654456e-06, - "loss": 2.1665, + "grad_norm": 0.1552734375, + "learning_rate": 0.00012956026079097272, + "loss": 2.1475, "step": 2740 }, { "epoch": 0.93, - "grad_norm": 0.1728515625, - "learning_rate": 1.4912607255042787e-06, - "loss": 2.1871, + "grad_norm": 0.1533203125, + "learning_rate": 0.000129277692897581, + "loss": 2.1373, "step": 2745 }, { "epoch": 0.93, - "grad_norm": 0.1806640625, - "learning_rate": 1.4204281708954437e-06, - "loss": 2.1737, + "grad_norm": 0.158203125, + "learning_rate": 0.00012899486908865012, + "loss": 2.138, "step": 2750 }, { "epoch": 0.93, - "grad_norm": 0.1767578125, - "learning_rate": 1.3512947831480217e-06, - "loss": 2.1824, + "grad_norm": 0.154296875, + "learning_rate": 0.0001287117918363356, + "loss": 2.1417, "step": 2755 }, { "epoch": 0.93, - "grad_norm": 0.171875, - "learning_rate": 1.2838629803393342e-06, - "loss": 2.1608, + "grad_norm": 0.15625, + "learning_rate": 0.00012842846361500816, + "loss": 2.1108, "step": 2760 }, { "epoch": 0.94, - "grad_norm": 0.169921875, - "learning_rate": 1.218135121030406e-06, - "loss": 2.1652, + "grad_norm": 0.1572265625, + "learning_rate": 0.00012814488690123226, + "loss": 2.1408, "step": 2765 }, { "epoch": 0.94, - "grad_norm": 0.1669921875, - "learning_rate": 1.1541135041834628e-06, - "loss": 2.1705, + "grad_norm": 0.158203125, + "learning_rate": 0.00012786106417374455, + "loss": 2.1762, "step": 2770 }, { "epoch": 0.94, - "grad_norm": 0.1748046875, - "learning_rate": 1.0918003690815138e-06, - "loss": 2.1644, + "grad_norm": 0.158203125, + "learning_rate": 0.00012757699791343186, + "loss": 2.1445, "step": 2775 }, { "epoch": 0.94, - "grad_norm": 0.1796875, - "learning_rate": 1.03119789525003e-06, - "loss": 2.1674, + "grad_norm": 0.15625, + "learning_rate": 0.00012729269060330999, + "loss": 2.1307, "step": 2780 }, { "epoch": 0.94, - "grad_norm": 0.173828125, - "learning_rate": 9.723082023807118e-07, - "loss": 2.158, + "grad_norm": 0.15234375, + "learning_rate": 0.0001270081447285015, + "loss": 2.1431, "step": 2785 }, { - "epoch": 0.95, - "grad_norm": 0.1748046875, - "learning_rate": 9.151333502573467e-07, - "loss": 2.2049, + "epoch": 0.94, + "grad_norm": 0.158203125, + "learning_rate": 0.00012672336277621442, + "loss": 2.1351, "step": 2790 }, { "epoch": 0.95, - "grad_norm": 0.1689453125, - "learning_rate": 8.596753386837797e-07, - "loss": 2.1732, + "grad_norm": 0.16015625, + "learning_rate": 0.0001264383472357202, + "loss": 2.1395, "step": 2795 }, { "epoch": 0.95, - "grad_norm": 0.171875, - "learning_rate": 8.059361074139293e-07, - "loss": 2.1608, + "grad_norm": 0.1611328125, + "learning_rate": 0.0001261531005983322, + "loss": 2.1492, "step": 2800 }, { "epoch": 0.95, - "grad_norm": 0.1845703125, - "learning_rate": 7.539175360839812e-07, - "loss": 2.1728, + "grad_norm": 0.158203125, + "learning_rate": 0.00012586762535738374, + "loss": 2.1344, "step": 2805 }, { "epoch": 0.95, - "grad_norm": 0.1669921875, - "learning_rate": 7.036214441466348e-07, - "loss": 2.1575, + "grad_norm": 0.158203125, + "learning_rate": 0.0001255819240082063, + "loss": 2.1352, "step": 2810 }, { "epoch": 0.95, - "grad_norm": 0.1708984375, - "learning_rate": 6.550495908074328e-07, - "loss": 2.179, + "grad_norm": 0.158203125, + "learning_rate": 0.00012529599904810784, + "loss": 2.1554, "step": 2815 }, { "epoch": 0.96, - "grad_norm": 0.171875, - "learning_rate": 6.082036749632703e-07, - "loss": 2.1894, + "grad_norm": 0.16015625, + "learning_rate": 0.00012500985297635088, + "loss": 2.125, "step": 2820 }, { "epoch": 0.96, - "grad_norm": 0.177734375, - "learning_rate": 5.630853351429599e-07, - "loss": 2.1787, + "grad_norm": 0.1591796875, + "learning_rate": 0.00012472348829413064, + "loss": 2.1478, "step": 2825 }, { "epoch": 0.96, - "grad_norm": 0.17578125, - "learning_rate": 5.196961494498997e-07, - "loss": 2.1696, + "grad_norm": 0.1591796875, + "learning_rate": 0.00012443690750455326, + "loss": 2.116, "step": 2830 }, { "epoch": 0.96, - "grad_norm": 0.1689453125, - "learning_rate": 4.780376355069172e-07, - "loss": 2.1568, + "grad_norm": 0.1533203125, + "learning_rate": 0.0001241501131126138, + "loss": 2.1667, "step": 2835 }, { "epoch": 0.96, - "grad_norm": 0.173828125, - "learning_rate": 4.381112504031337e-07, - "loss": 2.1699, + "grad_norm": 0.1572265625, + "learning_rate": 0.00012386310762517452, + "loss": 2.1338, "step": 2840 }, { "epoch": 0.96, - "grad_norm": 0.169921875, - "learning_rate": 3.9991839064305035e-07, - "loss": 2.1795, + "grad_norm": 0.158203125, + "learning_rate": 0.00012357589355094275, + "loss": 2.109, "step": 2845 }, { "epoch": 0.97, - "grad_norm": 0.169921875, - "learning_rate": 3.634603920976809e-07, - "loss": 2.1702, + "grad_norm": 0.1572265625, + "learning_rate": 0.0001232884734004491, + "loss": 2.1318, "step": 2850 }, { "epoch": 0.97, - "grad_norm": 0.1787109375, - "learning_rate": 3.2873852995781716e-07, - "loss": 2.1864, + "grad_norm": 0.154296875, + "learning_rate": 0.00012300084968602549, + "loss": 2.1117, "step": 2855 }, { "epoch": 0.97, - "grad_norm": 0.169921875, - "learning_rate": 2.957540186894481e-07, - "loss": 2.1621, + "grad_norm": 0.154296875, + "learning_rate": 0.00012271302492178327, + "loss": 2.1379, "step": 2860 }, { "epoch": 0.97, - "grad_norm": 0.1748046875, - "learning_rate": 2.6450801199126573e-07, - "loss": 2.2217, + "grad_norm": 0.1552734375, + "learning_rate": 0.00012242500162359105, + "loss": 2.122, "step": 2865 }, { "epoch": 0.97, - "grad_norm": 0.1689453125, - "learning_rate": 2.3500160275430893e-07, - "loss": 2.1848, + "grad_norm": 0.1591796875, + "learning_rate": 0.00012213678230905284, + "loss": 2.1651, "step": 2870 }, { "epoch": 0.97, - "grad_norm": 0.17578125, - "learning_rate": 2.0723582302376588e-07, - "loss": 2.187, + "grad_norm": 0.154296875, + "learning_rate": 0.00012184836949748608, + "loss": 2.1159, "step": 2875 }, { "epoch": 0.98, - "grad_norm": 0.1689453125, - "learning_rate": 1.812116439628364e-07, - "loss": 2.1751, + "grad_norm": 0.1611328125, + "learning_rate": 0.00012155976570989949, + "loss": 2.1241, "step": 2880 }, { "epoch": 0.98, - "grad_norm": 0.171875, - "learning_rate": 1.569299758187981e-07, - "loss": 2.1874, + "grad_norm": 0.1591796875, + "learning_rate": 0.0001212709734689712, + "loss": 2.1343, "step": 2885 }, { "epoch": 0.98, - "grad_norm": 0.1748046875, - "learning_rate": 1.3439166789113722e-07, - "loss": 2.158, + "grad_norm": 0.158203125, + "learning_rate": 0.00012098199529902648, + "loss": 2.134, "step": 2890 }, { "epoch": 0.98, - "grad_norm": 0.1669921875, - "learning_rate": 1.1359750850187256e-07, - "loss": 2.1644, + "grad_norm": 0.154296875, + "learning_rate": 0.0001206928337260159, + "loss": 2.1265, "step": 2895 }, { "epoch": 0.98, - "grad_norm": 0.17578125, - "learning_rate": 9.454822496796634e-08, - "loss": 2.1859, + "grad_norm": 0.158203125, + "learning_rate": 0.00012040349127749313, + "loss": 2.1531, "step": 2900 }, { "epoch": 0.98, - "grad_norm": 0.171875, - "learning_rate": 7.724448357588898e-08, - "loss": 2.1779, + "grad_norm": 0.1552734375, + "learning_rate": 0.00012011397048259285, + "loss": 2.1379, "step": 2905 }, { "epoch": 0.99, - "grad_norm": 0.1728515625, - "learning_rate": 6.168688955830448e-08, - "loss": 2.1375, + "grad_norm": 0.154296875, + "learning_rate": 0.00011982427387200867, + "loss": 2.1186, "step": 2910 }, { "epoch": 0.99, - "grad_norm": 0.171875, - "learning_rate": 4.7875987072915075e-08, - "loss": 2.1857, + "grad_norm": 0.158203125, + "learning_rate": 0.00011953440397797097, + "loss": 2.1584, "step": 2915 }, { "epoch": 0.99, - "grad_norm": 0.1708984375, - "learning_rate": 3.581225918342646e-08, - "loss": 2.1608, + "grad_norm": 0.15625, + "learning_rate": 0.00011924436333422489, + "loss": 2.1315, "step": 2920 }, { "epoch": 0.99, - "grad_norm": 0.169921875, - "learning_rate": 2.5496127842644658e-08, - "loss": 2.1335, + "grad_norm": 0.1552734375, + "learning_rate": 0.000118954154476008, + "loss": 2.1224, "step": 2925 }, { "epoch": 0.99, - "grad_norm": 0.171875, - "learning_rate": 1.692795387772117e-08, - "loss": 2.1946, + "grad_norm": 0.1572265625, + "learning_rate": 0.0001186637799400282, + "loss": 2.14, "step": 2930 }, { "epoch": 0.99, - "grad_norm": 0.1728515625, - "learning_rate": 1.0108036977535262e-08, - "loss": 2.1851, + "grad_norm": 0.1552734375, + "learning_rate": 0.00011837324226444169, + "loss": 2.1388, "step": 2935 }, { "epoch": 1.0, - "grad_norm": 0.171875, - "learning_rate": 5.03661568219127e-09, - "loss": 2.1765, + "grad_norm": 0.1552734375, + "learning_rate": 0.00011808254398883056, + "loss": 2.1444, "step": 2940 }, { "epoch": 1.0, - "grad_norm": 0.173828125, - "learning_rate": 1.7138673747196799e-09, - "loss": 2.2174, + "grad_norm": 0.1552734375, + "learning_rate": 0.00011779168765418079, + "loss": 2.1314, "step": 2945 }, { "epoch": 1.0, - "grad_norm": 0.1728515625, - "learning_rate": 1.3990827483212344e-10, - "loss": 2.1651, + "grad_norm": 0.1572265625, + "learning_rate": 0.00011750067580285988, + "loss": 2.1288, "step": 2950 }, { "epoch": 1.0, - "eval_loss": 2.1763968467712402, - "eval_runtime": 177.8602, - "eval_samples_per_second": 14.939, - "eval_steps_per_second": 1.872, + "eval_loss": 2.1319899559020996, + "eval_runtime": 173.5199, + "eval_samples_per_second": 7.653, + "eval_steps_per_second": 0.957, "step": 2952 }, { "epoch": 1.0, - "step": 2952, - "total_flos": 3.038696615046021e+17, - "train_loss": 2.200481324176478, - "train_runtime": 12060.097, - "train_samples_per_second": 3.917, - "train_steps_per_second": 0.245 + "grad_norm": 0.1572265625, + "learning_rate": 0.00011720951097859476, + "loss": 2.1113, + "step": 2955 + }, + { + "epoch": 1.0, + "grad_norm": 0.154296875, + "learning_rate": 0.00011691819572644939, + "loss": 2.0925, + "step": 2960 + }, + { + "epoch": 1.0, + "grad_norm": 0.15234375, + "learning_rate": 0.00011662673259280276, + "loss": 2.1001, + "step": 2965 + }, + { + "epoch": 1.01, + "grad_norm": 0.1572265625, + "learning_rate": 0.00011633512412532637, + "loss": 2.1235, + "step": 2970 + }, + { + "epoch": 1.01, + "grad_norm": 0.158203125, + "learning_rate": 0.0001160433728729621, + "loss": 2.1128, + "step": 2975 + }, + { + "epoch": 1.01, + "grad_norm": 0.158203125, + "learning_rate": 0.00011575148138589996, + "loss": 2.1216, + "step": 2980 + }, + { + "epoch": 1.01, + "grad_norm": 0.158203125, + "learning_rate": 0.0001154594522155557, + "loss": 2.1345, + "step": 2985 + }, + { + "epoch": 1.01, + "grad_norm": 0.1630859375, + "learning_rate": 0.00011516728791454861, + "loss": 2.1045, + "step": 2990 + }, + { + "epoch": 1.01, + "grad_norm": 0.1611328125, + "learning_rate": 0.00011487499103667904, + "loss": 2.0973, + "step": 2995 + }, + { + "epoch": 1.02, + "grad_norm": 0.16015625, + "learning_rate": 0.00011458256413690633, + "loss": 2.0934, + "step": 3000 + }, + { + "epoch": 1.02, + "grad_norm": 0.1630859375, + "learning_rate": 0.00011429000977132629, + "loss": 2.1147, + "step": 3005 + }, + { + "epoch": 1.02, + "grad_norm": 0.162109375, + "learning_rate": 0.00011399733049714884, + "loss": 2.1049, + "step": 3010 + }, + { + "epoch": 1.02, + "grad_norm": 0.1611328125, + "learning_rate": 0.00011370452887267582, + "loss": 2.0878, + "step": 3015 + }, + { + "epoch": 1.02, + "grad_norm": 0.1650390625, + "learning_rate": 0.00011341160745727844, + "loss": 2.1226, + "step": 3020 + }, + { + "epoch": 1.02, + "grad_norm": 0.16015625, + "learning_rate": 0.00011311856881137506, + "loss": 2.1125, + "step": 3025 + }, + { + "epoch": 1.03, + "grad_norm": 0.162109375, + "learning_rate": 0.00011282541549640873, + "loss": 2.0914, + "step": 3030 + }, + { + "epoch": 1.03, + "grad_norm": 0.1640625, + "learning_rate": 0.0001125321500748248, + "loss": 2.0832, + "step": 3035 + }, + { + "epoch": 1.03, + "grad_norm": 0.162109375, + "learning_rate": 0.00011223877511004863, + "loss": 2.1107, + "step": 3040 + }, + { + "epoch": 1.03, + "grad_norm": 0.1591796875, + "learning_rate": 0.00011194529316646293, + "loss": 2.1353, + "step": 3045 + }, + { + "epoch": 1.03, + "grad_norm": 0.1591796875, + "learning_rate": 0.00011165170680938572, + "loss": 2.1065, + "step": 3050 + }, + { + "epoch": 1.03, + "grad_norm": 0.1572265625, + "learning_rate": 0.00011135801860504749, + "loss": 2.1289, + "step": 3055 + }, + { + "epoch": 1.04, + "grad_norm": 0.1630859375, + "learning_rate": 0.00011106423112056911, + "loss": 2.1507, + "step": 3060 + }, + { + "epoch": 1.04, + "grad_norm": 0.16015625, + "learning_rate": 0.00011077034692393917, + "loss": 2.1053, + "step": 3065 + }, + { + "epoch": 1.04, + "grad_norm": 0.1591796875, + "learning_rate": 0.00011047636858399169, + "loss": 2.1196, + "step": 3070 + }, + { + "epoch": 1.04, + "grad_norm": 0.1611328125, + "learning_rate": 0.00011018229867038356, + "loss": 2.0859, + "step": 3075 + }, + { + "epoch": 1.04, + "grad_norm": 0.1650390625, + "learning_rate": 0.00010988813975357208, + "loss": 2.1273, + "step": 3080 + }, + { + "epoch": 1.04, + "grad_norm": 0.1591796875, + "learning_rate": 0.00010959389440479264, + "loss": 2.1134, + "step": 3085 + }, + { + "epoch": 1.05, + "grad_norm": 0.1611328125, + "learning_rate": 0.00010929956519603594, + "loss": 2.1101, + "step": 3090 + }, + { + "epoch": 1.05, + "grad_norm": 0.1650390625, + "learning_rate": 0.00010900515470002595, + "loss": 2.1221, + "step": 3095 + }, + { + "epoch": 1.05, + "grad_norm": 0.1630859375, + "learning_rate": 0.00010871066549019688, + "loss": 2.1026, + "step": 3100 + }, + { + "epoch": 1.05, + "grad_norm": 0.162109375, + "learning_rate": 0.0001084161001406712, + "loss": 2.1148, + "step": 3105 + }, + { + "epoch": 1.05, + "grad_norm": 0.16015625, + "learning_rate": 0.00010812146122623683, + "loss": 2.0942, + "step": 3110 + }, + { + "epoch": 1.06, + "grad_norm": 0.162109375, + "learning_rate": 0.00010782675132232474, + "loss": 2.1331, + "step": 3115 + }, + { + "epoch": 1.06, + "grad_norm": 0.1611328125, + "learning_rate": 0.00010753197300498638, + "loss": 2.1119, + "step": 3120 + }, + { + "epoch": 1.06, + "grad_norm": 0.1650390625, + "learning_rate": 0.00010723712885087123, + "loss": 2.0959, + "step": 3125 + }, + { + "epoch": 1.06, + "grad_norm": 0.162109375, + "learning_rate": 0.00010694222143720423, + "loss": 2.0654, + "step": 3130 + }, + { + "epoch": 1.06, + "grad_norm": 0.1611328125, + "learning_rate": 0.00010664725334176331, + "loss": 2.1195, + "step": 3135 + }, + { + "epoch": 1.06, + "grad_norm": 0.166015625, + "learning_rate": 0.00010635222714285676, + "loss": 2.1052, + "step": 3140 + }, + { + "epoch": 1.07, + "grad_norm": 0.16796875, + "learning_rate": 0.0001060571454193008, + "loss": 2.1174, + "step": 3145 + }, + { + "epoch": 1.07, + "grad_norm": 0.1640625, + "learning_rate": 0.00010576201075039696, + "loss": 2.0997, + "step": 3150 + }, + { + "epoch": 1.07, + "grad_norm": 0.166015625, + "learning_rate": 0.00010546682571590958, + "loss": 2.1202, + "step": 3155 + }, + { + "epoch": 1.07, + "grad_norm": 0.1640625, + "learning_rate": 0.00010517159289604324, + "loss": 2.1057, + "step": 3160 + }, + { + "epoch": 1.07, + "grad_norm": 0.162109375, + "learning_rate": 0.00010487631487142017, + "loss": 2.135, + "step": 3165 + }, + { + "epoch": 1.07, + "grad_norm": 0.162109375, + "learning_rate": 0.00010458099422305785, + "loss": 2.1585, + "step": 3170 + }, + { + "epoch": 1.08, + "grad_norm": 0.1669921875, + "learning_rate": 0.0001042856335323462, + "loss": 2.1096, + "step": 3175 + }, + { + "epoch": 1.08, + "grad_norm": 0.16015625, + "learning_rate": 0.00010399023538102522, + "loss": 2.1008, + "step": 3180 + }, + { + "epoch": 1.08, + "grad_norm": 0.16015625, + "learning_rate": 0.00010369480235116229, + "loss": 2.1126, + "step": 3185 + }, + { + "epoch": 1.08, + "grad_norm": 0.1650390625, + "learning_rate": 0.00010339933702512979, + "loss": 2.156, + "step": 3190 + }, + { + "epoch": 1.08, + "grad_norm": 0.1640625, + "learning_rate": 0.00010310384198558225, + "loss": 2.1252, + "step": 3195 + }, + { + "epoch": 1.08, + "grad_norm": 0.16015625, + "learning_rate": 0.00010280831981543405, + "loss": 2.0947, + "step": 3200 + }, + { + "epoch": 1.09, + "grad_norm": 0.1640625, + "learning_rate": 0.00010251277309783663, + "loss": 2.0965, + "step": 3205 + }, + { + "epoch": 1.09, + "grad_norm": 0.1591796875, + "learning_rate": 0.00010221720441615599, + "loss": 2.1106, + "step": 3210 + }, + { + "epoch": 1.09, + "grad_norm": 0.1630859375, + "learning_rate": 0.00010192161635395026, + "loss": 2.0715, + "step": 3215 + }, + { + "epoch": 1.09, + "grad_norm": 0.1591796875, + "learning_rate": 0.00010162601149494676, + "loss": 2.1249, + "step": 3220 + }, + { + "epoch": 1.09, + "grad_norm": 0.1640625, + "learning_rate": 0.00010133039242301985, + "loss": 2.0997, + "step": 3225 + }, + { + "epoch": 1.09, + "grad_norm": 0.1611328125, + "learning_rate": 0.00010103476172216792, + "loss": 2.1182, + "step": 3230 + }, + { + "epoch": 1.1, + "grad_norm": 0.162109375, + "learning_rate": 0.00010073912197649116, + "loss": 2.0769, + "step": 3235 + }, + { + "epoch": 1.1, + "grad_norm": 0.162109375, + "learning_rate": 0.0001004434757701688, + "loss": 2.1404, + "step": 3240 + }, + { + "epoch": 1.1, + "grad_norm": 0.1640625, + "learning_rate": 0.00010014782568743641, + "loss": 2.1354, + "step": 3245 + }, + { + "epoch": 1.1, + "grad_norm": 0.1630859375, + "learning_rate": 9.98521743125636e-05, + "loss": 2.1234, + "step": 3250 + }, + { + "epoch": 1.1, + "grad_norm": 0.1630859375, + "learning_rate": 9.955652422983122e-05, + "loss": 2.1039, + "step": 3255 + }, + { + "epoch": 1.1, + "grad_norm": 0.16796875, + "learning_rate": 9.926087802350886e-05, + "loss": 2.1161, + "step": 3260 + }, + { + "epoch": 1.11, + "grad_norm": 0.1650390625, + "learning_rate": 9.896523827783207e-05, + "loss": 2.1135, + "step": 3265 + }, + { + "epoch": 1.11, + "grad_norm": 0.1640625, + "learning_rate": 9.866960757698017e-05, + "loss": 2.1019, + "step": 3270 + }, + { + "epoch": 1.11, + "grad_norm": 0.162109375, + "learning_rate": 9.837398850505324e-05, + "loss": 2.1151, + "step": 3275 + }, + { + "epoch": 1.11, + "grad_norm": 0.162109375, + "learning_rate": 9.807838364604978e-05, + "loss": 2.1191, + "step": 3280 + }, + { + "epoch": 1.11, + "grad_norm": 0.1611328125, + "learning_rate": 9.7782795583844e-05, + "loss": 2.1127, + "step": 3285 + }, + { + "epoch": 1.11, + "grad_norm": 0.1630859375, + "learning_rate": 9.748722690216341e-05, + "loss": 2.1007, + "step": 3290 + }, + { + "epoch": 1.12, + "grad_norm": 0.1630859375, + "learning_rate": 9.719168018456598e-05, + "loss": 2.1043, + "step": 3295 + }, + { + "epoch": 1.12, + "grad_norm": 0.1611328125, + "learning_rate": 9.689615801441774e-05, + "loss": 2.1124, + "step": 3300 + }, + { + "epoch": 1.12, + "grad_norm": 0.1640625, + "learning_rate": 9.660066297487022e-05, + "loss": 2.126, + "step": 3305 + }, + { + "epoch": 1.12, + "grad_norm": 0.1611328125, + "learning_rate": 9.630519764883772e-05, + "loss": 2.1014, + "step": 3310 + }, + { + "epoch": 1.12, + "grad_norm": 0.1640625, + "learning_rate": 9.600976461897483e-05, + "loss": 2.1159, + "step": 3315 + }, + { + "epoch": 1.12, + "grad_norm": 0.1650390625, + "learning_rate": 9.571436646765382e-05, + "loss": 2.113, + "step": 3320 + }, + { + "epoch": 1.13, + "grad_norm": 0.162109375, + "learning_rate": 9.541900577694217e-05, + "loss": 2.1211, + "step": 3325 + }, + { + "epoch": 1.13, + "grad_norm": 0.1611328125, + "learning_rate": 9.512368512857984e-05, + "loss": 2.1062, + "step": 3330 + }, + { + "epoch": 1.13, + "grad_norm": 0.1630859375, + "learning_rate": 9.482840710395675e-05, + "loss": 2.1036, + "step": 3335 + }, + { + "epoch": 1.13, + "grad_norm": 0.1650390625, + "learning_rate": 9.453317428409044e-05, + "loss": 2.1095, + "step": 3340 + }, + { + "epoch": 1.13, + "grad_norm": 0.166015625, + "learning_rate": 9.423798924960306e-05, + "loss": 2.1065, + "step": 3345 + }, + { + "epoch": 1.13, + "grad_norm": 0.16796875, + "learning_rate": 9.394285458069923e-05, + "loss": 2.1398, + "step": 3350 + }, + { + "epoch": 1.14, + "grad_norm": 0.1669921875, + "learning_rate": 9.364777285714324e-05, + "loss": 2.1007, + "step": 3355 + }, + { + "epoch": 1.14, + "grad_norm": 0.1650390625, + "learning_rate": 9.33527466582367e-05, + "loss": 2.1188, + "step": 3360 + }, + { + "epoch": 1.14, + "grad_norm": 0.166015625, + "learning_rate": 9.30577785627958e-05, + "loss": 2.1267, + "step": 3365 + }, + { + "epoch": 1.14, + "grad_norm": 0.1630859375, + "learning_rate": 9.276287114912878e-05, + "loss": 2.1252, + "step": 3370 + }, + { + "epoch": 1.14, + "grad_norm": 0.16796875, + "learning_rate": 9.246802699501363e-05, + "loss": 2.1334, + "step": 3375 + }, + { + "epoch": 1.14, + "grad_norm": 0.1630859375, + "learning_rate": 9.217324867767527e-05, + "loss": 2.1118, + "step": 3380 + }, + { + "epoch": 1.15, + "grad_norm": 0.1650390625, + "learning_rate": 9.187853877376318e-05, + "loss": 2.1104, + "step": 3385 + }, + { + "epoch": 1.15, + "grad_norm": 0.1669921875, + "learning_rate": 9.158389985932881e-05, + "loss": 2.1238, + "step": 3390 + }, + { + "epoch": 1.15, + "grad_norm": 0.1640625, + "learning_rate": 9.128933450980314e-05, + "loss": 2.1228, + "step": 3395 + }, + { + "epoch": 1.15, + "grad_norm": 0.1611328125, + "learning_rate": 9.099484529997409e-05, + "loss": 2.0859, + "step": 3400 + }, + { + "epoch": 1.15, + "grad_norm": 0.166015625, + "learning_rate": 9.070043480396404e-05, + "loss": 2.0926, + "step": 3405 + }, + { + "epoch": 1.15, + "grad_norm": 0.1640625, + "learning_rate": 9.04061055952074e-05, + "loss": 2.1346, + "step": 3410 + }, + { + "epoch": 1.16, + "grad_norm": 0.162109375, + "learning_rate": 9.011186024642793e-05, + "loss": 2.096, + "step": 3415 + }, + { + "epoch": 1.16, + "grad_norm": 0.162109375, + "learning_rate": 8.981770132961649e-05, + "loss": 2.1154, + "step": 3420 + }, + { + "epoch": 1.16, + "grad_norm": 0.162109375, + "learning_rate": 8.952363141600834e-05, + "loss": 2.1073, + "step": 3425 + }, + { + "epoch": 1.16, + "grad_norm": 0.1728515625, + "learning_rate": 8.922965307606086e-05, + "loss": 2.1085, + "step": 3430 + }, + { + "epoch": 1.16, + "grad_norm": 0.1630859375, + "learning_rate": 8.893576887943094e-05, + "loss": 2.0935, + "step": 3435 + }, + { + "epoch": 1.17, + "grad_norm": 0.1669921875, + "learning_rate": 8.86419813949525e-05, + "loss": 2.0911, + "step": 3440 + }, + { + "epoch": 1.17, + "grad_norm": 0.166015625, + "learning_rate": 8.834829319061431e-05, + "loss": 2.1139, + "step": 3445 + }, + { + "epoch": 1.17, + "grad_norm": 0.16796875, + "learning_rate": 8.805470683353708e-05, + "loss": 2.1199, + "step": 3450 + }, + { + "epoch": 1.17, + "grad_norm": 0.1728515625, + "learning_rate": 8.77612248899514e-05, + "loss": 2.1038, + "step": 3455 + }, + { + "epoch": 1.17, + "grad_norm": 0.1630859375, + "learning_rate": 8.746784992517518e-05, + "loss": 2.0909, + "step": 3460 + }, + { + "epoch": 1.17, + "grad_norm": 0.1640625, + "learning_rate": 8.71745845035913e-05, + "loss": 2.1016, + "step": 3465 + }, + { + "epoch": 1.18, + "grad_norm": 0.1640625, + "learning_rate": 8.688143118862499e-05, + "loss": 2.1213, + "step": 3470 + }, + { + "epoch": 1.18, + "grad_norm": 0.16796875, + "learning_rate": 8.658839254272157e-05, + "loss": 2.1115, + "step": 3475 + }, + { + "epoch": 1.18, + "grad_norm": 0.169921875, + "learning_rate": 8.62954711273242e-05, + "loss": 2.1116, + "step": 3480 + }, + { + "epoch": 1.18, + "grad_norm": 0.162109375, + "learning_rate": 8.600266950285117e-05, + "loss": 2.1493, + "step": 3485 + }, + { + "epoch": 1.18, + "grad_norm": 0.1689453125, + "learning_rate": 8.570999022867373e-05, + "loss": 2.1146, + "step": 3490 + }, + { + "epoch": 1.18, + "grad_norm": 0.16796875, + "learning_rate": 8.541743586309365e-05, + "loss": 2.0887, + "step": 3495 + }, + { + "epoch": 1.19, + "grad_norm": 0.166015625, + "learning_rate": 8.512500896332097e-05, + "loss": 2.1072, + "step": 3500 + }, + { + "epoch": 1.19, + "grad_norm": 0.16796875, + "learning_rate": 8.483271208545144e-05, + "loss": 2.0959, + "step": 3505 + }, + { + "epoch": 1.19, + "grad_norm": 0.169921875, + "learning_rate": 8.454054778444431e-05, + "loss": 2.1305, + "step": 3510 + }, + { + "epoch": 1.19, + "grad_norm": 0.166015625, + "learning_rate": 8.424851861410007e-05, + "loss": 2.0803, + "step": 3515 + }, + { + "epoch": 1.19, + "grad_norm": 0.1728515625, + "learning_rate": 8.395662712703793e-05, + "loss": 2.0963, + "step": 3520 + }, + { + "epoch": 1.19, + "grad_norm": 0.16796875, + "learning_rate": 8.366487587467368e-05, + "loss": 2.1136, + "step": 3525 + }, + { + "epoch": 1.2, + "grad_norm": 0.16796875, + "learning_rate": 8.337326740719726e-05, + "loss": 2.0984, + "step": 3530 + }, + { + "epoch": 1.2, + "grad_norm": 0.1640625, + "learning_rate": 8.308180427355062e-05, + "loss": 2.1252, + "step": 3535 + }, + { + "epoch": 1.2, + "grad_norm": 0.1640625, + "learning_rate": 8.279048902140528e-05, + "loss": 2.1101, + "step": 3540 + }, + { + "epoch": 1.2, + "grad_norm": 0.16796875, + "learning_rate": 8.24993241971401e-05, + "loss": 2.1124, + "step": 3545 + }, + { + "epoch": 1.2, + "grad_norm": 0.1669921875, + "learning_rate": 8.220831234581922e-05, + "loss": 2.1135, + "step": 3550 + }, + { + "epoch": 1.2, + "grad_norm": 0.166015625, + "learning_rate": 8.191745601116947e-05, + "loss": 2.1134, + "step": 3555 + }, + { + "epoch": 1.21, + "grad_norm": 0.16015625, + "learning_rate": 8.162675773555836e-05, + "loss": 2.098, + "step": 3560 + }, + { + "epoch": 1.21, + "grad_norm": 0.162109375, + "learning_rate": 8.133622005997181e-05, + "loss": 2.095, + "step": 3565 + }, + { + "epoch": 1.21, + "grad_norm": 0.1689453125, + "learning_rate": 8.104584552399204e-05, + "loss": 2.1298, + "step": 3570 + }, + { + "epoch": 1.21, + "grad_norm": 0.166015625, + "learning_rate": 8.075563666577515e-05, + "loss": 2.1018, + "step": 3575 + }, + { + "epoch": 1.21, + "grad_norm": 0.1708984375, + "learning_rate": 8.046559602202901e-05, + "loss": 2.1074, + "step": 3580 + }, + { + "epoch": 1.21, + "grad_norm": 0.16796875, + "learning_rate": 8.017572612799135e-05, + "loss": 2.1308, + "step": 3585 + }, + { + "epoch": 1.22, + "grad_norm": 0.162109375, + "learning_rate": 7.988602951740717e-05, + "loss": 2.1215, + "step": 3590 + }, + { + "epoch": 1.22, + "grad_norm": 0.166015625, + "learning_rate": 7.959650872250688e-05, + "loss": 2.1063, + "step": 3595 + }, + { + "epoch": 1.22, + "grad_norm": 0.1650390625, + "learning_rate": 7.930716627398412e-05, + "loss": 2.1156, + "step": 3600 + }, + { + "epoch": 1.22, + "grad_norm": 0.16796875, + "learning_rate": 7.901800470097355e-05, + "loss": 2.1169, + "step": 3605 + }, + { + "epoch": 1.22, + "grad_norm": 0.166015625, + "learning_rate": 7.872902653102884e-05, + "loss": 2.0942, + "step": 3610 + }, + { + "epoch": 1.22, + "grad_norm": 0.169921875, + "learning_rate": 7.84402342901005e-05, + "loss": 2.1382, + "step": 3615 + }, + { + "epoch": 1.23, + "grad_norm": 0.169921875, + "learning_rate": 7.815163050251395e-05, + "loss": 2.1234, + "step": 3620 + }, + { + "epoch": 1.23, + "grad_norm": 0.1630859375, + "learning_rate": 7.786321769094717e-05, + "loss": 2.1158, + "step": 3625 + }, + { + "epoch": 1.23, + "grad_norm": 0.16796875, + "learning_rate": 7.7574998376409e-05, + "loss": 2.1071, + "step": 3630 + }, + { + "epoch": 1.23, + "grad_norm": 0.1689453125, + "learning_rate": 7.728697507821674e-05, + "loss": 2.1165, + "step": 3635 + }, + { + "epoch": 1.23, + "grad_norm": 0.1689453125, + "learning_rate": 7.699915031397452e-05, + "loss": 2.1276, + "step": 3640 + }, + { + "epoch": 1.23, + "grad_norm": 0.1689453125, + "learning_rate": 7.671152659955096e-05, + "loss": 2.12, + "step": 3645 + }, + { + "epoch": 1.24, + "grad_norm": 0.1630859375, + "learning_rate": 7.642410644905726e-05, + "loss": 2.1144, + "step": 3650 + }, + { + "epoch": 1.24, + "grad_norm": 0.162109375, + "learning_rate": 7.613689237482551e-05, + "loss": 2.1058, + "step": 3655 + }, + { + "epoch": 1.24, + "grad_norm": 0.1650390625, + "learning_rate": 7.584988688738622e-05, + "loss": 2.0953, + "step": 3660 + }, + { + "epoch": 1.24, + "grad_norm": 0.1650390625, + "learning_rate": 7.556309249544678e-05, + "loss": 2.1459, + "step": 3665 + }, + { + "epoch": 1.24, + "grad_norm": 0.16796875, + "learning_rate": 7.527651170586936e-05, + "loss": 2.1061, + "step": 3670 + }, + { + "epoch": 1.24, + "grad_norm": 0.1640625, + "learning_rate": 7.499014702364913e-05, + "loss": 2.1119, + "step": 3675 + }, + { + "epoch": 1.25, + "grad_norm": 0.166015625, + "learning_rate": 7.470400095189219e-05, + "loss": 2.1395, + "step": 3680 + }, + { + "epoch": 1.25, + "grad_norm": 0.1640625, + "learning_rate": 7.44180759917937e-05, + "loss": 2.0965, + "step": 3685 + }, + { + "epoch": 1.25, + "grad_norm": 0.1708984375, + "learning_rate": 7.413237464261627e-05, + "loss": 2.1035, + "step": 3690 + }, + { + "epoch": 1.25, + "grad_norm": 0.1689453125, + "learning_rate": 7.38468994016678e-05, + "loss": 2.1184, + "step": 3695 + }, + { + "epoch": 1.25, + "grad_norm": 0.1669921875, + "learning_rate": 7.356165276427983e-05, + "loss": 2.1213, + "step": 3700 + }, + { + "epoch": 1.25, + "grad_norm": 0.1611328125, + "learning_rate": 7.327663722378561e-05, + "loss": 2.0926, + "step": 3705 + }, + { + "epoch": 1.26, + "grad_norm": 0.1640625, + "learning_rate": 7.299185527149853e-05, + "loss": 2.1019, + "step": 3710 + }, + { + "epoch": 1.26, + "grad_norm": 0.166015625, + "learning_rate": 7.270730939669006e-05, + "loss": 2.0903, + "step": 3715 + }, + { + "epoch": 1.26, + "grad_norm": 0.166015625, + "learning_rate": 7.242300208656814e-05, + "loss": 2.1252, + "step": 3720 + }, + { + "epoch": 1.26, + "grad_norm": 0.162109375, + "learning_rate": 7.213893582625548e-05, + "loss": 2.1357, + "step": 3725 + }, + { + "epoch": 1.26, + "grad_norm": 0.169921875, + "learning_rate": 7.185511309876775e-05, + "loss": 2.0908, + "step": 3730 + }, + { + "epoch": 1.27, + "grad_norm": 0.166015625, + "learning_rate": 7.157153638499188e-05, + "loss": 2.0689, + "step": 3735 + }, + { + "epoch": 1.27, + "grad_norm": 0.1640625, + "learning_rate": 7.128820816366442e-05, + "loss": 2.1063, + "step": 3740 + }, + { + "epoch": 1.27, + "grad_norm": 0.1689453125, + "learning_rate": 7.100513091134989e-05, + "loss": 2.1295, + "step": 3745 + }, + { + "epoch": 1.27, + "grad_norm": 0.173828125, + "learning_rate": 7.072230710241905e-05, + "loss": 2.1386, + "step": 3750 + }, + { + "epoch": 1.27, + "grad_norm": 0.1689453125, + "learning_rate": 7.043973920902729e-05, + "loss": 2.1216, + "step": 3755 + }, + { + "epoch": 1.27, + "grad_norm": 0.1640625, + "learning_rate": 7.015742970109317e-05, + "loss": 2.1305, + "step": 3760 + }, + { + "epoch": 1.28, + "grad_norm": 0.1630859375, + "learning_rate": 6.98753810462766e-05, + "loss": 2.1098, + "step": 3765 + }, + { + "epoch": 1.28, + "grad_norm": 0.16796875, + "learning_rate": 6.959359570995738e-05, + "loss": 2.1271, + "step": 3770 + }, + { + "epoch": 1.28, + "grad_norm": 0.169921875, + "learning_rate": 6.931207615521366e-05, + "loss": 2.1041, + "step": 3775 + }, + { + "epoch": 1.28, + "grad_norm": 0.16796875, + "learning_rate": 6.903082484280053e-05, + "loss": 2.114, + "step": 3780 + }, + { + "epoch": 1.28, + "grad_norm": 0.169921875, + "learning_rate": 6.874984423112819e-05, + "loss": 2.1364, + "step": 3785 + }, + { + "epoch": 1.28, + "grad_norm": 0.1708984375, + "learning_rate": 6.84691367762407e-05, + "loss": 2.117, + "step": 3790 + }, + { + "epoch": 1.29, + "grad_norm": 0.16796875, + "learning_rate": 6.818870493179458e-05, + "loss": 2.0934, + "step": 3795 + }, + { + "epoch": 1.29, + "grad_norm": 0.1650390625, + "learning_rate": 6.790855114903714e-05, + "loss": 2.1207, + "step": 3800 + }, + { + "epoch": 1.29, + "grad_norm": 0.1640625, + "learning_rate": 6.762867787678512e-05, + "loss": 2.121, + "step": 3805 + }, + { + "epoch": 1.29, + "grad_norm": 0.1650390625, + "learning_rate": 6.73490875614035e-05, + "loss": 2.1113, + "step": 3810 + }, + { + "epoch": 1.29, + "grad_norm": 0.171875, + "learning_rate": 6.706978264678376e-05, + "loss": 2.1088, + "step": 3815 + }, + { + "epoch": 1.29, + "grad_norm": 0.1708984375, + "learning_rate": 6.679076557432278e-05, + "loss": 2.1221, + "step": 3820 + }, + { + "epoch": 1.3, + "grad_norm": 0.16796875, + "learning_rate": 6.651203878290139e-05, + "loss": 2.1239, + "step": 3825 + }, + { + "epoch": 1.3, + "grad_norm": 0.1650390625, + "learning_rate": 6.623360470886314e-05, + "loss": 2.1229, + "step": 3830 + }, + { + "epoch": 1.3, + "grad_norm": 0.1669921875, + "learning_rate": 6.59554657859929e-05, + "loss": 2.0984, + "step": 3835 + }, + { + "epoch": 1.3, + "grad_norm": 0.16796875, + "learning_rate": 6.567762444549558e-05, + "loss": 2.1042, + "step": 3840 + }, + { + "epoch": 1.3, + "grad_norm": 0.1630859375, + "learning_rate": 6.540008311597507e-05, + "loss": 2.1203, + "step": 3845 + }, + { + "epoch": 1.3, + "grad_norm": 0.1669921875, + "learning_rate": 6.512284422341275e-05, + "loss": 2.0983, + "step": 3850 + }, + { + "epoch": 1.31, + "grad_norm": 0.166015625, + "learning_rate": 6.484591019114646e-05, + "loss": 2.112, + "step": 3855 + }, + { + "epoch": 1.31, + "grad_norm": 0.1650390625, + "learning_rate": 6.456928343984919e-05, + "loss": 2.1393, + "step": 3860 + }, + { + "epoch": 1.31, + "grad_norm": 0.16796875, + "learning_rate": 6.429296638750814e-05, + "loss": 2.1089, + "step": 3865 + }, + { + "epoch": 1.31, + "grad_norm": 0.16796875, + "learning_rate": 6.401696144940332e-05, + "loss": 2.0942, + "step": 3870 + }, + { + "epoch": 1.31, + "grad_norm": 0.1669921875, + "learning_rate": 6.374127103808654e-05, + "loss": 2.1108, + "step": 3875 + }, + { + "epoch": 1.31, + "grad_norm": 0.1669921875, + "learning_rate": 6.34658975633605e-05, + "loss": 2.1063, + "step": 3880 + }, + { + "epoch": 1.32, + "grad_norm": 0.1689453125, + "learning_rate": 6.319084343225738e-05, + "loss": 2.1309, + "step": 3885 + }, + { + "epoch": 1.32, + "grad_norm": 0.1689453125, + "learning_rate": 6.291611104901812e-05, + "loss": 2.1287, + "step": 3890 + }, + { + "epoch": 1.32, + "grad_norm": 0.1650390625, + "learning_rate": 6.264170281507111e-05, + "loss": 2.1196, + "step": 3895 + }, + { + "epoch": 1.32, + "grad_norm": 0.166015625, + "learning_rate": 6.236762112901158e-05, + "loss": 2.1202, + "step": 3900 + }, + { + "epoch": 1.32, + "grad_norm": 0.1689453125, + "learning_rate": 6.209386838658024e-05, + "loss": 2.1264, + "step": 3905 + }, + { + "epoch": 1.32, + "grad_norm": 0.169921875, + "learning_rate": 6.182044698064256e-05, + "loss": 2.1083, + "step": 3910 + }, + { + "epoch": 1.33, + "grad_norm": 0.166015625, + "learning_rate": 6.154735930116786e-05, + "loss": 2.1208, + "step": 3915 + }, + { + "epoch": 1.33, + "grad_norm": 0.169921875, + "learning_rate": 6.12746077352083e-05, + "loss": 2.0995, + "step": 3920 + }, + { + "epoch": 1.33, + "grad_norm": 0.1650390625, + "learning_rate": 6.1002194666878106e-05, + "loss": 2.1249, + "step": 3925 + }, + { + "epoch": 1.33, + "grad_norm": 0.1669921875, + "learning_rate": 6.0730122477332675e-05, + "loss": 2.1024, + "step": 3930 + }, + { + "epoch": 1.33, + "grad_norm": 0.169921875, + "learning_rate": 6.045839354474786e-05, + "loss": 2.09, + "step": 3935 + }, + { + "epoch": 1.33, + "grad_norm": 0.1630859375, + "learning_rate": 6.0187010244299046e-05, + "loss": 2.0995, + "step": 3940 + }, + { + "epoch": 1.34, + "grad_norm": 0.169921875, + "learning_rate": 5.9915974948140474e-05, + "loss": 2.101, + "step": 3945 + }, + { + "epoch": 1.34, + "grad_norm": 0.1669921875, + "learning_rate": 5.964529002538455e-05, + "loss": 2.0827, + "step": 3950 + }, + { + "epoch": 1.34, + "grad_norm": 0.1669921875, + "learning_rate": 5.937495784208096e-05, + "loss": 2.1324, + "step": 3955 + }, + { + "epoch": 1.34, + "grad_norm": 0.1728515625, + "learning_rate": 5.910498076119622e-05, + "loss": 2.1116, + "step": 3960 + }, + { + "epoch": 1.34, + "grad_norm": 0.1669921875, + "learning_rate": 5.883536114259277e-05, + "loss": 2.0813, + "step": 3965 + }, + { + "epoch": 1.34, + "grad_norm": 0.166015625, + "learning_rate": 5.8566101343008687e-05, + "loss": 2.1031, + "step": 3970 + }, + { + "epoch": 1.35, + "grad_norm": 0.1708984375, + "learning_rate": 5.829720371603664e-05, + "loss": 2.1116, + "step": 3975 + }, + { + "epoch": 1.35, + "grad_norm": 0.1689453125, + "learning_rate": 5.802867061210375e-05, + "loss": 2.1094, + "step": 3980 + }, + { + "epoch": 1.35, + "grad_norm": 0.169921875, + "learning_rate": 5.776050437845075e-05, + "loss": 2.1008, + "step": 3985 + }, + { + "epoch": 1.35, + "grad_norm": 0.1708984375, + "learning_rate": 5.749270735911158e-05, + "loss": 2.1074, + "step": 3990 + }, + { + "epoch": 1.35, + "grad_norm": 0.16796875, + "learning_rate": 5.7225281894892935e-05, + "loss": 2.1045, + "step": 3995 + }, + { + "epoch": 1.35, + "grad_norm": 0.1669921875, + "learning_rate": 5.695823032335366e-05, + "loss": 2.1119, + "step": 4000 + }, + { + "epoch": 1.36, + "grad_norm": 0.1689453125, + "learning_rate": 5.669155497878454e-05, + "loss": 2.1185, + "step": 4005 + }, + { + "epoch": 1.36, + "grad_norm": 0.1650390625, + "learning_rate": 5.642525819218769e-05, + "loss": 2.0826, + "step": 4010 + }, + { + "epoch": 1.36, + "grad_norm": 0.166015625, + "learning_rate": 5.6159342291256254e-05, + "loss": 2.0952, + "step": 4015 + }, + { + "epoch": 1.36, + "grad_norm": 0.16796875, + "learning_rate": 5.589380960035417e-05, + "loss": 2.1267, + "step": 4020 + }, + { + "epoch": 1.36, + "grad_norm": 0.1640625, + "learning_rate": 5.562866244049557e-05, + "loss": 2.1282, + "step": 4025 + }, + { + "epoch": 1.36, + "grad_norm": 0.16796875, + "learning_rate": 5.53639031293248e-05, + "loss": 2.1114, + "step": 4030 + }, + { + "epoch": 1.37, + "grad_norm": 0.1728515625, + "learning_rate": 5.509953398109594e-05, + "loss": 2.1173, + "step": 4035 + }, + { + "epoch": 1.37, + "grad_norm": 0.166015625, + "learning_rate": 5.483555730665282e-05, + "loss": 2.0859, + "step": 4040 + }, + { + "epoch": 1.37, + "grad_norm": 0.16796875, + "learning_rate": 5.457197541340853e-05, + "loss": 2.1138, + "step": 4045 + }, + { + "epoch": 1.37, + "grad_norm": 0.1669921875, + "learning_rate": 5.4308790605325364e-05, + "loss": 2.1407, + "step": 4050 + }, + { + "epoch": 1.37, + "grad_norm": 0.166015625, + "learning_rate": 5.404600518289487e-05, + "loss": 2.1156, + "step": 4055 + }, + { + "epoch": 1.38, + "grad_norm": 0.166015625, + "learning_rate": 5.3783621443117414e-05, + "loss": 2.1036, + "step": 4060 + }, + { + "epoch": 1.38, + "grad_norm": 0.1650390625, + "learning_rate": 5.352164167948233e-05, + "loss": 2.1022, + "step": 4065 + }, + { + "epoch": 1.38, + "grad_norm": 0.1689453125, + "learning_rate": 5.326006818194782e-05, + "loss": 2.1037, + "step": 4070 + }, + { + "epoch": 1.38, + "grad_norm": 0.1669921875, + "learning_rate": 5.2998903236920895e-05, + "loss": 2.1189, + "step": 4075 + }, + { + "epoch": 1.38, + "grad_norm": 0.171875, + "learning_rate": 5.273814912723742e-05, + "loss": 2.1133, + "step": 4080 + }, + { + "epoch": 1.38, + "grad_norm": 0.169921875, + "learning_rate": 5.247780813214214e-05, + "loss": 2.1261, + "step": 4085 + }, + { + "epoch": 1.39, + "grad_norm": 0.171875, + "learning_rate": 5.221788252726889e-05, + "loss": 2.1114, + "step": 4090 + }, + { + "epoch": 1.39, + "grad_norm": 0.1689453125, + "learning_rate": 5.195837458462045e-05, + "loss": 2.1119, + "step": 4095 + }, + { + "epoch": 1.39, + "grad_norm": 0.16796875, + "learning_rate": 5.16992865725489e-05, + "loss": 2.125, + "step": 4100 + }, + { + "epoch": 1.39, + "grad_norm": 0.1669921875, + "learning_rate": 5.14406207557357e-05, + "loss": 2.0831, + "step": 4105 + }, + { + "epoch": 1.39, + "grad_norm": 0.1708984375, + "learning_rate": 5.11823793951719e-05, + "loss": 2.1304, + "step": 4110 + }, + { + "epoch": 1.39, + "grad_norm": 0.17578125, + "learning_rate": 5.092456474813841e-05, + "loss": 2.1066, + "step": 4115 + }, + { + "epoch": 1.4, + "grad_norm": 0.1640625, + "learning_rate": 5.066717906818618e-05, + "loss": 2.1206, + "step": 4120 + }, + { + "epoch": 1.4, + "grad_norm": 0.1611328125, + "learning_rate": 5.041022460511673e-05, + "loss": 2.1133, + "step": 4125 + }, + { + "epoch": 1.4, + "grad_norm": 0.169921875, + "learning_rate": 5.015370360496219e-05, + "loss": 2.1384, + "step": 4130 + }, + { + "epoch": 1.4, + "grad_norm": 0.166015625, + "learning_rate": 4.989761830996581e-05, + "loss": 2.1042, + "step": 4135 + }, + { + "epoch": 1.4, + "grad_norm": 0.1640625, + "learning_rate": 4.9641970958562366e-05, + "loss": 2.1241, + "step": 4140 + }, + { + "epoch": 1.4, + "grad_norm": 0.171875, + "learning_rate": 4.938676378535866e-05, + "loss": 2.1237, + "step": 4145 + }, + { + "epoch": 1.41, + "grad_norm": 0.1787109375, + "learning_rate": 4.913199902111385e-05, + "loss": 2.1187, + "step": 4150 + }, + { + "epoch": 1.41, + "grad_norm": 0.166015625, + "learning_rate": 4.8877678892719866e-05, + "loss": 2.0729, + "step": 4155 + }, + { + "epoch": 1.41, + "grad_norm": 0.171875, + "learning_rate": 4.862380562318236e-05, + "loss": 2.1087, + "step": 4160 + }, + { + "epoch": 1.41, + "grad_norm": 0.16796875, + "learning_rate": 4.837038143160082e-05, + "loss": 2.118, + "step": 4165 + }, + { + "epoch": 1.41, + "grad_norm": 0.166015625, + "learning_rate": 4.811740853314939e-05, + "loss": 2.118, + "step": 4170 + }, + { + "epoch": 1.41, + "grad_norm": 0.169921875, + "learning_rate": 4.786488913905745e-05, + "loss": 2.1102, + "step": 4175 + }, + { + "epoch": 1.42, + "grad_norm": 0.169921875, + "learning_rate": 4.7612825456590435e-05, + "loss": 2.0915, + "step": 4180 + }, + { + "epoch": 1.42, + "grad_norm": 0.16796875, + "learning_rate": 4.736121968903027e-05, + "loss": 2.1321, + "step": 4185 + }, + { + "epoch": 1.42, + "grad_norm": 0.16796875, + "learning_rate": 4.7110074035656316e-05, + "loss": 2.0898, + "step": 4190 + }, + { + "epoch": 1.42, + "grad_norm": 0.16796875, + "learning_rate": 4.685939069172609e-05, + "loss": 2.1212, + "step": 4195 + }, + { + "epoch": 1.42, + "grad_norm": 0.1669921875, + "learning_rate": 4.6609171848456066e-05, + "loss": 2.0965, + "step": 4200 + }, + { + "epoch": 1.42, + "grad_norm": 0.1669921875, + "learning_rate": 4.6359419693002534e-05, + "loss": 2.095, + "step": 4205 + }, + { + "epoch": 1.43, + "grad_norm": 0.1728515625, + "learning_rate": 4.611013640844245e-05, + "loss": 2.118, + "step": 4210 + }, + { + "epoch": 1.43, + "grad_norm": 0.1689453125, + "learning_rate": 4.5861324173754484e-05, + "loss": 2.1134, + "step": 4215 + }, + { + "epoch": 1.43, + "grad_norm": 0.1689453125, + "learning_rate": 4.561298516379974e-05, + "loss": 2.1027, + "step": 4220 + }, + { + "epoch": 1.43, + "grad_norm": 0.166015625, + "learning_rate": 4.5365121549302914e-05, + "loss": 2.0939, + "step": 4225 + }, + { + "epoch": 1.43, + "grad_norm": 0.169921875, + "learning_rate": 4.5117735496833415e-05, + "loss": 2.1279, + "step": 4230 + }, + { + "epoch": 1.43, + "grad_norm": 0.166015625, + "learning_rate": 4.487082916878606e-05, + "loss": 2.1078, + "step": 4235 + }, + { + "epoch": 1.44, + "grad_norm": 0.1650390625, + "learning_rate": 4.4624404723362576e-05, + "loss": 2.1143, + "step": 4240 + }, + { + "epoch": 1.44, + "grad_norm": 0.1650390625, + "learning_rate": 4.437846431455249e-05, + "loss": 2.0946, + "step": 4245 + }, + { + "epoch": 1.44, + "grad_norm": 0.171875, + "learning_rate": 4.4133010092114494e-05, + "loss": 2.1203, + "step": 4250 + }, + { + "epoch": 1.44, + "grad_norm": 0.166015625, + "learning_rate": 4.3888044201557376e-05, + "loss": 2.1112, + "step": 4255 + }, + { + "epoch": 1.44, + "grad_norm": 0.1708984375, + "learning_rate": 4.36435687841215e-05, + "loss": 2.0873, + "step": 4260 + }, + { + "epoch": 1.44, + "grad_norm": 0.173828125, + "learning_rate": 4.3399585976760105e-05, + "loss": 2.1056, + "step": 4265 + }, + { + "epoch": 1.45, + "grad_norm": 0.169921875, + "learning_rate": 4.3156097912120385e-05, + "loss": 2.101, + "step": 4270 + }, + { + "epoch": 1.45, + "grad_norm": 0.1728515625, + "learning_rate": 4.29131067185251e-05, + "loss": 2.1302, + "step": 4275 + }, + { + "epoch": 1.45, + "grad_norm": 0.1689453125, + "learning_rate": 4.2670614519953834e-05, + "loss": 2.0862, + "step": 4280 + }, + { + "epoch": 1.45, + "grad_norm": 0.169921875, + "learning_rate": 4.242862343602447e-05, + "loss": 2.0792, + "step": 4285 + }, + { + "epoch": 1.45, + "grad_norm": 0.1689453125, + "learning_rate": 4.21871355819747e-05, + "loss": 2.112, + "step": 4290 + }, + { + "epoch": 1.45, + "grad_norm": 0.1708984375, + "learning_rate": 4.19461530686434e-05, + "loss": 2.1081, + "step": 4295 + }, + { + "epoch": 1.46, + "grad_norm": 0.177734375, + "learning_rate": 4.170567800245244e-05, + "loss": 2.1193, + "step": 4300 + }, + { + "epoch": 1.46, + "grad_norm": 0.169921875, + "learning_rate": 4.1465712485387966e-05, + "loss": 2.1102, + "step": 4305 + }, + { + "epoch": 1.46, + "grad_norm": 0.169921875, + "learning_rate": 4.1226258614982214e-05, + "loss": 2.1125, + "step": 4310 + }, + { + "epoch": 1.46, + "grad_norm": 0.171875, + "learning_rate": 4.0987318484295135e-05, + "loss": 2.103, + "step": 4315 + }, + { + "epoch": 1.46, + "grad_norm": 0.1748046875, + "learning_rate": 4.074889418189608e-05, + "loss": 2.1172, + "step": 4320 + }, + { + "epoch": 1.46, + "grad_norm": 0.1640625, + "learning_rate": 4.051098779184559e-05, + "loss": 2.0825, + "step": 4325 + }, + { + "epoch": 1.47, + "grad_norm": 0.1640625, + "learning_rate": 4.0273601393677064e-05, + "loss": 2.1347, + "step": 4330 + }, + { + "epoch": 1.47, + "grad_norm": 0.171875, + "learning_rate": 4.0036737062378823e-05, + "loss": 2.1057, + "step": 4335 + }, + { + "epoch": 1.47, + "grad_norm": 0.16796875, + "learning_rate": 3.980039686837568e-05, + "loss": 2.1365, + "step": 4340 + }, + { + "epoch": 1.47, + "grad_norm": 0.171875, + "learning_rate": 3.956458287751097e-05, + "loss": 2.1176, + "step": 4345 + }, + { + "epoch": 1.47, + "grad_norm": 0.1669921875, + "learning_rate": 3.932929715102863e-05, + "loss": 2.0994, + "step": 4350 + }, + { + "epoch": 1.48, + "grad_norm": 0.1728515625, + "learning_rate": 3.9094541745554946e-05, + "loss": 2.0961, + "step": 4355 + }, + { + "epoch": 1.48, + "grad_norm": 0.1708984375, + "learning_rate": 3.8860318713080725e-05, + "loss": 2.1128, + "step": 4360 + }, + { + "epoch": 1.48, + "grad_norm": 0.16796875, + "learning_rate": 3.8626630100943196e-05, + "loss": 2.1145, + "step": 4365 + }, + { + "epoch": 1.48, + "grad_norm": 0.169921875, + "learning_rate": 3.8393477951808444e-05, + "loss": 2.1021, + "step": 4370 + }, + { + "epoch": 1.48, + "grad_norm": 0.16796875, + "learning_rate": 3.816086430365321e-05, + "loss": 2.1185, + "step": 4375 + }, + { + "epoch": 1.48, + "grad_norm": 0.1669921875, + "learning_rate": 3.7928791189747195e-05, + "loss": 2.1078, + "step": 4380 + }, + { + "epoch": 1.49, + "grad_norm": 0.16796875, + "learning_rate": 3.769726063863541e-05, + "loss": 2.1133, + "step": 4385 + }, + { + "epoch": 1.49, + "grad_norm": 0.1650390625, + "learning_rate": 3.746627467412026e-05, + "loss": 2.1013, + "step": 4390 + }, + { + "epoch": 1.49, + "grad_norm": 0.16796875, + "learning_rate": 3.723583531524394e-05, + "loss": 2.115, + "step": 4395 + }, + { + "epoch": 1.49, + "grad_norm": 0.1689453125, + "learning_rate": 3.700594457627079e-05, + "loss": 2.1374, + "step": 4400 + }, + { + "epoch": 1.49, + "grad_norm": 0.1669921875, + "learning_rate": 3.6776604466669686e-05, + "loss": 2.1082, + "step": 4405 + }, + { + "epoch": 1.49, + "grad_norm": 0.1708984375, + "learning_rate": 3.654781699109645e-05, + "loss": 2.1171, + "step": 4410 + }, + { + "epoch": 1.5, + "grad_norm": 0.16796875, + "learning_rate": 3.631958414937633e-05, + "loss": 2.1055, + "step": 4415 + }, + { + "epoch": 1.5, + "grad_norm": 0.166015625, + "learning_rate": 3.609190793648661e-05, + "loss": 2.1367, + "step": 4420 + }, + { + "epoch": 1.5, + "grad_norm": 0.171875, + "learning_rate": 3.586479034253902e-05, + "loss": 2.1161, + "step": 4425 + }, + { + "epoch": 1.5, + "grad_norm": 0.16796875, + "learning_rate": 3.563823335276244e-05, + "loss": 2.1004, + "step": 4430 + }, + { + "epoch": 1.5, + "grad_norm": 0.1689453125, + "learning_rate": 3.541223894748553e-05, + "loss": 2.0959, + "step": 4435 + }, + { + "epoch": 1.5, + "grad_norm": 0.16796875, + "learning_rate": 3.51868091021194e-05, + "loss": 2.1085, + "step": 4440 + }, + { + "epoch": 1.51, + "grad_norm": 0.1689453125, + "learning_rate": 3.496194578714036e-05, + "loss": 2.1207, + "step": 4445 + }, + { + "epoch": 1.51, + "grad_norm": 0.1689453125, + "learning_rate": 3.473765096807269e-05, + "loss": 2.1181, + "step": 4450 + }, + { + "epoch": 1.51, + "grad_norm": 0.166015625, + "learning_rate": 3.45139266054715e-05, + "loss": 2.1069, + "step": 4455 + }, + { + "epoch": 1.51, + "grad_norm": 0.16796875, + "learning_rate": 3.429077465490551e-05, + "loss": 2.1213, + "step": 4460 + }, + { + "epoch": 1.51, + "grad_norm": 0.1728515625, + "learning_rate": 3.406819706694003e-05, + "loss": 2.0987, + "step": 4465 + }, + { + "epoch": 1.51, + "grad_norm": 0.1708984375, + "learning_rate": 3.3846195787119814e-05, + "loss": 2.0996, + "step": 4470 + }, + { + "epoch": 1.52, + "grad_norm": 0.1689453125, + "learning_rate": 3.362477275595225e-05, + "loss": 2.0933, + "step": 4475 + }, + { + "epoch": 1.52, + "grad_norm": 0.171875, + "learning_rate": 3.340392990889018e-05, + "loss": 2.0977, + "step": 4480 + }, + { + "epoch": 1.52, + "grad_norm": 0.1748046875, + "learning_rate": 3.3183669176315045e-05, + "loss": 2.1336, + "step": 4485 + }, + { + "epoch": 1.52, + "grad_norm": 0.166015625, + "learning_rate": 3.296399248352012e-05, + "loss": 2.0963, + "step": 4490 + }, + { + "epoch": 1.52, + "grad_norm": 0.1689453125, + "learning_rate": 3.2744901750693556e-05, + "loss": 2.0951, + "step": 4495 + }, + { + "epoch": 1.52, + "grad_norm": 0.1728515625, + "learning_rate": 3.2526398892901654e-05, + "loss": 2.0882, + "step": 4500 + }, + { + "epoch": 1.53, + "grad_norm": 0.169921875, + "learning_rate": 3.2308485820072075e-05, + "loss": 2.1044, + "step": 4505 + }, + { + "epoch": 1.53, + "grad_norm": 0.1669921875, + "learning_rate": 3.2091164436977294e-05, + "loss": 2.1091, + "step": 4510 + }, + { + "epoch": 1.53, + "grad_norm": 0.1650390625, + "learning_rate": 3.187443664321773e-05, + "loss": 2.0628, + "step": 4515 + }, + { + "epoch": 1.53, + "grad_norm": 0.16796875, + "learning_rate": 3.165830433320531e-05, + "loss": 2.1029, + "step": 4520 + }, + { + "epoch": 1.53, + "grad_norm": 0.162109375, + "learning_rate": 3.144276939614683e-05, + "loss": 2.1297, + "step": 4525 + }, + { + "epoch": 1.53, + "grad_norm": 0.16796875, + "learning_rate": 3.122783371602747e-05, + "loss": 2.1324, + "step": 4530 + }, + { + "epoch": 1.54, + "grad_norm": 0.166015625, + "learning_rate": 3.101349917159433e-05, + "loss": 2.1366, + "step": 4535 + }, + { + "epoch": 1.54, + "grad_norm": 0.16796875, + "learning_rate": 3.079976763633996e-05, + "loss": 2.1136, + "step": 4540 + }, + { + "epoch": 1.54, + "grad_norm": 0.16796875, + "learning_rate": 3.058664097848612e-05, + "loss": 2.1107, + "step": 4545 + }, + { + "epoch": 1.54, + "grad_norm": 0.171875, + "learning_rate": 3.0374121060967255e-05, + "loss": 2.1024, + "step": 4550 + }, + { + "epoch": 1.54, + "grad_norm": 0.1669921875, + "learning_rate": 3.0162209741414304e-05, + "loss": 2.1044, + "step": 4555 + }, + { + "epoch": 1.54, + "grad_norm": 0.1708984375, + "learning_rate": 2.9950908872138584e-05, + "loss": 2.0755, + "step": 4560 + }, + { + "epoch": 1.55, + "grad_norm": 0.1669921875, + "learning_rate": 2.9740220300115386e-05, + "loss": 2.1191, + "step": 4565 + }, + { + "epoch": 1.55, + "grad_norm": 0.1708984375, + "learning_rate": 2.9530145866967895e-05, + "loss": 2.1415, + "step": 4570 + }, + { + "epoch": 1.55, + "grad_norm": 0.1650390625, + "learning_rate": 2.9320687408951162e-05, + "loss": 2.0845, + "step": 4575 + }, + { + "epoch": 1.55, + "grad_norm": 0.166015625, + "learning_rate": 2.9111846756936113e-05, + "loss": 2.118, + "step": 4580 + }, + { + "epoch": 1.55, + "grad_norm": 0.1669921875, + "learning_rate": 2.8903625736393304e-05, + "loss": 2.0881, + "step": 4585 + }, + { + "epoch": 1.55, + "grad_norm": 0.1689453125, + "learning_rate": 2.8696026167377155e-05, + "loss": 2.1497, + "step": 4590 + }, + { + "epoch": 1.56, + "grad_norm": 0.1748046875, + "learning_rate": 2.8489049864510054e-05, + "loss": 2.1046, + "step": 4595 + }, + { + "epoch": 1.56, + "grad_norm": 0.1708984375, + "learning_rate": 2.8282698636966375e-05, + "loss": 2.139, + "step": 4600 + }, + { + "epoch": 1.56, + "grad_norm": 0.169921875, + "learning_rate": 2.8076974288456726e-05, + "loss": 2.0716, + "step": 4605 + }, + { + "epoch": 1.56, + "grad_norm": 0.171875, + "learning_rate": 2.78718786172122e-05, + "loss": 2.1047, + "step": 4610 + }, + { + "epoch": 1.56, + "grad_norm": 0.169921875, + "learning_rate": 2.7667413415968635e-05, + "loss": 2.0918, + "step": 4615 + }, + { + "epoch": 1.56, + "grad_norm": 0.1689453125, + "learning_rate": 2.7463580471950943e-05, + "loss": 2.1356, + "step": 4620 + }, + { + "epoch": 1.57, + "grad_norm": 0.1650390625, + "learning_rate": 2.7260381566857473e-05, + "loss": 2.0997, + "step": 4625 + }, + { + "epoch": 1.57, + "grad_norm": 0.166015625, + "learning_rate": 2.7057818476844533e-05, + "loss": 2.0962, + "step": 4630 + }, + { + "epoch": 1.57, + "grad_norm": 0.16796875, + "learning_rate": 2.68558929725107e-05, + "loss": 2.095, + "step": 4635 + }, + { + "epoch": 1.57, + "grad_norm": 0.166015625, + "learning_rate": 2.6654606818881465e-05, + "loss": 2.1209, + "step": 4640 + }, + { + "epoch": 1.57, + "grad_norm": 0.1640625, + "learning_rate": 2.645396177539379e-05, + "loss": 2.1148, + "step": 4645 + }, + { + "epoch": 1.57, + "grad_norm": 0.1650390625, + "learning_rate": 2.6253959595880673e-05, + "loss": 2.1075, + "step": 4650 + }, + { + "epoch": 1.58, + "grad_norm": 0.1689453125, + "learning_rate": 2.6054602028555885e-05, + "loss": 2.1129, + "step": 4655 + }, + { + "epoch": 1.58, + "grad_norm": 0.1689453125, + "learning_rate": 2.585589081599862e-05, + "loss": 2.1114, + "step": 4660 + }, + { + "epoch": 1.58, + "grad_norm": 0.1689453125, + "learning_rate": 2.5657827695138372e-05, + "loss": 2.1017, + "step": 4665 + }, + { + "epoch": 1.58, + "grad_norm": 0.177734375, + "learning_rate": 2.546041439723963e-05, + "loss": 2.1269, + "step": 4670 + }, + { + "epoch": 1.58, + "grad_norm": 0.16796875, + "learning_rate": 2.5263652647886803e-05, + "loss": 2.1149, + "step": 4675 + }, + { + "epoch": 1.59, + "grad_norm": 0.16796875, + "learning_rate": 2.5067544166969114e-05, + "loss": 2.1177, + "step": 4680 + }, + { + "epoch": 1.59, + "grad_norm": 0.1708984375, + "learning_rate": 2.487209066866565e-05, + "loss": 2.0814, + "step": 4685 + }, + { + "epoch": 1.59, + "grad_norm": 0.16796875, + "learning_rate": 2.467729386143025e-05, + "loss": 2.1058, + "step": 4690 + }, + { + "epoch": 1.59, + "grad_norm": 0.1689453125, + "learning_rate": 2.4483155447976657e-05, + "loss": 2.1239, + "step": 4695 + }, + { + "epoch": 1.59, + "grad_norm": 0.1708984375, + "learning_rate": 2.42896771252636e-05, + "loss": 2.0943, + "step": 4700 + }, + { + "epoch": 1.59, + "grad_norm": 0.1669921875, + "learning_rate": 2.4096860584479974e-05, + "loss": 2.135, + "step": 4705 + }, + { + "epoch": 1.6, + "grad_norm": 0.16796875, + "learning_rate": 2.390470751103008e-05, + "loss": 2.0901, + "step": 4710 + }, + { + "epoch": 1.6, + "grad_norm": 0.1669921875, + "learning_rate": 2.37132195845188e-05, + "loss": 2.1333, + "step": 4715 + }, + { + "epoch": 1.6, + "grad_norm": 0.16796875, + "learning_rate": 2.3522398478737108e-05, + "loss": 2.0971, + "step": 4720 + }, + { + "epoch": 1.6, + "grad_norm": 0.1689453125, + "learning_rate": 2.3332245861647206e-05, + "loss": 2.0869, + "step": 4725 + }, + { + "epoch": 1.6, + "grad_norm": 0.1728515625, + "learning_rate": 2.3142763395368095e-05, + "loss": 2.0907, + "step": 4730 + }, + { + "epoch": 1.6, + "grad_norm": 0.1669921875, + "learning_rate": 2.295395273616099e-05, + "loss": 2.1023, + "step": 4735 + }, + { + "epoch": 1.61, + "grad_norm": 0.171875, + "learning_rate": 2.2765815534414868e-05, + "loss": 2.0932, + "step": 4740 + }, + { + "epoch": 1.61, + "grad_norm": 0.1630859375, + "learning_rate": 2.257835343463205e-05, + "loss": 2.0842, + "step": 4745 + }, + { + "epoch": 1.61, + "grad_norm": 0.173828125, + "learning_rate": 2.239156807541375e-05, + "loss": 2.1045, + "step": 4750 + }, + { + "epoch": 1.61, + "grad_norm": 0.173828125, + "learning_rate": 2.22054610894459e-05, + "loss": 2.1369, + "step": 4755 + }, + { + "epoch": 1.61, + "grad_norm": 0.1806640625, + "learning_rate": 2.202003410348473e-05, + "loss": 2.0935, + "step": 4760 + }, + { + "epoch": 1.61, + "grad_norm": 0.16796875, + "learning_rate": 2.1835288738342596e-05, + "loss": 2.1332, + "step": 4765 + }, + { + "epoch": 1.62, + "grad_norm": 0.169921875, + "learning_rate": 2.1651226608873877e-05, + "loss": 2.0987, + "step": 4770 + }, + { + "epoch": 1.62, + "grad_norm": 0.16796875, + "learning_rate": 2.1467849323960797e-05, + "loss": 2.136, + "step": 4775 + }, + { + "epoch": 1.62, + "grad_norm": 0.1748046875, + "learning_rate": 2.128515848649929e-05, + "loss": 2.1241, + "step": 4780 + }, + { + "epoch": 1.62, + "grad_norm": 0.16796875, + "learning_rate": 2.1103155693385136e-05, + "loss": 2.137, + "step": 4785 + }, + { + "epoch": 1.62, + "grad_norm": 0.1748046875, + "learning_rate": 2.092184253549998e-05, + "loss": 2.1204, + "step": 4790 + }, + { + "epoch": 1.62, + "grad_norm": 0.1708984375, + "learning_rate": 2.074122059769733e-05, + "loss": 2.0799, + "step": 4795 + }, + { + "epoch": 1.63, + "grad_norm": 0.16796875, + "learning_rate": 2.0561291458788733e-05, + "loss": 2.1012, + "step": 4800 + }, + { + "epoch": 1.63, + "grad_norm": 0.171875, + "learning_rate": 2.0382056691530084e-05, + "loss": 2.1381, + "step": 4805 + }, + { + "epoch": 1.63, + "grad_norm": 0.16796875, + "learning_rate": 2.02035178626077e-05, + "loss": 2.1153, + "step": 4810 + }, + { + "epoch": 1.63, + "grad_norm": 0.166015625, + "learning_rate": 2.002567653262479e-05, + "loss": 2.0965, + "step": 4815 + }, + { + "epoch": 1.63, + "grad_norm": 0.1640625, + "learning_rate": 1.984853425608769e-05, + "loss": 2.1355, + "step": 4820 + }, + { + "epoch": 1.63, + "grad_norm": 0.169921875, + "learning_rate": 1.9672092581392375e-05, + "loss": 2.1273, + "step": 4825 + }, + { + "epoch": 1.64, + "grad_norm": 0.171875, + "learning_rate": 1.9496353050810843e-05, + "loss": 2.1183, + "step": 4830 + }, + { + "epoch": 1.64, + "grad_norm": 0.1630859375, + "learning_rate": 1.9321317200477653e-05, + "loss": 2.1448, + "step": 4835 + }, + { + "epoch": 1.64, + "grad_norm": 0.16796875, + "learning_rate": 1.91469865603766e-05, + "loss": 2.0943, + "step": 4840 + }, + { + "epoch": 1.64, + "grad_norm": 0.16796875, + "learning_rate": 1.8973362654327175e-05, + "loss": 2.117, + "step": 4845 + }, + { + "epoch": 1.64, + "grad_norm": 0.166015625, + "learning_rate": 1.8800446999971346e-05, + "loss": 2.1076, + "step": 4850 + }, + { + "epoch": 1.64, + "grad_norm": 0.17578125, + "learning_rate": 1.8628241108760268e-05, + "loss": 2.0966, + "step": 4855 + }, + { + "epoch": 1.65, + "grad_norm": 0.169921875, + "learning_rate": 1.845674648594108e-05, + "loss": 2.1312, + "step": 4860 + }, + { + "epoch": 1.65, + "grad_norm": 0.1669921875, + "learning_rate": 1.828596463054375e-05, + "loss": 2.0882, + "step": 4865 + }, + { + "epoch": 1.65, + "grad_norm": 0.1689453125, + "learning_rate": 1.8115897035367934e-05, + "loss": 2.0856, + "step": 4870 + }, + { + "epoch": 1.65, + "grad_norm": 0.1689453125, + "learning_rate": 1.7946545186970022e-05, + "loss": 2.1279, + "step": 4875 + }, + { + "epoch": 1.65, + "grad_norm": 0.1708984375, + "learning_rate": 1.7777910565650024e-05, + "loss": 2.0998, + "step": 4880 + }, + { + "epoch": 1.65, + "grad_norm": 0.169921875, + "learning_rate": 1.760999464543869e-05, + "loss": 2.0973, + "step": 4885 + }, + { + "epoch": 1.66, + "grad_norm": 0.1728515625, + "learning_rate": 1.7442798894084655e-05, + "loss": 2.1295, + "step": 4890 + }, + { + "epoch": 1.66, + "grad_norm": 0.1689453125, + "learning_rate": 1.7276324773041565e-05, + "loss": 2.1326, + "step": 4895 + }, + { + "epoch": 1.66, + "grad_norm": 0.171875, + "learning_rate": 1.7110573737455295e-05, + "loss": 2.1087, + "step": 4900 + }, + { + "epoch": 1.66, + "grad_norm": 0.1669921875, + "learning_rate": 1.694554723615126e-05, + "loss": 2.1013, + "step": 4905 + }, + { + "epoch": 1.66, + "grad_norm": 0.16796875, + "learning_rate": 1.6781246711621744e-05, + "loss": 2.1268, + "step": 4910 + }, + { + "epoch": 1.66, + "grad_norm": 0.1650390625, + "learning_rate": 1.6617673600013296e-05, + "loss": 2.1262, + "step": 4915 + }, + { + "epoch": 1.67, + "grad_norm": 0.16796875, + "learning_rate": 1.645482933111412e-05, + "loss": 2.0923, + "step": 4920 + }, + { + "epoch": 1.67, + "grad_norm": 0.16796875, + "learning_rate": 1.6292715328341712e-05, + "loss": 2.0922, + "step": 4925 + }, + { + "epoch": 1.67, + "grad_norm": 0.1728515625, + "learning_rate": 1.6131333008730277e-05, + "loss": 2.1355, + "step": 4930 + }, + { + "epoch": 1.67, + "grad_norm": 0.1728515625, + "learning_rate": 1.5970683782918374e-05, + "loss": 2.108, + "step": 4935 + }, + { + "epoch": 1.67, + "grad_norm": 0.16796875, + "learning_rate": 1.5810769055136644e-05, + "loss": 2.1115, + "step": 4940 + }, + { + "epoch": 1.67, + "grad_norm": 0.171875, + "learning_rate": 1.56515902231955e-05, + "loss": 2.1024, + "step": 4945 + }, + { + "epoch": 1.68, + "grad_norm": 0.169921875, + "learning_rate": 1.5493148678472903e-05, + "loss": 2.1268, + "step": 4950 + }, + { + "epoch": 1.68, + "grad_norm": 0.171875, + "learning_rate": 1.533544580590217e-05, + "loss": 2.1084, + "step": 4955 + }, + { + "epoch": 1.68, + "grad_norm": 0.1728515625, + "learning_rate": 1.5178482983959985e-05, + "loss": 2.0965, + "step": 4960 + }, + { + "epoch": 1.68, + "grad_norm": 0.169921875, + "learning_rate": 1.5022261584654207e-05, + "loss": 2.1289, + "step": 4965 + }, + { + "epoch": 1.68, + "grad_norm": 0.1669921875, + "learning_rate": 1.4866782973511962e-05, + "loss": 2.0724, + "step": 4970 + }, + { + "epoch": 1.69, + "grad_norm": 0.16796875, + "learning_rate": 1.4712048509567634e-05, + "loss": 2.1429, + "step": 4975 + }, + { + "epoch": 1.69, + "grad_norm": 0.16796875, + "learning_rate": 1.4558059545351143e-05, + "loss": 2.1328, + "step": 4980 + }, + { + "epoch": 1.69, + "grad_norm": 0.16796875, + "learning_rate": 1.4404817426875938e-05, + "loss": 2.0846, + "step": 4985 + }, + { + "epoch": 1.69, + "grad_norm": 0.1689453125, + "learning_rate": 1.4252323493627251e-05, + "loss": 2.1035, + "step": 4990 + }, + { + "epoch": 1.69, + "grad_norm": 0.169921875, + "learning_rate": 1.4100579078550613e-05, + "loss": 2.0901, + "step": 4995 + }, + { + "epoch": 1.69, + "grad_norm": 0.1748046875, + "learning_rate": 1.3949585508039886e-05, + "loss": 2.1229, + "step": 5000 + }, + { + "epoch": 1.7, + "grad_norm": 0.1669921875, + "learning_rate": 1.3799344101925904e-05, + "loss": 2.1345, + "step": 5005 + }, + { + "epoch": 1.7, + "grad_norm": 0.177734375, + "learning_rate": 1.3649856173464781e-05, + "loss": 2.1067, + "step": 5010 + }, + { + "epoch": 1.7, + "grad_norm": 0.1669921875, + "learning_rate": 1.3501123029326601e-05, + "loss": 2.1274, + "step": 5015 + }, + { + "epoch": 1.7, + "grad_norm": 0.169921875, + "learning_rate": 1.3353145969583813e-05, + "loss": 2.1117, + "step": 5020 + }, + { + "epoch": 1.7, + "grad_norm": 0.1669921875, + "learning_rate": 1.3205926287699988e-05, + "loss": 2.1139, + "step": 5025 + }, + { + "epoch": 1.7, + "grad_norm": 0.1728515625, + "learning_rate": 1.3059465270518468e-05, + "loss": 2.1131, + "step": 5030 + }, + { + "epoch": 1.71, + "grad_norm": 0.16796875, + "learning_rate": 1.2913764198251132e-05, + "loss": 2.1569, + "step": 5035 + }, + { + "epoch": 1.71, + "grad_norm": 0.169921875, + "learning_rate": 1.276882434446719e-05, + "loss": 2.1104, + "step": 5040 + }, + { + "epoch": 1.71, + "grad_norm": 0.169921875, + "learning_rate": 1.2624646976082066e-05, + "loss": 2.1107, + "step": 5045 + }, + { + "epoch": 1.71, + "grad_norm": 0.1689453125, + "learning_rate": 1.2481233353346344e-05, + "loss": 2.1026, + "step": 5050 + }, + { + "epoch": 1.71, + "grad_norm": 0.1689453125, + "learning_rate": 1.2338584729834701e-05, + "loss": 2.0853, + "step": 5055 + }, + { + "epoch": 1.71, + "grad_norm": 0.16796875, + "learning_rate": 1.2196702352434953e-05, + "loss": 2.1188, + "step": 5060 + }, + { + "epoch": 1.72, + "grad_norm": 0.169921875, + "learning_rate": 1.205558746133727e-05, + "loss": 2.1077, + "step": 5065 + }, + { + "epoch": 1.72, + "grad_norm": 0.16796875, + "learning_rate": 1.1915241290023115e-05, + "loss": 2.097, + "step": 5070 + }, + { + "epoch": 1.72, + "grad_norm": 0.1728515625, + "learning_rate": 1.1775665065254704e-05, + "loss": 2.1289, + "step": 5075 + }, + { + "epoch": 1.72, + "grad_norm": 0.169921875, + "learning_rate": 1.1636860007064076e-05, + "loss": 2.082, + "step": 5080 + }, + { + "epoch": 1.72, + "grad_norm": 0.1708984375, + "learning_rate": 1.1498827328742623e-05, + "loss": 2.108, + "step": 5085 + }, + { + "epoch": 1.72, + "grad_norm": 0.1689453125, + "learning_rate": 1.1361568236830323e-05, + "loss": 2.1289, + "step": 5090 + }, + { + "epoch": 1.73, + "grad_norm": 0.166015625, + "learning_rate": 1.122508393110524e-05, + "loss": 2.1245, + "step": 5095 + }, + { + "epoch": 1.73, + "grad_norm": 0.16796875, + "learning_rate": 1.1089375604573116e-05, + "loss": 2.0823, + "step": 5100 + }, + { + "epoch": 1.73, + "grad_norm": 0.1708984375, + "learning_rate": 1.0954444443456824e-05, + "loss": 2.0788, + "step": 5105 + }, + { + "epoch": 1.73, + "grad_norm": 0.169921875, + "learning_rate": 1.0820291627186107e-05, + "loss": 2.1059, + "step": 5110 + }, + { + "epoch": 1.73, + "grad_norm": 0.1640625, + "learning_rate": 1.0686918328387118e-05, + "loss": 2.1058, + "step": 5115 + }, + { + "epoch": 1.73, + "grad_norm": 0.16796875, + "learning_rate": 1.0554325712872381e-05, + "loss": 2.1066, + "step": 5120 + }, + { + "epoch": 1.74, + "grad_norm": 0.1728515625, + "learning_rate": 1.0422514939630424e-05, + "loss": 2.1088, + "step": 5125 + }, + { + "epoch": 1.74, + "grad_norm": 0.1708984375, + "learning_rate": 1.0291487160815726e-05, + "loss": 2.1371, + "step": 5130 + }, + { + "epoch": 1.74, + "grad_norm": 0.1669921875, + "learning_rate": 1.0161243521738661e-05, + "loss": 2.1256, + "step": 5135 + }, + { + "epoch": 1.74, + "grad_norm": 0.1650390625, + "learning_rate": 1.003178516085541e-05, + "loss": 2.1213, + "step": 5140 + }, + { + "epoch": 1.74, + "grad_norm": 0.169921875, + "learning_rate": 9.903113209758096e-06, + "loss": 2.1239, + "step": 5145 + }, + { + "epoch": 1.74, + "grad_norm": 0.16796875, + "learning_rate": 9.775228793164826e-06, + "loss": 2.0883, + "step": 5150 + }, + { + "epoch": 1.75, + "grad_norm": 0.1689453125, + "learning_rate": 9.6481330289099e-06, + "loss": 2.0986, + "step": 5155 + }, + { + "epoch": 1.75, + "grad_norm": 0.1669921875, + "learning_rate": 9.521827027934038e-06, + "loss": 2.1189, + "step": 5160 + }, + { + "epoch": 1.75, + "grad_norm": 0.1689453125, + "learning_rate": 9.396311894274645e-06, + "loss": 2.1267, + "step": 5165 + }, + { + "epoch": 1.75, + "grad_norm": 0.1669921875, + "learning_rate": 9.271588725056201e-06, + "loss": 2.1026, + "step": 5170 + }, + { + "epoch": 1.75, + "grad_norm": 0.169921875, + "learning_rate": 9.147658610480625e-06, + "loss": 2.097, + "step": 5175 + }, + { + "epoch": 1.75, + "grad_norm": 0.1689453125, + "learning_rate": 9.024522633817756e-06, + "loss": 2.1243, + "step": 5180 + }, + { + "epoch": 1.76, + "grad_norm": 0.169921875, + "learning_rate": 8.90218187139591e-06, + "loss": 2.1412, + "step": 5185 + }, + { + "epoch": 1.76, + "grad_norm": 0.1689453125, + "learning_rate": 8.780637392592495e-06, + "loss": 2.1046, + "step": 5190 + }, + { + "epoch": 1.76, + "grad_norm": 0.16796875, + "learning_rate": 8.659890259824532e-06, + "loss": 2.1173, + "step": 5195 + }, + { + "epoch": 1.76, + "grad_norm": 0.1640625, + "learning_rate": 8.53994152853952e-06, + "loss": 2.1192, + "step": 5200 + }, + { + "epoch": 1.76, + "grad_norm": 0.1689453125, + "learning_rate": 8.420792247206177e-06, + "loss": 2.1264, + "step": 5205 + }, + { + "epoch": 1.76, + "grad_norm": 0.16796875, + "learning_rate": 8.302443457305209e-06, + "loss": 2.0895, + "step": 5210 + }, + { + "epoch": 1.77, + "grad_norm": 0.1708984375, + "learning_rate": 8.184896193320246e-06, + "loss": 2.1079, + "step": 5215 + }, + { + "epoch": 1.77, + "grad_norm": 0.1689453125, + "learning_rate": 8.068151482728802e-06, + "loss": 2.0993, + "step": 5220 + }, + { + "epoch": 1.77, + "grad_norm": 0.16796875, + "learning_rate": 7.952210345993339e-06, + "loss": 2.1263, + "step": 5225 + }, + { + "epoch": 1.77, + "grad_norm": 0.16796875, + "learning_rate": 7.837073796552241e-06, + "loss": 2.0958, + "step": 5230 + }, + { + "epoch": 1.77, + "grad_norm": 0.169921875, + "learning_rate": 7.72274284081106e-06, + "loss": 2.1071, + "step": 5235 + }, + { + "epoch": 1.77, + "grad_norm": 0.169921875, + "learning_rate": 7.609218478133628e-06, + "loss": 2.1182, + "step": 5240 + }, + { + "epoch": 1.78, + "grad_norm": 0.1669921875, + "learning_rate": 7.4965017008334245e-06, + "loss": 2.0934, + "step": 5245 + }, + { + "epoch": 1.78, + "grad_norm": 0.169921875, + "learning_rate": 7.3845934941648046e-06, + "loss": 2.1125, + "step": 5250 + }, + { + "epoch": 1.78, + "grad_norm": 0.16796875, + "learning_rate": 7.2734948363144206e-06, + "loss": 2.0973, + "step": 5255 + }, + { + "epoch": 1.78, + "grad_norm": 0.16796875, + "learning_rate": 7.163206698392744e-06, + "loss": 2.1164, + "step": 5260 + }, + { + "epoch": 1.78, + "grad_norm": 0.1650390625, + "learning_rate": 7.0537300444254435e-06, + "loss": 2.1232, + "step": 5265 + }, + { + "epoch": 1.78, + "grad_norm": 0.169921875, + "learning_rate": 6.945065831345077e-06, + "loss": 2.1179, + "step": 5270 + }, + { + "epoch": 1.79, + "grad_norm": 0.1728515625, + "learning_rate": 6.837215008982633e-06, + "loss": 2.1167, + "step": 5275 + }, + { + "epoch": 1.79, + "grad_norm": 0.169921875, + "learning_rate": 6.7301785200593046e-06, + "loss": 2.1473, + "step": 5280 + }, + { + "epoch": 1.79, + "grad_norm": 0.1748046875, + "learning_rate": 6.623957300178207e-06, + "loss": 2.0955, + "step": 5285 + }, + { + "epoch": 1.79, + "grad_norm": 0.16796875, + "learning_rate": 6.518552277816215e-06, + "loss": 2.128, + "step": 5290 + }, + { + "epoch": 1.79, + "grad_norm": 0.169921875, + "learning_rate": 6.413964374315851e-06, + "loss": 2.1291, + "step": 5295 + }, + { + "epoch": 1.8, + "grad_norm": 0.16796875, + "learning_rate": 6.31019450387721e-06, + "loss": 2.103, + "step": 5300 + }, + { + "epoch": 1.8, + "grad_norm": 0.17578125, + "learning_rate": 6.207243573549959e-06, + "loss": 2.1046, + "step": 5305 + }, + { + "epoch": 1.8, + "grad_norm": 0.1689453125, + "learning_rate": 6.1051124832254944e-06, + "loss": 2.0974, + "step": 5310 + }, + { + "epoch": 1.8, + "grad_norm": 0.1689453125, + "learning_rate": 6.003802125628966e-06, + "loss": 2.1328, + "step": 5315 + }, + { + "epoch": 1.8, + "grad_norm": 0.1728515625, + "learning_rate": 5.903313386311527e-06, + "loss": 2.0939, + "step": 5320 + }, + { + "epoch": 1.8, + "grad_norm": 0.1689453125, + "learning_rate": 5.803647143642554e-06, + "loss": 2.1194, + "step": 5325 + }, + { + "epoch": 1.81, + "grad_norm": 0.169921875, + "learning_rate": 5.704804268802077e-06, + "loss": 2.1237, + "step": 5330 + }, + { + "epoch": 1.81, + "grad_norm": 0.1748046875, + "learning_rate": 5.606785625773048e-06, + "loss": 2.1483, + "step": 5335 + }, + { + "epoch": 1.81, + "grad_norm": 0.17578125, + "learning_rate": 5.5095920713338134e-06, + "loss": 2.14, + "step": 5340 + }, + { + "epoch": 1.81, + "grad_norm": 0.17578125, + "learning_rate": 5.413224455050692e-06, + "loss": 2.1279, + "step": 5345 + }, + { + "epoch": 1.81, + "grad_norm": 0.16796875, + "learning_rate": 5.3176836192704414e-06, + "loss": 2.1356, + "step": 5350 + }, + { + "epoch": 1.81, + "grad_norm": 0.16796875, + "learning_rate": 5.222970399112981e-06, + "loss": 2.1192, + "step": 5355 + }, + { + "epoch": 1.82, + "grad_norm": 0.16796875, + "learning_rate": 5.12908562246408e-06, + "loss": 2.1321, + "step": 5360 + }, + { + "epoch": 1.82, + "grad_norm": 0.171875, + "learning_rate": 5.036030109968082e-06, + "loss": 2.1285, + "step": 5365 + }, + { + "epoch": 1.82, + "grad_norm": 0.1650390625, + "learning_rate": 4.9438046750207465e-06, + "loss": 2.117, + "step": 5370 + }, + { + "epoch": 1.82, + "grad_norm": 0.16796875, + "learning_rate": 4.8524101237621635e-06, + "loss": 2.1062, + "step": 5375 + }, + { + "epoch": 1.82, + "grad_norm": 0.1689453125, + "learning_rate": 4.7618472550696954e-06, + "loss": 2.1149, + "step": 5380 + }, + { + "epoch": 1.82, + "grad_norm": 0.1728515625, + "learning_rate": 4.67211686055099e-06, + "loss": 2.1403, + "step": 5385 + }, + { + "epoch": 1.83, + "grad_norm": 0.1728515625, + "learning_rate": 4.583219724537046e-06, + "loss": 2.1146, + "step": 5390 + }, + { + "epoch": 1.83, + "grad_norm": 0.16796875, + "learning_rate": 4.495156624075347e-06, + "loss": 2.1065, + "step": 5395 + }, + { + "epoch": 1.83, + "grad_norm": 0.169921875, + "learning_rate": 4.407928328923194e-06, + "loss": 2.1347, + "step": 5400 + }, + { + "epoch": 1.83, + "grad_norm": 0.1728515625, + "learning_rate": 4.321535601540738e-06, + "loss": 2.1006, + "step": 5405 + }, + { + "epoch": 1.83, + "grad_norm": 0.16796875, + "learning_rate": 4.2359791970845496e-06, + "loss": 2.0962, + "step": 5410 + }, + { + "epoch": 1.83, + "grad_norm": 0.1708984375, + "learning_rate": 4.1512598634009025e-06, + "loss": 2.0732, + "step": 5415 + }, + { + "epoch": 1.84, + "grad_norm": 0.1689453125, + "learning_rate": 4.067378341019257e-06, + "loss": 2.1268, + "step": 5420 + }, + { + "epoch": 1.84, + "grad_norm": 0.1748046875, + "learning_rate": 3.984335363145752e-06, + "loss": 2.1028, + "step": 5425 + }, + { + "epoch": 1.84, + "grad_norm": 0.1708984375, + "learning_rate": 3.902131655656893e-06, + "loss": 2.0817, + "step": 5430 + }, + { + "epoch": 1.84, + "grad_norm": 0.16796875, + "learning_rate": 3.820767937093095e-06, + "loss": 2.1152, + "step": 5435 + }, + { + "epoch": 1.84, + "grad_norm": 0.166015625, + "learning_rate": 3.740244918652469e-06, + "loss": 2.1147, + "step": 5440 + }, + { + "epoch": 1.84, + "grad_norm": 0.169921875, + "learning_rate": 3.6605633041846053e-06, + "loss": 2.0832, + "step": 5445 + }, + { + "epoch": 1.85, + "grad_norm": 0.16796875, + "learning_rate": 3.581723790184377e-06, + "loss": 2.1154, + "step": 5450 + }, + { + "epoch": 1.85, + "grad_norm": 0.1650390625, + "learning_rate": 3.503727065785878e-06, + "loss": 2.1344, + "step": 5455 + }, + { + "epoch": 1.85, + "grad_norm": 0.173828125, + "learning_rate": 3.4265738127564286e-06, + "loss": 2.1045, + "step": 5460 + }, + { + "epoch": 1.85, + "grad_norm": 0.171875, + "learning_rate": 3.350264705490569e-06, + "loss": 2.1143, + "step": 5465 + }, + { + "epoch": 1.85, + "grad_norm": 0.1708984375, + "learning_rate": 3.2748004110041863e-06, + "loss": 2.1074, + "step": 5470 + }, + { + "epoch": 1.85, + "grad_norm": 0.1669921875, + "learning_rate": 3.2001815889286856e-06, + "loss": 2.0837, + "step": 5475 + }, + { + "epoch": 1.86, + "grad_norm": 0.1689453125, + "learning_rate": 3.126408891505217e-06, + "loss": 2.0874, + "step": 5480 + }, + { + "epoch": 1.86, + "grad_norm": 0.16796875, + "learning_rate": 3.0534829635789684e-06, + "loss": 2.1232, + "step": 5485 + }, + { + "epoch": 1.86, + "grad_norm": 0.1669921875, + "learning_rate": 2.9814044425935606e-06, + "loss": 2.1074, + "step": 5490 + }, + { + "epoch": 1.86, + "grad_norm": 0.1689453125, + "learning_rate": 2.910173958585416e-06, + "loss": 2.082, + "step": 5495 + }, + { + "epoch": 1.86, + "grad_norm": 0.169921875, + "learning_rate": 2.8397921341783317e-06, + "loss": 2.0771, + "step": 5500 + }, + { + "epoch": 1.86, + "grad_norm": 0.1669921875, + "learning_rate": 2.770259584577972e-06, + "loss": 2.1078, + "step": 5505 + }, + { + "epoch": 1.87, + "grad_norm": 0.1650390625, + "learning_rate": 2.7015769175665063e-06, + "loss": 2.0803, + "step": 5510 + }, + { + "epoch": 1.87, + "grad_norm": 0.169921875, + "learning_rate": 2.633744733497312e-06, + "loss": 2.1227, + "step": 5515 + }, + { + "epoch": 1.87, + "grad_norm": 0.1767578125, + "learning_rate": 2.5667636252897143e-06, + "loss": 2.1089, + "step": 5520 + }, + { + "epoch": 1.87, + "grad_norm": 0.1630859375, + "learning_rate": 2.5006341784238107e-06, + "loss": 2.0958, + "step": 5525 + }, + { + "epoch": 1.87, + "grad_norm": 0.171875, + "learning_rate": 2.435356970935354e-06, + "loss": 2.1027, + "step": 5530 + }, + { + "epoch": 1.87, + "grad_norm": 0.171875, + "learning_rate": 2.370932573410667e-06, + "loss": 2.1082, + "step": 5535 + }, + { + "epoch": 1.88, + "grad_norm": 0.171875, + "learning_rate": 2.3073615489817235e-06, + "loss": 2.1316, + "step": 5540 + }, + { + "epoch": 1.88, + "grad_norm": 0.16796875, + "learning_rate": 2.244644453321165e-06, + "loss": 2.0948, + "step": 5545 + }, + { + "epoch": 1.88, + "grad_norm": 0.1728515625, + "learning_rate": 2.1827818346374482e-06, + "loss": 2.1408, + "step": 5550 + }, + { + "epoch": 1.88, + "grad_norm": 0.169921875, + "learning_rate": 2.121774233670104e-06, + "loss": 2.1094, + "step": 5555 + }, + { + "epoch": 1.88, + "grad_norm": 0.16796875, + "learning_rate": 2.0616221836849638e-06, + "loss": 2.117, + "step": 5560 + }, + { + "epoch": 1.88, + "grad_norm": 0.1708984375, + "learning_rate": 2.0023262104694852e-06, + "loss": 2.1125, + "step": 5565 + }, + { + "epoch": 1.89, + "grad_norm": 0.1708984375, + "learning_rate": 1.9438868323282124e-06, + "loss": 2.1178, + "step": 5570 + }, + { + "epoch": 1.89, + "grad_norm": 0.1689453125, + "learning_rate": 1.8863045600782003e-06, + "loss": 2.1206, + "step": 5575 + }, + { + "epoch": 1.89, + "grad_norm": 0.1748046875, + "learning_rate": 1.8295798970445754e-06, + "loss": 2.1231, + "step": 5580 + }, + { + "epoch": 1.89, + "grad_norm": 0.1728515625, + "learning_rate": 1.7737133390561046e-06, + "loss": 2.1121, + "step": 5585 + }, + { + "epoch": 1.89, + "grad_norm": 0.16796875, + "learning_rate": 1.7187053744409098e-06, + "loss": 2.0981, + "step": 5590 + }, + { + "epoch": 1.9, + "grad_norm": 0.1708984375, + "learning_rate": 1.6645564840221396e-06, + "loss": 2.1042, + "step": 5595 + }, + { + "epoch": 1.9, + "grad_norm": 0.1669921875, + "learning_rate": 1.6112671411138036e-06, + "loss": 2.0861, + "step": 5600 + }, + { + "epoch": 1.9, + "grad_norm": 0.1669921875, + "learning_rate": 1.5588378115166669e-06, + "loss": 2.0713, + "step": 5605 + }, + { + "epoch": 1.9, + "grad_norm": 0.16796875, + "learning_rate": 1.5072689535141072e-06, + "loss": 2.1116, + "step": 5610 + }, + { + "epoch": 1.9, + "grad_norm": 0.16796875, + "learning_rate": 1.4565610178681299e-06, + "loss": 2.1319, + "step": 5615 + }, + { + "epoch": 1.9, + "grad_norm": 0.169921875, + "learning_rate": 1.4067144478154604e-06, + "loss": 2.1159, + "step": 5620 + }, + { + "epoch": 1.91, + "grad_norm": 0.1689453125, + "learning_rate": 1.3577296790636684e-06, + "loss": 2.1301, + "step": 5625 + }, + { + "epoch": 1.91, + "grad_norm": 0.1708984375, + "learning_rate": 1.3096071397873056e-06, + "loss": 2.0978, + "step": 5630 + }, + { + "epoch": 1.91, + "grad_norm": 0.166015625, + "learning_rate": 1.2623472506242184e-06, + "loss": 2.1322, + "step": 5635 + }, + { + "epoch": 1.91, + "grad_norm": 0.171875, + "learning_rate": 1.2159504246718522e-06, + "loss": 2.1104, + "step": 5640 + }, + { + "epoch": 1.91, + "grad_norm": 0.16796875, + "learning_rate": 1.1704170674836313e-06, + "loss": 2.1059, + "step": 5645 + }, + { + "epoch": 1.91, + "grad_norm": 0.1630859375, + "learning_rate": 1.125747577065428e-06, + "loss": 2.1137, + "step": 5650 + }, + { + "epoch": 1.92, + "grad_norm": 0.169921875, + "learning_rate": 1.0819423438720665e-06, + "loss": 2.1135, + "step": 5655 + }, + { + "epoch": 1.92, + "grad_norm": 0.1748046875, + "learning_rate": 1.0390017508039473e-06, + "loss": 2.1428, + "step": 5660 + }, + { + "epoch": 1.92, + "grad_norm": 0.17578125, + "learning_rate": 9.969261732036605e-07, + "loss": 2.1313, + "step": 5665 + }, + { + "epoch": 1.92, + "grad_norm": 0.1650390625, + "learning_rate": 9.557159788526892e-07, + "loss": 2.115, + "step": 5670 + }, + { + "epoch": 1.92, + "grad_norm": 0.1708984375, + "learning_rate": 9.153715279682784e-07, + "loss": 2.1337, + "step": 5675 + }, + { + "epoch": 1.92, + "grad_norm": 0.1689453125, + "learning_rate": 8.75893173200204e-07, + "loss": 2.107, + "step": 5680 + }, + { + "epoch": 1.93, + "grad_norm": 0.169921875, + "learning_rate": 8.372812596277091e-07, + "loss": 2.1094, + "step": 5685 + }, + { + "epoch": 1.93, + "grad_norm": 0.1689453125, + "learning_rate": 7.99536124756517e-07, + "loss": 2.093, + "step": 5690 + }, + { + "epoch": 1.93, + "grad_norm": 0.171875, + "learning_rate": 7.62658098515856e-07, + "loss": 2.0956, + "step": 5695 + }, + { + "epoch": 1.93, + "grad_norm": 0.1689453125, + "learning_rate": 7.266475032555619e-07, + "loss": 2.12, + "step": 5700 + }, + { + "epoch": 1.93, + "grad_norm": 0.1708984375, + "learning_rate": 6.915046537433023e-07, + "loss": 2.1025, + "step": 5705 + }, + { + "epoch": 1.93, + "grad_norm": 0.1669921875, + "learning_rate": 6.572298571618118e-07, + "loss": 2.1076, + "step": 5710 + }, + { + "epoch": 1.94, + "grad_norm": 0.1689453125, + "learning_rate": 6.238234131061616e-07, + "loss": 2.0958, + "step": 5715 + }, + { + "epoch": 1.94, + "grad_norm": 0.1689453125, + "learning_rate": 5.912856135812051e-07, + "loss": 2.102, + "step": 5720 + }, + { + "epoch": 1.94, + "grad_norm": 0.16796875, + "learning_rate": 5.596167429989807e-07, + "loss": 2.1054, + "step": 5725 + }, + { + "epoch": 1.94, + "grad_norm": 0.169921875, + "learning_rate": 5.288170781762469e-07, + "loss": 2.1115, + "step": 5730 + }, + { + "epoch": 1.94, + "grad_norm": 0.1689453125, + "learning_rate": 4.98886888332073e-07, + "loss": 2.1091, + "step": 5735 + }, + { + "epoch": 1.94, + "grad_norm": 0.16796875, + "learning_rate": 4.698264350854409e-07, + "loss": 2.1136, + "step": 5740 + }, + { + "epoch": 1.95, + "grad_norm": 0.1767578125, + "learning_rate": 4.416359724530139e-07, + "loss": 2.1142, + "step": 5745 + }, + { + "epoch": 1.95, + "grad_norm": 0.1708984375, + "learning_rate": 4.143157468468717e-07, + "loss": 2.1154, + "step": 5750 + }, + { + "epoch": 1.95, + "grad_norm": 0.1689453125, + "learning_rate": 3.878659970724008e-07, + "loss": 2.1069, + "step": 5755 + }, + { + "epoch": 1.95, + "grad_norm": 0.16796875, + "learning_rate": 3.622869543261298e-07, + "loss": 2.0822, + "step": 5760 + }, + { + "epoch": 1.95, + "grad_norm": 0.1748046875, + "learning_rate": 3.3757884219383085e-07, + "loss": 2.1078, + "step": 5765 + }, + { + "epoch": 1.95, + "grad_norm": 0.1728515625, + "learning_rate": 3.1374187664844346e-07, + "loss": 2.115, + "step": 5770 + }, + { + "epoch": 1.96, + "grad_norm": 0.16796875, + "learning_rate": 2.907762660482538e-07, + "loss": 2.1022, + "step": 5775 + }, + { + "epoch": 1.96, + "grad_norm": 0.171875, + "learning_rate": 2.6868221113505175e-07, + "loss": 2.1222, + "step": 5780 + }, + { + "epoch": 1.96, + "grad_norm": 0.166015625, + "learning_rate": 2.474599050323989e-07, + "loss": 2.1228, + "step": 5785 + }, + { + "epoch": 1.96, + "grad_norm": 0.169921875, + "learning_rate": 2.271095332438966e-07, + "loss": 2.0893, + "step": 5790 + }, + { + "epoch": 1.96, + "grad_norm": 0.166015625, + "learning_rate": 2.076312736516206e-07, + "loss": 2.1281, + "step": 5795 + }, + { + "epoch": 1.96, + "grad_norm": 0.16796875, + "learning_rate": 1.890252965145112e-07, + "loss": 2.0929, + "step": 5800 + }, + { + "epoch": 1.97, + "grad_norm": 0.166015625, + "learning_rate": 1.7129176446692984e-07, + "loss": 2.0959, + "step": 5805 + }, + { + "epoch": 1.97, + "grad_norm": 0.1689453125, + "learning_rate": 1.5443083251720503e-07, + "loss": 2.133, + "step": 5810 + }, + { + "epoch": 1.97, + "grad_norm": 0.1689453125, + "learning_rate": 1.384426480462997e-07, + "loss": 2.0971, + "step": 5815 + }, + { + "epoch": 1.97, + "grad_norm": 0.16796875, + "learning_rate": 1.2332735080651248e-07, + "loss": 2.0809, + "step": 5820 + }, + { + "epoch": 1.97, + "grad_norm": 0.171875, + "learning_rate": 1.0908507292026748e-07, + "loss": 2.133, + "step": 5825 + }, + { + "epoch": 1.97, + "grad_norm": 0.1650390625, + "learning_rate": 9.571593887891528e-08, + "loss": 2.1076, + "step": 5830 + }, + { + "epoch": 1.98, + "grad_norm": 0.171875, + "learning_rate": 8.322006554171146e-08, + "loss": 2.1422, + "step": 5835 + }, + { + "epoch": 1.98, + "grad_norm": 0.1669921875, + "learning_rate": 7.159756213476199e-08, + "loss": 2.0922, + "step": 5840 + }, + { + "epoch": 1.98, + "grad_norm": 0.1640625, + "learning_rate": 6.084853025005721e-08, + "loss": 2.1117, + "step": 5845 + }, + { + "epoch": 1.98, + "grad_norm": 0.169921875, + "learning_rate": 5.0973063844605986e-08, + "loss": 2.0981, + "step": 5850 + }, + { + "epoch": 1.98, + "grad_norm": 0.1708984375, + "learning_rate": 4.1971249239591834e-08, + "loss": 2.0941, + "step": 5855 + }, + { + "epoch": 1.98, + "grad_norm": 0.1640625, + "learning_rate": 3.384316511964025e-08, + "loss": 2.1075, + "step": 5860 + }, + { + "epoch": 1.99, + "grad_norm": 0.1689453125, + "learning_rate": 2.658888253211922e-08, + "loss": 2.107, + "step": 5865 + }, + { + "epoch": 1.99, + "grad_norm": 0.171875, + "learning_rate": 2.0208464886517508e-08, + "loss": 2.105, + "step": 5870 + }, + { + "epoch": 1.99, + "grad_norm": 0.16796875, + "learning_rate": 1.4701967953911766e-08, + "loss": 2.1027, + "step": 5875 + }, + { + "epoch": 1.99, + "grad_norm": 0.1689453125, + "learning_rate": 1.0069439866422503e-08, + "loss": 2.1107, + "step": 5880 + }, + { + "epoch": 1.99, + "grad_norm": 0.171875, + "learning_rate": 6.3109211168699275e-09, + "loss": 2.1134, + "step": 5885 + }, + { + "epoch": 1.99, + "grad_norm": 0.1630859375, + "learning_rate": 3.4264445583631622e-09, + "loss": 2.1139, + "step": 5890 + }, + { + "epoch": 2.0, + "grad_norm": 0.169921875, + "learning_rate": 1.4160354040448908e-09, + "loss": 2.0934, + "step": 5895 + }, + { + "epoch": 2.0, + "grad_norm": 0.169921875, + "learning_rate": 2.7971122683601023e-10, + "loss": 2.1251, + "step": 5900 + }, + { + "epoch": 2.0, + "eval_loss": 2.12707781791687, + "eval_runtime": 173.3977, + "eval_samples_per_second": 7.659, + "eval_steps_per_second": 0.957, + "step": 5904 + }, + { + "epoch": 2.0, + "step": 5904, + "total_flos": 6.077264558911652e+17, + "train_loss": 2.1395558682560596, + "train_runtime": 23055.2544, + "train_samples_per_second": 2.049, + "train_steps_per_second": 0.256 } ], "logging_steps": 5, - "max_steps": 2952, + "max_steps": 5904, "num_input_tokens_seen": 0, - "num_train_epochs": 1, + "num_train_epochs": 2, "save_steps": 100, - "total_flos": 3.038696615046021e+17, + "total_flos": 6.077264558911652e+17, "train_batch_size": 4, "trial_name": null, "trial_params": null