{ "best_metric": 0.024681543931365013, "best_model_checkpoint": "./mnist_outputs/checkpoint-31875", "epoch": 5.0, "eval_steps": 500, "global_step": 31875, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 1.999372549019608e-05, "loss": 2.2789, "step": 10 }, { "epoch": 0.0, "learning_rate": 1.9987450980392157e-05, "loss": 2.2295, "step": 20 }, { "epoch": 0.0, "learning_rate": 1.9981176470588238e-05, "loss": 2.1727, "step": 30 }, { "epoch": 0.01, "learning_rate": 1.9974901960784316e-05, "loss": 2.1001, "step": 40 }, { "epoch": 0.01, "learning_rate": 1.9968627450980393e-05, "loss": 2.1057, "step": 50 }, { "epoch": 0.01, "learning_rate": 1.9962352941176474e-05, "loss": 2.0562, "step": 60 }, { "epoch": 0.01, "learning_rate": 1.995607843137255e-05, "loss": 1.9778, "step": 70 }, { "epoch": 0.01, "learning_rate": 1.994980392156863e-05, "loss": 1.892, "step": 80 }, { "epoch": 0.01, "learning_rate": 1.9943529411764707e-05, "loss": 1.9218, "step": 90 }, { "epoch": 0.02, "learning_rate": 1.9937254901960785e-05, "loss": 1.7873, "step": 100 }, { "epoch": 0.02, "learning_rate": 1.9930980392156866e-05, "loss": 1.8123, "step": 110 }, { "epoch": 0.02, "learning_rate": 1.9924705882352943e-05, "loss": 1.6605, "step": 120 }, { "epoch": 0.02, "learning_rate": 1.991843137254902e-05, "loss": 1.612, "step": 130 }, { "epoch": 0.02, "learning_rate": 1.99121568627451e-05, "loss": 1.7091, "step": 140 }, { "epoch": 0.02, "learning_rate": 1.9905882352941176e-05, "loss": 1.5283, "step": 150 }, { "epoch": 0.03, "learning_rate": 1.9899607843137257e-05, "loss": 1.5758, "step": 160 }, { "epoch": 0.03, "learning_rate": 1.9893333333333335e-05, "loss": 1.5249, "step": 170 }, { "epoch": 0.03, "learning_rate": 1.9887058823529412e-05, "loss": 1.4885, "step": 180 }, { "epoch": 0.03, "learning_rate": 1.9880784313725494e-05, "loss": 1.3615, "step": 190 }, { "epoch": 0.03, "learning_rate": 1.987450980392157e-05, "loss": 1.4021, "step": 200 }, { "epoch": 0.03, "learning_rate": 1.986823529411765e-05, "loss": 1.4081, "step": 210 }, { "epoch": 0.03, "learning_rate": 1.9861960784313726e-05, "loss": 1.2757, "step": 220 }, { "epoch": 0.04, "learning_rate": 1.9855686274509804e-05, "loss": 1.3967, "step": 230 }, { "epoch": 0.04, "learning_rate": 1.9849411764705885e-05, "loss": 1.3048, "step": 240 }, { "epoch": 0.04, "learning_rate": 1.9843137254901963e-05, "loss": 1.2477, "step": 250 }, { "epoch": 0.04, "learning_rate": 1.983686274509804e-05, "loss": 1.214, "step": 260 }, { "epoch": 0.04, "learning_rate": 1.9830588235294118e-05, "loss": 1.1004, "step": 270 }, { "epoch": 0.04, "learning_rate": 1.98243137254902e-05, "loss": 1.0673, "step": 280 }, { "epoch": 0.05, "learning_rate": 1.9818039215686277e-05, "loss": 1.1414, "step": 290 }, { "epoch": 0.05, "learning_rate": 1.9811764705882354e-05, "loss": 1.0788, "step": 300 }, { "epoch": 0.05, "learning_rate": 1.9805490196078435e-05, "loss": 1.1828, "step": 310 }, { "epoch": 0.05, "learning_rate": 1.979921568627451e-05, "loss": 1.0224, "step": 320 }, { "epoch": 0.05, "learning_rate": 1.979294117647059e-05, "loss": 1.0185, "step": 330 }, { "epoch": 0.05, "learning_rate": 1.9786666666666668e-05, "loss": 0.9836, "step": 340 }, { "epoch": 0.05, "learning_rate": 1.9780392156862746e-05, "loss": 1.0133, "step": 350 }, { "epoch": 0.06, "learning_rate": 1.9774117647058827e-05, "loss": 1.0461, "step": 360 }, { "epoch": 0.06, "learning_rate": 1.9767843137254904e-05, "loss": 1.0084, "step": 370 }, { "epoch": 0.06, "learning_rate": 1.9761568627450982e-05, "loss": 0.9133, "step": 380 }, { "epoch": 0.06, "learning_rate": 1.9755294117647063e-05, "loss": 0.8537, "step": 390 }, { "epoch": 0.06, "learning_rate": 1.9749019607843137e-05, "loss": 0.8897, "step": 400 }, { "epoch": 0.06, "learning_rate": 1.9742745098039218e-05, "loss": 1.0199, "step": 410 }, { "epoch": 0.07, "learning_rate": 1.9736470588235296e-05, "loss": 0.9376, "step": 420 }, { "epoch": 0.07, "learning_rate": 1.9730196078431374e-05, "loss": 0.9112, "step": 430 }, { "epoch": 0.07, "learning_rate": 1.9723921568627455e-05, "loss": 0.9608, "step": 440 }, { "epoch": 0.07, "learning_rate": 1.971764705882353e-05, "loss": 1.1142, "step": 450 }, { "epoch": 0.07, "learning_rate": 1.971137254901961e-05, "loss": 0.9409, "step": 460 }, { "epoch": 0.07, "learning_rate": 1.9705098039215687e-05, "loss": 0.8876, "step": 470 }, { "epoch": 0.08, "learning_rate": 1.9698823529411765e-05, "loss": 0.9268, "step": 480 }, { "epoch": 0.08, "learning_rate": 1.9692549019607846e-05, "loss": 0.8593, "step": 490 }, { "epoch": 0.08, "learning_rate": 1.9686274509803924e-05, "loss": 0.83, "step": 500 }, { "epoch": 0.08, "learning_rate": 1.968e-05, "loss": 0.9209, "step": 510 }, { "epoch": 0.08, "learning_rate": 1.967372549019608e-05, "loss": 0.8907, "step": 520 }, { "epoch": 0.08, "learning_rate": 1.9667450980392157e-05, "loss": 0.9432, "step": 530 }, { "epoch": 0.08, "learning_rate": 1.9661176470588238e-05, "loss": 0.6645, "step": 540 }, { "epoch": 0.09, "learning_rate": 1.9654901960784315e-05, "loss": 0.9797, "step": 550 }, { "epoch": 0.09, "learning_rate": 1.9648627450980393e-05, "loss": 0.9579, "step": 560 }, { "epoch": 0.09, "learning_rate": 1.9642352941176474e-05, "loss": 0.7805, "step": 570 }, { "epoch": 0.09, "learning_rate": 1.963607843137255e-05, "loss": 0.9182, "step": 580 }, { "epoch": 0.09, "learning_rate": 1.962980392156863e-05, "loss": 1.0498, "step": 590 }, { "epoch": 0.09, "learning_rate": 1.9623529411764707e-05, "loss": 0.7954, "step": 600 }, { "epoch": 0.1, "learning_rate": 1.9617254901960784e-05, "loss": 0.8751, "step": 610 }, { "epoch": 0.1, "learning_rate": 1.9610980392156865e-05, "loss": 0.8905, "step": 620 }, { "epoch": 0.1, "learning_rate": 1.9604705882352943e-05, "loss": 0.7198, "step": 630 }, { "epoch": 0.1, "learning_rate": 1.959843137254902e-05, "loss": 0.6002, "step": 640 }, { "epoch": 0.1, "learning_rate": 1.9592156862745098e-05, "loss": 0.7463, "step": 650 }, { "epoch": 0.1, "learning_rate": 1.958588235294118e-05, "loss": 0.6625, "step": 660 }, { "epoch": 0.11, "learning_rate": 1.9579607843137257e-05, "loss": 0.7436, "step": 670 }, { "epoch": 0.11, "learning_rate": 1.9573333333333335e-05, "loss": 0.8655, "step": 680 }, { "epoch": 0.11, "learning_rate": 1.9567058823529412e-05, "loss": 0.6307, "step": 690 }, { "epoch": 0.11, "learning_rate": 1.9560784313725493e-05, "loss": 0.8614, "step": 700 }, { "epoch": 0.11, "learning_rate": 1.955450980392157e-05, "loss": 0.7712, "step": 710 }, { "epoch": 0.11, "learning_rate": 1.954823529411765e-05, "loss": 0.84, "step": 720 }, { "epoch": 0.11, "learning_rate": 1.9541960784313726e-05, "loss": 0.6539, "step": 730 }, { "epoch": 0.12, "learning_rate": 1.9535686274509807e-05, "loss": 0.7001, "step": 740 }, { "epoch": 0.12, "learning_rate": 1.9529411764705885e-05, "loss": 0.726, "step": 750 }, { "epoch": 0.12, "learning_rate": 1.9523137254901962e-05, "loss": 0.8089, "step": 760 }, { "epoch": 0.12, "learning_rate": 1.9516862745098043e-05, "loss": 0.7691, "step": 770 }, { "epoch": 0.12, "learning_rate": 1.9510588235294118e-05, "loss": 0.6764, "step": 780 }, { "epoch": 0.12, "learning_rate": 1.95043137254902e-05, "loss": 0.5204, "step": 790 }, { "epoch": 0.13, "learning_rate": 1.9498039215686276e-05, "loss": 0.6394, "step": 800 }, { "epoch": 0.13, "learning_rate": 1.9491764705882354e-05, "loss": 0.6033, "step": 810 }, { "epoch": 0.13, "learning_rate": 1.9485490196078435e-05, "loss": 0.6866, "step": 820 }, { "epoch": 0.13, "learning_rate": 1.947921568627451e-05, "loss": 0.6494, "step": 830 }, { "epoch": 0.13, "learning_rate": 1.947294117647059e-05, "loss": 0.7317, "step": 840 }, { "epoch": 0.13, "learning_rate": 1.9466666666666668e-05, "loss": 0.6113, "step": 850 }, { "epoch": 0.13, "learning_rate": 1.9460392156862745e-05, "loss": 0.6295, "step": 860 }, { "epoch": 0.14, "learning_rate": 1.9454117647058826e-05, "loss": 0.754, "step": 870 }, { "epoch": 0.14, "learning_rate": 1.9447843137254904e-05, "loss": 0.489, "step": 880 }, { "epoch": 0.14, "learning_rate": 1.944156862745098e-05, "loss": 0.6729, "step": 890 }, { "epoch": 0.14, "learning_rate": 1.9435294117647063e-05, "loss": 0.5822, "step": 900 }, { "epoch": 0.14, "learning_rate": 1.9429019607843137e-05, "loss": 0.6276, "step": 910 }, { "epoch": 0.14, "learning_rate": 1.9422745098039218e-05, "loss": 0.752, "step": 920 }, { "epoch": 0.15, "learning_rate": 1.9416470588235296e-05, "loss": 0.5482, "step": 930 }, { "epoch": 0.15, "learning_rate": 1.9410196078431373e-05, "loss": 0.4547, "step": 940 }, { "epoch": 0.15, "learning_rate": 1.9403921568627454e-05, "loss": 0.6781, "step": 950 }, { "epoch": 0.15, "learning_rate": 1.939764705882353e-05, "loss": 0.7318, "step": 960 }, { "epoch": 0.15, "learning_rate": 1.939137254901961e-05, "loss": 0.587, "step": 970 }, { "epoch": 0.15, "learning_rate": 1.9385098039215687e-05, "loss": 0.5052, "step": 980 }, { "epoch": 0.16, "learning_rate": 1.9378823529411765e-05, "loss": 0.6496, "step": 990 }, { "epoch": 0.16, "learning_rate": 1.9372549019607846e-05, "loss": 0.5834, "step": 1000 }, { "epoch": 0.16, "learning_rate": 1.9366274509803923e-05, "loss": 0.6174, "step": 1010 }, { "epoch": 0.16, "learning_rate": 1.936e-05, "loss": 0.7505, "step": 1020 }, { "epoch": 0.16, "learning_rate": 1.935372549019608e-05, "loss": 0.5905, "step": 1030 }, { "epoch": 0.16, "learning_rate": 1.934745098039216e-05, "loss": 0.4801, "step": 1040 }, { "epoch": 0.16, "learning_rate": 1.9341176470588237e-05, "loss": 0.6023, "step": 1050 }, { "epoch": 0.17, "learning_rate": 1.9334901960784315e-05, "loss": 0.5718, "step": 1060 }, { "epoch": 0.17, "learning_rate": 1.9328627450980393e-05, "loss": 0.5193, "step": 1070 }, { "epoch": 0.17, "learning_rate": 1.9322352941176474e-05, "loss": 0.4298, "step": 1080 }, { "epoch": 0.17, "learning_rate": 1.931607843137255e-05, "loss": 0.6194, "step": 1090 }, { "epoch": 0.17, "learning_rate": 1.930980392156863e-05, "loss": 0.702, "step": 1100 }, { "epoch": 0.17, "learning_rate": 1.9303529411764706e-05, "loss": 0.6076, "step": 1110 }, { "epoch": 0.18, "learning_rate": 1.9297254901960787e-05, "loss": 0.5032, "step": 1120 }, { "epoch": 0.18, "learning_rate": 1.9290980392156865e-05, "loss": 0.5564, "step": 1130 }, { "epoch": 0.18, "learning_rate": 1.9284705882352943e-05, "loss": 0.6277, "step": 1140 }, { "epoch": 0.18, "learning_rate": 1.927843137254902e-05, "loss": 0.4949, "step": 1150 }, { "epoch": 0.18, "learning_rate": 1.9272156862745098e-05, "loss": 0.6322, "step": 1160 }, { "epoch": 0.18, "learning_rate": 1.926588235294118e-05, "loss": 0.5901, "step": 1170 }, { "epoch": 0.19, "learning_rate": 1.9259607843137257e-05, "loss": 0.5491, "step": 1180 }, { "epoch": 0.19, "learning_rate": 1.9253333333333334e-05, "loss": 0.4269, "step": 1190 }, { "epoch": 0.19, "learning_rate": 1.9247058823529415e-05, "loss": 0.6592, "step": 1200 }, { "epoch": 0.19, "learning_rate": 1.9240784313725493e-05, "loss": 0.3821, "step": 1210 }, { "epoch": 0.19, "learning_rate": 1.923450980392157e-05, "loss": 0.3673, "step": 1220 }, { "epoch": 0.19, "learning_rate": 1.9228235294117648e-05, "loss": 0.4858, "step": 1230 }, { "epoch": 0.19, "learning_rate": 1.9221960784313726e-05, "loss": 0.7259, "step": 1240 }, { "epoch": 0.2, "learning_rate": 1.9215686274509807e-05, "loss": 0.512, "step": 1250 }, { "epoch": 0.2, "learning_rate": 1.9209411764705884e-05, "loss": 0.409, "step": 1260 }, { "epoch": 0.2, "learning_rate": 1.9203137254901962e-05, "loss": 0.7868, "step": 1270 }, { "epoch": 0.2, "learning_rate": 1.9196862745098043e-05, "loss": 0.4972, "step": 1280 }, { "epoch": 0.2, "learning_rate": 1.9190588235294117e-05, "loss": 0.5942, "step": 1290 }, { "epoch": 0.2, "learning_rate": 1.9184313725490198e-05, "loss": 0.4825, "step": 1300 }, { "epoch": 0.21, "learning_rate": 1.9178039215686276e-05, "loss": 0.5038, "step": 1310 }, { "epoch": 0.21, "learning_rate": 1.9171764705882354e-05, "loss": 0.5512, "step": 1320 }, { "epoch": 0.21, "learning_rate": 1.9165490196078435e-05, "loss": 0.4745, "step": 1330 }, { "epoch": 0.21, "learning_rate": 1.915921568627451e-05, "loss": 0.532, "step": 1340 }, { "epoch": 0.21, "learning_rate": 1.915294117647059e-05, "loss": 0.5329, "step": 1350 }, { "epoch": 0.21, "learning_rate": 1.9146666666666667e-05, "loss": 0.5939, "step": 1360 }, { "epoch": 0.21, "learning_rate": 1.9140392156862745e-05, "loss": 0.7166, "step": 1370 }, { "epoch": 0.22, "learning_rate": 1.9134117647058826e-05, "loss": 0.5228, "step": 1380 }, { "epoch": 0.22, "learning_rate": 1.9127843137254904e-05, "loss": 0.6598, "step": 1390 }, { "epoch": 0.22, "learning_rate": 1.912156862745098e-05, "loss": 0.4777, "step": 1400 }, { "epoch": 0.22, "learning_rate": 1.9115294117647062e-05, "loss": 0.6095, "step": 1410 }, { "epoch": 0.22, "learning_rate": 1.9109019607843137e-05, "loss": 0.545, "step": 1420 }, { "epoch": 0.22, "learning_rate": 1.9102745098039218e-05, "loss": 0.5326, "step": 1430 }, { "epoch": 0.23, "learning_rate": 1.9096470588235295e-05, "loss": 0.6153, "step": 1440 }, { "epoch": 0.23, "learning_rate": 1.9090196078431373e-05, "loss": 0.3665, "step": 1450 }, { "epoch": 0.23, "learning_rate": 1.9083921568627454e-05, "loss": 0.5732, "step": 1460 }, { "epoch": 0.23, "learning_rate": 1.907764705882353e-05, "loss": 0.4221, "step": 1470 }, { "epoch": 0.23, "learning_rate": 1.907137254901961e-05, "loss": 0.742, "step": 1480 }, { "epoch": 0.23, "learning_rate": 1.9065098039215687e-05, "loss": 0.7577, "step": 1490 }, { "epoch": 0.24, "learning_rate": 1.9058823529411764e-05, "loss": 0.8295, "step": 1500 }, { "epoch": 0.24, "learning_rate": 1.9052549019607845e-05, "loss": 0.6034, "step": 1510 }, { "epoch": 0.24, "learning_rate": 1.9046274509803923e-05, "loss": 0.4211, "step": 1520 }, { "epoch": 0.24, "learning_rate": 1.904e-05, "loss": 0.6399, "step": 1530 }, { "epoch": 0.24, "learning_rate": 1.9033725490196078e-05, "loss": 0.5496, "step": 1540 }, { "epoch": 0.24, "learning_rate": 1.902745098039216e-05, "loss": 0.4106, "step": 1550 }, { "epoch": 0.24, "learning_rate": 1.9021176470588237e-05, "loss": 0.4963, "step": 1560 }, { "epoch": 0.25, "learning_rate": 1.9014901960784315e-05, "loss": 0.4513, "step": 1570 }, { "epoch": 0.25, "learning_rate": 1.9008627450980396e-05, "loss": 0.5222, "step": 1580 }, { "epoch": 0.25, "learning_rate": 1.9002352941176473e-05, "loss": 0.6034, "step": 1590 }, { "epoch": 0.25, "learning_rate": 1.899607843137255e-05, "loss": 0.3557, "step": 1600 }, { "epoch": 0.25, "learning_rate": 1.898980392156863e-05, "loss": 0.6165, "step": 1610 }, { "epoch": 0.25, "learning_rate": 1.8983529411764706e-05, "loss": 0.6132, "step": 1620 }, { "epoch": 0.26, "learning_rate": 1.8977254901960787e-05, "loss": 0.6371, "step": 1630 }, { "epoch": 0.26, "learning_rate": 1.8970980392156865e-05, "loss": 0.3825, "step": 1640 }, { "epoch": 0.26, "learning_rate": 1.8964705882352942e-05, "loss": 0.6457, "step": 1650 }, { "epoch": 0.26, "learning_rate": 1.8958431372549023e-05, "loss": 0.2875, "step": 1660 }, { "epoch": 0.26, "learning_rate": 1.8952156862745098e-05, "loss": 0.6183, "step": 1670 }, { "epoch": 0.26, "learning_rate": 1.894588235294118e-05, "loss": 0.4728, "step": 1680 }, { "epoch": 0.27, "learning_rate": 1.8939607843137256e-05, "loss": 0.5733, "step": 1690 }, { "epoch": 0.27, "learning_rate": 1.8933333333333334e-05, "loss": 0.5763, "step": 1700 }, { "epoch": 0.27, "learning_rate": 1.8927058823529415e-05, "loss": 0.5293, "step": 1710 }, { "epoch": 0.27, "learning_rate": 1.8920784313725493e-05, "loss": 0.5856, "step": 1720 }, { "epoch": 0.27, "learning_rate": 1.891450980392157e-05, "loss": 0.5123, "step": 1730 }, { "epoch": 0.27, "learning_rate": 1.8908235294117648e-05, "loss": 0.4412, "step": 1740 }, { "epoch": 0.27, "learning_rate": 1.8901960784313725e-05, "loss": 0.652, "step": 1750 }, { "epoch": 0.28, "learning_rate": 1.8895686274509806e-05, "loss": 0.4982, "step": 1760 }, { "epoch": 0.28, "learning_rate": 1.8889411764705884e-05, "loss": 0.7407, "step": 1770 }, { "epoch": 0.28, "learning_rate": 1.888313725490196e-05, "loss": 0.6328, "step": 1780 }, { "epoch": 0.28, "learning_rate": 1.8876862745098043e-05, "loss": 0.5365, "step": 1790 }, { "epoch": 0.28, "learning_rate": 1.8870588235294117e-05, "loss": 0.2584, "step": 1800 }, { "epoch": 0.28, "learning_rate": 1.8864313725490198e-05, "loss": 0.6181, "step": 1810 }, { "epoch": 0.29, "learning_rate": 1.8858039215686276e-05, "loss": 0.7211, "step": 1820 }, { "epoch": 0.29, "learning_rate": 1.8851764705882353e-05, "loss": 0.5348, "step": 1830 }, { "epoch": 0.29, "learning_rate": 1.8845490196078434e-05, "loss": 0.6387, "step": 1840 }, { "epoch": 0.29, "learning_rate": 1.8839215686274512e-05, "loss": 0.5686, "step": 1850 }, { "epoch": 0.29, "learning_rate": 1.883294117647059e-05, "loss": 0.2804, "step": 1860 }, { "epoch": 0.29, "learning_rate": 1.8826666666666667e-05, "loss": 0.4973, "step": 1870 }, { "epoch": 0.29, "learning_rate": 1.8820392156862745e-05, "loss": 0.6983, "step": 1880 }, { "epoch": 0.3, "learning_rate": 1.8814117647058826e-05, "loss": 0.5673, "step": 1890 }, { "epoch": 0.3, "learning_rate": 1.8807843137254903e-05, "loss": 0.5574, "step": 1900 }, { "epoch": 0.3, "learning_rate": 1.880156862745098e-05, "loss": 0.7118, "step": 1910 }, { "epoch": 0.3, "learning_rate": 1.8795294117647062e-05, "loss": 0.5256, "step": 1920 }, { "epoch": 0.3, "learning_rate": 1.878901960784314e-05, "loss": 0.6654, "step": 1930 }, { "epoch": 0.3, "learning_rate": 1.8782745098039217e-05, "loss": 0.4579, "step": 1940 }, { "epoch": 0.31, "learning_rate": 1.8776470588235295e-05, "loss": 0.5034, "step": 1950 }, { "epoch": 0.31, "learning_rate": 1.8770196078431373e-05, "loss": 0.591, "step": 1960 }, { "epoch": 0.31, "learning_rate": 1.8763921568627454e-05, "loss": 0.5224, "step": 1970 }, { "epoch": 0.31, "learning_rate": 1.875764705882353e-05, "loss": 0.4024, "step": 1980 }, { "epoch": 0.31, "learning_rate": 1.875137254901961e-05, "loss": 0.5707, "step": 1990 }, { "epoch": 0.31, "learning_rate": 1.8745098039215686e-05, "loss": 0.5684, "step": 2000 }, { "epoch": 0.32, "learning_rate": 1.8738823529411767e-05, "loss": 0.3196, "step": 2010 }, { "epoch": 0.32, "learning_rate": 1.8732549019607845e-05, "loss": 0.6304, "step": 2020 }, { "epoch": 0.32, "learning_rate": 1.8726274509803923e-05, "loss": 0.3952, "step": 2030 }, { "epoch": 0.32, "learning_rate": 1.8720000000000004e-05, "loss": 0.6595, "step": 2040 }, { "epoch": 0.32, "learning_rate": 1.8713725490196078e-05, "loss": 0.6493, "step": 2050 }, { "epoch": 0.32, "learning_rate": 1.870745098039216e-05, "loss": 0.5847, "step": 2060 }, { "epoch": 0.32, "learning_rate": 1.8701176470588237e-05, "loss": 0.4594, "step": 2070 }, { "epoch": 0.33, "learning_rate": 1.8694901960784314e-05, "loss": 0.6175, "step": 2080 }, { "epoch": 0.33, "learning_rate": 1.8688627450980395e-05, "loss": 0.4409, "step": 2090 }, { "epoch": 0.33, "learning_rate": 1.8682352941176473e-05, "loss": 0.3169, "step": 2100 }, { "epoch": 0.33, "learning_rate": 1.867607843137255e-05, "loss": 0.5847, "step": 2110 }, { "epoch": 0.33, "learning_rate": 1.866980392156863e-05, "loss": 0.4628, "step": 2120 }, { "epoch": 0.33, "learning_rate": 1.8663529411764706e-05, "loss": 0.5075, "step": 2130 }, { "epoch": 0.34, "learning_rate": 1.8657254901960787e-05, "loss": 0.6629, "step": 2140 }, { "epoch": 0.34, "learning_rate": 1.8650980392156864e-05, "loss": 0.6518, "step": 2150 }, { "epoch": 0.34, "learning_rate": 1.8644705882352942e-05, "loss": 0.436, "step": 2160 }, { "epoch": 0.34, "learning_rate": 1.8638431372549023e-05, "loss": 0.5047, "step": 2170 }, { "epoch": 0.34, "learning_rate": 1.8632156862745097e-05, "loss": 0.3969, "step": 2180 }, { "epoch": 0.34, "learning_rate": 1.862588235294118e-05, "loss": 0.488, "step": 2190 }, { "epoch": 0.35, "learning_rate": 1.8619607843137256e-05, "loss": 0.3727, "step": 2200 }, { "epoch": 0.35, "learning_rate": 1.8613333333333334e-05, "loss": 0.5949, "step": 2210 }, { "epoch": 0.35, "learning_rate": 1.8607058823529415e-05, "loss": 0.3066, "step": 2220 }, { "epoch": 0.35, "learning_rate": 1.8600784313725492e-05, "loss": 0.322, "step": 2230 }, { "epoch": 0.35, "learning_rate": 1.859450980392157e-05, "loss": 0.4223, "step": 2240 }, { "epoch": 0.35, "learning_rate": 1.8588235294117647e-05, "loss": 0.4347, "step": 2250 }, { "epoch": 0.35, "learning_rate": 1.8581960784313725e-05, "loss": 0.4379, "step": 2260 }, { "epoch": 0.36, "learning_rate": 1.8575686274509806e-05, "loss": 0.356, "step": 2270 }, { "epoch": 0.36, "learning_rate": 1.8569411764705884e-05, "loss": 0.5919, "step": 2280 }, { "epoch": 0.36, "learning_rate": 1.856313725490196e-05, "loss": 0.6687, "step": 2290 }, { "epoch": 0.36, "learning_rate": 1.8556862745098042e-05, "loss": 0.2783, "step": 2300 }, { "epoch": 0.36, "learning_rate": 1.855058823529412e-05, "loss": 0.6347, "step": 2310 }, { "epoch": 0.36, "learning_rate": 1.8544313725490198e-05, "loss": 0.5836, "step": 2320 }, { "epoch": 0.37, "learning_rate": 1.8538039215686275e-05, "loss": 0.4925, "step": 2330 }, { "epoch": 0.37, "learning_rate": 1.8531764705882353e-05, "loss": 0.4425, "step": 2340 }, { "epoch": 0.37, "learning_rate": 1.8525490196078434e-05, "loss": 0.6163, "step": 2350 }, { "epoch": 0.37, "learning_rate": 1.851921568627451e-05, "loss": 0.5169, "step": 2360 }, { "epoch": 0.37, "learning_rate": 1.851294117647059e-05, "loss": 0.4921, "step": 2370 }, { "epoch": 0.37, "learning_rate": 1.8506666666666667e-05, "loss": 0.6734, "step": 2380 }, { "epoch": 0.37, "learning_rate": 1.8500392156862748e-05, "loss": 0.3967, "step": 2390 }, { "epoch": 0.38, "learning_rate": 1.8494117647058825e-05, "loss": 0.4749, "step": 2400 }, { "epoch": 0.38, "learning_rate": 1.8487843137254903e-05, "loss": 0.4886, "step": 2410 }, { "epoch": 0.38, "learning_rate": 1.848156862745098e-05, "loss": 0.53, "step": 2420 }, { "epoch": 0.38, "learning_rate": 1.8475294117647062e-05, "loss": 0.2947, "step": 2430 }, { "epoch": 0.38, "learning_rate": 1.846901960784314e-05, "loss": 0.6457, "step": 2440 }, { "epoch": 0.38, "learning_rate": 1.8462745098039217e-05, "loss": 0.5484, "step": 2450 }, { "epoch": 0.39, "learning_rate": 1.8456470588235295e-05, "loss": 0.4617, "step": 2460 }, { "epoch": 0.39, "learning_rate": 1.8450196078431376e-05, "loss": 0.4708, "step": 2470 }, { "epoch": 0.39, "learning_rate": 1.8443921568627453e-05, "loss": 0.3851, "step": 2480 }, { "epoch": 0.39, "learning_rate": 1.843764705882353e-05, "loss": 0.6457, "step": 2490 }, { "epoch": 0.39, "learning_rate": 1.843137254901961e-05, "loss": 0.5013, "step": 2500 }, { "epoch": 0.39, "learning_rate": 1.8425098039215686e-05, "loss": 0.4466, "step": 2510 }, { "epoch": 0.4, "learning_rate": 1.8418823529411767e-05, "loss": 0.4183, "step": 2520 }, { "epoch": 0.4, "learning_rate": 1.8412549019607845e-05, "loss": 0.4117, "step": 2530 }, { "epoch": 0.4, "learning_rate": 1.8406274509803922e-05, "loss": 0.4856, "step": 2540 }, { "epoch": 0.4, "learning_rate": 1.8400000000000003e-05, "loss": 0.4091, "step": 2550 }, { "epoch": 0.4, "learning_rate": 1.839372549019608e-05, "loss": 0.6208, "step": 2560 }, { "epoch": 0.4, "learning_rate": 1.838745098039216e-05, "loss": 0.4862, "step": 2570 }, { "epoch": 0.4, "learning_rate": 1.8381176470588236e-05, "loss": 0.5889, "step": 2580 }, { "epoch": 0.41, "learning_rate": 1.8374901960784314e-05, "loss": 0.5868, "step": 2590 }, { "epoch": 0.41, "learning_rate": 1.8368627450980395e-05, "loss": 0.6107, "step": 2600 }, { "epoch": 0.41, "learning_rate": 1.8362352941176473e-05, "loss": 0.5159, "step": 2610 }, { "epoch": 0.41, "learning_rate": 1.835607843137255e-05, "loss": 0.3524, "step": 2620 }, { "epoch": 0.41, "learning_rate": 1.834980392156863e-05, "loss": 0.3744, "step": 2630 }, { "epoch": 0.41, "learning_rate": 1.8343529411764705e-05, "loss": 0.5817, "step": 2640 }, { "epoch": 0.42, "learning_rate": 1.8337254901960786e-05, "loss": 0.3613, "step": 2650 }, { "epoch": 0.42, "learning_rate": 1.8330980392156864e-05, "loss": 0.426, "step": 2660 }, { "epoch": 0.42, "learning_rate": 1.8324705882352942e-05, "loss": 0.3909, "step": 2670 }, { "epoch": 0.42, "learning_rate": 1.8318431372549023e-05, "loss": 0.6015, "step": 2680 }, { "epoch": 0.42, "learning_rate": 1.8312156862745097e-05, "loss": 0.4558, "step": 2690 }, { "epoch": 0.42, "learning_rate": 1.8305882352941178e-05, "loss": 0.5472, "step": 2700 }, { "epoch": 0.43, "learning_rate": 1.8299607843137256e-05, "loss": 0.5786, "step": 2710 }, { "epoch": 0.43, "learning_rate": 1.8293333333333333e-05, "loss": 0.5748, "step": 2720 }, { "epoch": 0.43, "learning_rate": 1.8287058823529414e-05, "loss": 0.5519, "step": 2730 }, { "epoch": 0.43, "learning_rate": 1.8280784313725492e-05, "loss": 0.5655, "step": 2740 }, { "epoch": 0.43, "learning_rate": 1.827450980392157e-05, "loss": 0.5407, "step": 2750 }, { "epoch": 0.43, "learning_rate": 1.826823529411765e-05, "loss": 0.6617, "step": 2760 }, { "epoch": 0.43, "learning_rate": 1.8261960784313725e-05, "loss": 0.699, "step": 2770 }, { "epoch": 0.44, "learning_rate": 1.8255686274509806e-05, "loss": 0.487, "step": 2780 }, { "epoch": 0.44, "learning_rate": 1.8249411764705883e-05, "loss": 0.4486, "step": 2790 }, { "epoch": 0.44, "learning_rate": 1.824313725490196e-05, "loss": 0.5308, "step": 2800 }, { "epoch": 0.44, "learning_rate": 1.8236862745098042e-05, "loss": 0.5604, "step": 2810 }, { "epoch": 0.44, "learning_rate": 1.823058823529412e-05, "loss": 0.4307, "step": 2820 }, { "epoch": 0.44, "learning_rate": 1.8224313725490197e-05, "loss": 0.5339, "step": 2830 }, { "epoch": 0.45, "learning_rate": 1.8218039215686275e-05, "loss": 0.6893, "step": 2840 }, { "epoch": 0.45, "learning_rate": 1.8211764705882356e-05, "loss": 0.2871, "step": 2850 }, { "epoch": 0.45, "learning_rate": 1.8205490196078434e-05, "loss": 0.4795, "step": 2860 }, { "epoch": 0.45, "learning_rate": 1.819921568627451e-05, "loss": 0.403, "step": 2870 }, { "epoch": 0.45, "learning_rate": 1.819294117647059e-05, "loss": 0.5494, "step": 2880 }, { "epoch": 0.45, "learning_rate": 1.8186666666666666e-05, "loss": 0.336, "step": 2890 }, { "epoch": 0.45, "learning_rate": 1.8180392156862747e-05, "loss": 0.6093, "step": 2900 }, { "epoch": 0.46, "learning_rate": 1.8174117647058825e-05, "loss": 0.3189, "step": 2910 }, { "epoch": 0.46, "learning_rate": 1.8167843137254903e-05, "loss": 0.4858, "step": 2920 }, { "epoch": 0.46, "learning_rate": 1.8161568627450984e-05, "loss": 0.3438, "step": 2930 }, { "epoch": 0.46, "learning_rate": 1.815529411764706e-05, "loss": 0.2844, "step": 2940 }, { "epoch": 0.46, "learning_rate": 1.814901960784314e-05, "loss": 0.5394, "step": 2950 }, { "epoch": 0.46, "learning_rate": 1.8142745098039217e-05, "loss": 0.4374, "step": 2960 }, { "epoch": 0.47, "learning_rate": 1.8136470588235294e-05, "loss": 0.5053, "step": 2970 }, { "epoch": 0.47, "learning_rate": 1.8130196078431375e-05, "loss": 0.3314, "step": 2980 }, { "epoch": 0.47, "learning_rate": 1.8123921568627453e-05, "loss": 0.4635, "step": 2990 }, { "epoch": 0.47, "learning_rate": 1.811764705882353e-05, "loss": 0.4938, "step": 3000 }, { "epoch": 0.47, "learning_rate": 1.811137254901961e-05, "loss": 0.2762, "step": 3010 }, { "epoch": 0.47, "learning_rate": 1.8105098039215686e-05, "loss": 0.448, "step": 3020 }, { "epoch": 0.48, "learning_rate": 1.8098823529411767e-05, "loss": 0.4077, "step": 3030 }, { "epoch": 0.48, "learning_rate": 1.8092549019607844e-05, "loss": 0.4459, "step": 3040 }, { "epoch": 0.48, "learning_rate": 1.8086274509803922e-05, "loss": 0.532, "step": 3050 }, { "epoch": 0.48, "learning_rate": 1.8080000000000003e-05, "loss": 0.6144, "step": 3060 }, { "epoch": 0.48, "learning_rate": 1.807372549019608e-05, "loss": 0.3598, "step": 3070 }, { "epoch": 0.48, "learning_rate": 1.806745098039216e-05, "loss": 0.6793, "step": 3080 }, { "epoch": 0.48, "learning_rate": 1.8061176470588236e-05, "loss": 0.4201, "step": 3090 }, { "epoch": 0.49, "learning_rate": 1.8054901960784314e-05, "loss": 0.423, "step": 3100 }, { "epoch": 0.49, "learning_rate": 1.8048627450980395e-05, "loss": 0.0962, "step": 3110 }, { "epoch": 0.49, "learning_rate": 1.8042352941176472e-05, "loss": 0.4085, "step": 3120 }, { "epoch": 0.49, "learning_rate": 1.803607843137255e-05, "loss": 0.4184, "step": 3130 }, { "epoch": 0.49, "learning_rate": 1.802980392156863e-05, "loss": 0.4435, "step": 3140 }, { "epoch": 0.49, "learning_rate": 1.8023529411764705e-05, "loss": 0.6404, "step": 3150 }, { "epoch": 0.5, "learning_rate": 1.8017254901960786e-05, "loss": 0.3995, "step": 3160 }, { "epoch": 0.5, "learning_rate": 1.8010980392156864e-05, "loss": 0.4047, "step": 3170 }, { "epoch": 0.5, "learning_rate": 1.800470588235294e-05, "loss": 0.4449, "step": 3180 }, { "epoch": 0.5, "learning_rate": 1.7998431372549022e-05, "loss": 0.4696, "step": 3190 }, { "epoch": 0.5, "learning_rate": 1.79921568627451e-05, "loss": 0.5717, "step": 3200 }, { "epoch": 0.5, "learning_rate": 1.7985882352941178e-05, "loss": 0.5161, "step": 3210 }, { "epoch": 0.51, "learning_rate": 1.7979607843137255e-05, "loss": 0.5331, "step": 3220 }, { "epoch": 0.51, "learning_rate": 1.7973333333333333e-05, "loss": 0.5415, "step": 3230 }, { "epoch": 0.51, "learning_rate": 1.7967058823529414e-05, "loss": 0.3641, "step": 3240 }, { "epoch": 0.51, "learning_rate": 1.796078431372549e-05, "loss": 0.5812, "step": 3250 }, { "epoch": 0.51, "learning_rate": 1.795450980392157e-05, "loss": 0.4383, "step": 3260 }, { "epoch": 0.51, "learning_rate": 1.794823529411765e-05, "loss": 0.619, "step": 3270 }, { "epoch": 0.51, "learning_rate": 1.7941960784313728e-05, "loss": 0.5031, "step": 3280 }, { "epoch": 0.52, "learning_rate": 1.7935686274509805e-05, "loss": 0.2926, "step": 3290 }, { "epoch": 0.52, "learning_rate": 1.7929411764705883e-05, "loss": 0.4512, "step": 3300 }, { "epoch": 0.52, "learning_rate": 1.7923137254901964e-05, "loss": 0.4708, "step": 3310 }, { "epoch": 0.52, "learning_rate": 1.7916862745098042e-05, "loss": 0.5025, "step": 3320 }, { "epoch": 0.52, "learning_rate": 1.791058823529412e-05, "loss": 0.5146, "step": 3330 }, { "epoch": 0.52, "learning_rate": 1.7904313725490197e-05, "loss": 0.3929, "step": 3340 }, { "epoch": 0.53, "learning_rate": 1.7898039215686275e-05, "loss": 0.445, "step": 3350 }, { "epoch": 0.53, "learning_rate": 1.7891764705882356e-05, "loss": 0.4696, "step": 3360 }, { "epoch": 0.53, "learning_rate": 1.7885490196078433e-05, "loss": 0.7052, "step": 3370 }, { "epoch": 0.53, "learning_rate": 1.787921568627451e-05, "loss": 0.4187, "step": 3380 }, { "epoch": 0.53, "learning_rate": 1.7872941176470592e-05, "loss": 0.4291, "step": 3390 }, { "epoch": 0.53, "learning_rate": 1.7866666666666666e-05, "loss": 0.5806, "step": 3400 }, { "epoch": 0.53, "learning_rate": 1.7860392156862747e-05, "loss": 0.3835, "step": 3410 }, { "epoch": 0.54, "learning_rate": 1.7854117647058825e-05, "loss": 0.509, "step": 3420 }, { "epoch": 0.54, "learning_rate": 1.7847843137254902e-05, "loss": 0.3938, "step": 3430 }, { "epoch": 0.54, "learning_rate": 1.7841568627450983e-05, "loss": 0.5569, "step": 3440 }, { "epoch": 0.54, "learning_rate": 1.783529411764706e-05, "loss": 0.4211, "step": 3450 }, { "epoch": 0.54, "learning_rate": 1.782901960784314e-05, "loss": 0.3951, "step": 3460 }, { "epoch": 0.54, "learning_rate": 1.782274509803922e-05, "loss": 0.7058, "step": 3470 }, { "epoch": 0.55, "learning_rate": 1.7816470588235294e-05, "loss": 0.4147, "step": 3480 }, { "epoch": 0.55, "learning_rate": 1.7810196078431375e-05, "loss": 0.4316, "step": 3490 }, { "epoch": 0.55, "learning_rate": 1.7803921568627453e-05, "loss": 0.4652, "step": 3500 }, { "epoch": 0.55, "learning_rate": 1.779764705882353e-05, "loss": 0.3202, "step": 3510 }, { "epoch": 0.55, "learning_rate": 1.779137254901961e-05, "loss": 0.356, "step": 3520 }, { "epoch": 0.55, "learning_rate": 1.7785098039215685e-05, "loss": 0.3736, "step": 3530 }, { "epoch": 0.56, "learning_rate": 1.7778823529411766e-05, "loss": 0.5295, "step": 3540 }, { "epoch": 0.56, "learning_rate": 1.7772549019607844e-05, "loss": 0.4191, "step": 3550 }, { "epoch": 0.56, "learning_rate": 1.7766274509803922e-05, "loss": 0.5823, "step": 3560 }, { "epoch": 0.56, "learning_rate": 1.7760000000000003e-05, "loss": 0.5329, "step": 3570 }, { "epoch": 0.56, "learning_rate": 1.775372549019608e-05, "loss": 0.4112, "step": 3580 }, { "epoch": 0.56, "learning_rate": 1.7747450980392158e-05, "loss": 0.5879, "step": 3590 }, { "epoch": 0.56, "learning_rate": 1.7741176470588236e-05, "loss": 0.3301, "step": 3600 }, { "epoch": 0.57, "learning_rate": 1.7734901960784313e-05, "loss": 0.3224, "step": 3610 }, { "epoch": 0.57, "learning_rate": 1.7728627450980394e-05, "loss": 0.7505, "step": 3620 }, { "epoch": 0.57, "learning_rate": 1.7722352941176472e-05, "loss": 0.6838, "step": 3630 }, { "epoch": 0.57, "learning_rate": 1.771607843137255e-05, "loss": 0.3865, "step": 3640 }, { "epoch": 0.57, "learning_rate": 1.770980392156863e-05, "loss": 0.2459, "step": 3650 }, { "epoch": 0.57, "learning_rate": 1.7703529411764708e-05, "loss": 0.3261, "step": 3660 }, { "epoch": 0.58, "learning_rate": 1.7697254901960786e-05, "loss": 0.3476, "step": 3670 }, { "epoch": 0.58, "learning_rate": 1.7690980392156863e-05, "loss": 0.434, "step": 3680 }, { "epoch": 0.58, "learning_rate": 1.768470588235294e-05, "loss": 0.2215, "step": 3690 }, { "epoch": 0.58, "learning_rate": 1.7678431372549022e-05, "loss": 0.4827, "step": 3700 }, { "epoch": 0.58, "learning_rate": 1.76721568627451e-05, "loss": 0.4416, "step": 3710 }, { "epoch": 0.58, "learning_rate": 1.7665882352941177e-05, "loss": 0.3683, "step": 3720 }, { "epoch": 0.59, "learning_rate": 1.7659607843137255e-05, "loss": 0.2841, "step": 3730 }, { "epoch": 0.59, "learning_rate": 1.7653333333333336e-05, "loss": 0.5965, "step": 3740 }, { "epoch": 0.59, "learning_rate": 1.7647058823529414e-05, "loss": 0.5293, "step": 3750 }, { "epoch": 0.59, "learning_rate": 1.764078431372549e-05, "loss": 0.5331, "step": 3760 }, { "epoch": 0.59, "learning_rate": 1.763450980392157e-05, "loss": 0.4219, "step": 3770 }, { "epoch": 0.59, "learning_rate": 1.762823529411765e-05, "loss": 0.28, "step": 3780 }, { "epoch": 0.59, "learning_rate": 1.7621960784313728e-05, "loss": 0.284, "step": 3790 }, { "epoch": 0.6, "learning_rate": 1.7615686274509805e-05, "loss": 0.5983, "step": 3800 }, { "epoch": 0.6, "learning_rate": 1.7609411764705883e-05, "loss": 0.5753, "step": 3810 }, { "epoch": 0.6, "learning_rate": 1.7603137254901964e-05, "loss": 0.5758, "step": 3820 }, { "epoch": 0.6, "learning_rate": 1.759686274509804e-05, "loss": 0.5001, "step": 3830 }, { "epoch": 0.6, "learning_rate": 1.759058823529412e-05, "loss": 0.5661, "step": 3840 }, { "epoch": 0.6, "learning_rate": 1.75843137254902e-05, "loss": 0.4947, "step": 3850 }, { "epoch": 0.61, "learning_rate": 1.7578039215686274e-05, "loss": 0.4889, "step": 3860 }, { "epoch": 0.61, "learning_rate": 1.7571764705882355e-05, "loss": 0.6032, "step": 3870 }, { "epoch": 0.61, "learning_rate": 1.7565490196078433e-05, "loss": 0.4657, "step": 3880 }, { "epoch": 0.61, "learning_rate": 1.755921568627451e-05, "loss": 0.4748, "step": 3890 }, { "epoch": 0.61, "learning_rate": 1.755294117647059e-05, "loss": 0.6189, "step": 3900 }, { "epoch": 0.61, "learning_rate": 1.7546666666666666e-05, "loss": 0.3276, "step": 3910 }, { "epoch": 0.61, "learning_rate": 1.7540392156862747e-05, "loss": 0.3944, "step": 3920 }, { "epoch": 0.62, "learning_rate": 1.7534117647058824e-05, "loss": 0.4249, "step": 3930 }, { "epoch": 0.62, "learning_rate": 1.7527843137254902e-05, "loss": 0.4523, "step": 3940 }, { "epoch": 0.62, "learning_rate": 1.7521568627450983e-05, "loss": 0.4623, "step": 3950 }, { "epoch": 0.62, "learning_rate": 1.751529411764706e-05, "loss": 0.5576, "step": 3960 }, { "epoch": 0.62, "learning_rate": 1.750901960784314e-05, "loss": 0.3367, "step": 3970 }, { "epoch": 0.62, "learning_rate": 1.750274509803922e-05, "loss": 0.3559, "step": 3980 }, { "epoch": 0.63, "learning_rate": 1.7496470588235294e-05, "loss": 0.4657, "step": 3990 }, { "epoch": 0.63, "learning_rate": 1.7490196078431375e-05, "loss": 0.271, "step": 4000 }, { "epoch": 0.63, "learning_rate": 1.7483921568627452e-05, "loss": 0.371, "step": 4010 }, { "epoch": 0.63, "learning_rate": 1.747764705882353e-05, "loss": 0.456, "step": 4020 }, { "epoch": 0.63, "learning_rate": 1.747137254901961e-05, "loss": 0.4631, "step": 4030 }, { "epoch": 0.63, "learning_rate": 1.746509803921569e-05, "loss": 0.3277, "step": 4040 }, { "epoch": 0.64, "learning_rate": 1.7458823529411766e-05, "loss": 0.3229, "step": 4050 }, { "epoch": 0.64, "learning_rate": 1.7452549019607844e-05, "loss": 0.377, "step": 4060 }, { "epoch": 0.64, "learning_rate": 1.744627450980392e-05, "loss": 0.5313, "step": 4070 }, { "epoch": 0.64, "learning_rate": 1.7440000000000002e-05, "loss": 0.3063, "step": 4080 }, { "epoch": 0.64, "learning_rate": 1.743372549019608e-05, "loss": 0.573, "step": 4090 }, { "epoch": 0.64, "learning_rate": 1.7427450980392158e-05, "loss": 0.4461, "step": 4100 }, { "epoch": 0.64, "learning_rate": 1.7421176470588235e-05, "loss": 0.3573, "step": 4110 }, { "epoch": 0.65, "learning_rate": 1.7414901960784316e-05, "loss": 0.3093, "step": 4120 }, { "epoch": 0.65, "learning_rate": 1.7408627450980394e-05, "loss": 0.4877, "step": 4130 }, { "epoch": 0.65, "learning_rate": 1.740235294117647e-05, "loss": 0.2459, "step": 4140 }, { "epoch": 0.65, "learning_rate": 1.739607843137255e-05, "loss": 0.5925, "step": 4150 }, { "epoch": 0.65, "learning_rate": 1.738980392156863e-05, "loss": 0.2778, "step": 4160 }, { "epoch": 0.65, "learning_rate": 1.7383529411764708e-05, "loss": 0.6329, "step": 4170 }, { "epoch": 0.66, "learning_rate": 1.7377254901960785e-05, "loss": 0.4066, "step": 4180 }, { "epoch": 0.66, "learning_rate": 1.7370980392156863e-05, "loss": 0.4114, "step": 4190 }, { "epoch": 0.66, "learning_rate": 1.7364705882352944e-05, "loss": 0.5276, "step": 4200 }, { "epoch": 0.66, "learning_rate": 1.7358431372549022e-05, "loss": 0.2913, "step": 4210 }, { "epoch": 0.66, "learning_rate": 1.73521568627451e-05, "loss": 0.3409, "step": 4220 }, { "epoch": 0.66, "learning_rate": 1.7345882352941177e-05, "loss": 0.4262, "step": 4230 }, { "epoch": 0.67, "learning_rate": 1.7339607843137255e-05, "loss": 0.5843, "step": 4240 }, { "epoch": 0.67, "learning_rate": 1.7333333333333336e-05, "loss": 0.4588, "step": 4250 }, { "epoch": 0.67, "learning_rate": 1.7327058823529413e-05, "loss": 0.6901, "step": 4260 }, { "epoch": 0.67, "learning_rate": 1.732078431372549e-05, "loss": 0.6224, "step": 4270 }, { "epoch": 0.67, "learning_rate": 1.7314509803921572e-05, "loss": 0.4508, "step": 4280 }, { "epoch": 0.67, "learning_rate": 1.730823529411765e-05, "loss": 0.3203, "step": 4290 }, { "epoch": 0.67, "learning_rate": 1.7301960784313727e-05, "loss": 0.3556, "step": 4300 }, { "epoch": 0.68, "learning_rate": 1.7295686274509805e-05, "loss": 0.2536, "step": 4310 }, { "epoch": 0.68, "learning_rate": 1.7289411764705882e-05, "loss": 0.4612, "step": 4320 }, { "epoch": 0.68, "learning_rate": 1.7283137254901963e-05, "loss": 0.3292, "step": 4330 }, { "epoch": 0.68, "learning_rate": 1.727686274509804e-05, "loss": 0.3124, "step": 4340 }, { "epoch": 0.68, "learning_rate": 1.727058823529412e-05, "loss": 0.5192, "step": 4350 }, { "epoch": 0.68, "learning_rate": 1.72643137254902e-05, "loss": 0.4535, "step": 4360 }, { "epoch": 0.69, "learning_rate": 1.7258039215686274e-05, "loss": 0.3882, "step": 4370 }, { "epoch": 0.69, "learning_rate": 1.7251764705882355e-05, "loss": 0.5214, "step": 4380 }, { "epoch": 0.69, "learning_rate": 1.7245490196078433e-05, "loss": 0.4069, "step": 4390 }, { "epoch": 0.69, "learning_rate": 1.723921568627451e-05, "loss": 0.3989, "step": 4400 }, { "epoch": 0.69, "learning_rate": 1.723294117647059e-05, "loss": 0.3817, "step": 4410 }, { "epoch": 0.69, "learning_rate": 1.7226666666666665e-05, "loss": 0.3902, "step": 4420 }, { "epoch": 0.69, "learning_rate": 1.7220392156862747e-05, "loss": 0.341, "step": 4430 }, { "epoch": 0.7, "learning_rate": 1.7214117647058824e-05, "loss": 0.5587, "step": 4440 }, { "epoch": 0.7, "learning_rate": 1.7207843137254902e-05, "loss": 0.4374, "step": 4450 }, { "epoch": 0.7, "learning_rate": 1.7201568627450983e-05, "loss": 0.4623, "step": 4460 }, { "epoch": 0.7, "learning_rate": 1.719529411764706e-05, "loss": 0.3986, "step": 4470 }, { "epoch": 0.7, "learning_rate": 1.7189019607843138e-05, "loss": 0.6526, "step": 4480 }, { "epoch": 0.7, "learning_rate": 1.718274509803922e-05, "loss": 0.3689, "step": 4490 }, { "epoch": 0.71, "learning_rate": 1.7176470588235293e-05, "loss": 0.3612, "step": 4500 }, { "epoch": 0.71, "learning_rate": 1.7170196078431374e-05, "loss": 0.4184, "step": 4510 }, { "epoch": 0.71, "learning_rate": 1.7163921568627452e-05, "loss": 0.3404, "step": 4520 }, { "epoch": 0.71, "learning_rate": 1.715764705882353e-05, "loss": 0.3869, "step": 4530 }, { "epoch": 0.71, "learning_rate": 1.715137254901961e-05, "loss": 0.564, "step": 4540 }, { "epoch": 0.71, "learning_rate": 1.7145098039215688e-05, "loss": 0.2652, "step": 4550 }, { "epoch": 0.72, "learning_rate": 1.7138823529411766e-05, "loss": 0.5204, "step": 4560 }, { "epoch": 0.72, "learning_rate": 1.7132549019607843e-05, "loss": 0.3922, "step": 4570 }, { "epoch": 0.72, "learning_rate": 1.7126274509803924e-05, "loss": 0.3079, "step": 4580 }, { "epoch": 0.72, "learning_rate": 1.7120000000000002e-05, "loss": 0.3528, "step": 4590 }, { "epoch": 0.72, "learning_rate": 1.711372549019608e-05, "loss": 0.571, "step": 4600 }, { "epoch": 0.72, "learning_rate": 1.7107450980392157e-05, "loss": 0.4107, "step": 4610 }, { "epoch": 0.72, "learning_rate": 1.7101176470588235e-05, "loss": 0.6316, "step": 4620 }, { "epoch": 0.73, "learning_rate": 1.7094901960784316e-05, "loss": 0.4084, "step": 4630 }, { "epoch": 0.73, "learning_rate": 1.7088627450980394e-05, "loss": 0.4082, "step": 4640 }, { "epoch": 0.73, "learning_rate": 1.708235294117647e-05, "loss": 0.1875, "step": 4650 }, { "epoch": 0.73, "learning_rate": 1.7076078431372552e-05, "loss": 0.4833, "step": 4660 }, { "epoch": 0.73, "learning_rate": 1.706980392156863e-05, "loss": 0.3911, "step": 4670 }, { "epoch": 0.73, "learning_rate": 1.7063529411764708e-05, "loss": 0.2973, "step": 4680 }, { "epoch": 0.74, "learning_rate": 1.7057254901960785e-05, "loss": 0.4119, "step": 4690 }, { "epoch": 0.74, "learning_rate": 1.7050980392156863e-05, "loss": 0.5331, "step": 4700 }, { "epoch": 0.74, "learning_rate": 1.7044705882352944e-05, "loss": 0.5294, "step": 4710 }, { "epoch": 0.74, "learning_rate": 1.703843137254902e-05, "loss": 0.2866, "step": 4720 }, { "epoch": 0.74, "learning_rate": 1.70321568627451e-05, "loss": 0.5146, "step": 4730 }, { "epoch": 0.74, "learning_rate": 1.702588235294118e-05, "loss": 0.3898, "step": 4740 }, { "epoch": 0.75, "learning_rate": 1.7019607843137254e-05, "loss": 0.3546, "step": 4750 }, { "epoch": 0.75, "learning_rate": 1.7013333333333335e-05, "loss": 0.3873, "step": 4760 }, { "epoch": 0.75, "learning_rate": 1.7007058823529413e-05, "loss": 0.4679, "step": 4770 }, { "epoch": 0.75, "learning_rate": 1.700078431372549e-05, "loss": 0.4024, "step": 4780 }, { "epoch": 0.75, "learning_rate": 1.699450980392157e-05, "loss": 0.3911, "step": 4790 }, { "epoch": 0.75, "learning_rate": 1.698823529411765e-05, "loss": 0.4108, "step": 4800 }, { "epoch": 0.75, "learning_rate": 1.6981960784313727e-05, "loss": 0.4102, "step": 4810 }, { "epoch": 0.76, "learning_rate": 1.6975686274509804e-05, "loss": 0.5272, "step": 4820 }, { "epoch": 0.76, "learning_rate": 1.6969411764705882e-05, "loss": 0.3563, "step": 4830 }, { "epoch": 0.76, "learning_rate": 1.6963137254901963e-05, "loss": 0.6685, "step": 4840 }, { "epoch": 0.76, "learning_rate": 1.695686274509804e-05, "loss": 0.6152, "step": 4850 }, { "epoch": 0.76, "learning_rate": 1.695058823529412e-05, "loss": 0.412, "step": 4860 }, { "epoch": 0.76, "learning_rate": 1.69443137254902e-05, "loss": 0.4334, "step": 4870 }, { "epoch": 0.77, "learning_rate": 1.6938039215686274e-05, "loss": 0.3752, "step": 4880 }, { "epoch": 0.77, "learning_rate": 1.6931764705882355e-05, "loss": 0.3005, "step": 4890 }, { "epoch": 0.77, "learning_rate": 1.6925490196078432e-05, "loss": 0.4969, "step": 4900 }, { "epoch": 0.77, "learning_rate": 1.691921568627451e-05, "loss": 0.3894, "step": 4910 }, { "epoch": 0.77, "learning_rate": 1.691294117647059e-05, "loss": 0.478, "step": 4920 }, { "epoch": 0.77, "learning_rate": 1.690666666666667e-05, "loss": 0.3421, "step": 4930 }, { "epoch": 0.77, "learning_rate": 1.6900392156862746e-05, "loss": 0.4749, "step": 4940 }, { "epoch": 0.78, "learning_rate": 1.6894117647058824e-05, "loss": 0.5884, "step": 4950 }, { "epoch": 0.78, "learning_rate": 1.68878431372549e-05, "loss": 0.4297, "step": 4960 }, { "epoch": 0.78, "learning_rate": 1.6881568627450982e-05, "loss": 0.2578, "step": 4970 }, { "epoch": 0.78, "learning_rate": 1.687529411764706e-05, "loss": 0.615, "step": 4980 }, { "epoch": 0.78, "learning_rate": 1.6869019607843138e-05, "loss": 0.3926, "step": 4990 }, { "epoch": 0.78, "learning_rate": 1.686274509803922e-05, "loss": 0.3479, "step": 5000 }, { "epoch": 0.79, "learning_rate": 1.6856470588235296e-05, "loss": 0.4077, "step": 5010 }, { "epoch": 0.79, "learning_rate": 1.6850196078431374e-05, "loss": 0.3062, "step": 5020 }, { "epoch": 0.79, "learning_rate": 1.684392156862745e-05, "loss": 0.2684, "step": 5030 }, { "epoch": 0.79, "learning_rate": 1.6837647058823533e-05, "loss": 0.6147, "step": 5040 }, { "epoch": 0.79, "learning_rate": 1.683137254901961e-05, "loss": 0.4932, "step": 5050 }, { "epoch": 0.79, "learning_rate": 1.6825098039215688e-05, "loss": 0.5261, "step": 5060 }, { "epoch": 0.8, "learning_rate": 1.6818823529411766e-05, "loss": 0.1751, "step": 5070 }, { "epoch": 0.8, "learning_rate": 1.6812549019607843e-05, "loss": 0.5453, "step": 5080 }, { "epoch": 0.8, "learning_rate": 1.6806274509803924e-05, "loss": 0.4395, "step": 5090 }, { "epoch": 0.8, "learning_rate": 1.6800000000000002e-05, "loss": 0.4138, "step": 5100 }, { "epoch": 0.8, "learning_rate": 1.679372549019608e-05, "loss": 0.5219, "step": 5110 }, { "epoch": 0.8, "learning_rate": 1.678745098039216e-05, "loss": 0.2728, "step": 5120 }, { "epoch": 0.8, "learning_rate": 1.6781176470588235e-05, "loss": 0.4497, "step": 5130 }, { "epoch": 0.81, "learning_rate": 1.6774901960784316e-05, "loss": 0.5375, "step": 5140 }, { "epoch": 0.81, "learning_rate": 1.6768627450980393e-05, "loss": 0.4299, "step": 5150 }, { "epoch": 0.81, "learning_rate": 1.676235294117647e-05, "loss": 0.5703, "step": 5160 }, { "epoch": 0.81, "learning_rate": 1.6756078431372552e-05, "loss": 0.5133, "step": 5170 }, { "epoch": 0.81, "learning_rate": 1.674980392156863e-05, "loss": 0.301, "step": 5180 }, { "epoch": 0.81, "learning_rate": 1.6743529411764707e-05, "loss": 0.3641, "step": 5190 }, { "epoch": 0.82, "learning_rate": 1.6737254901960788e-05, "loss": 0.3872, "step": 5200 }, { "epoch": 0.82, "learning_rate": 1.6730980392156862e-05, "loss": 0.3031, "step": 5210 }, { "epoch": 0.82, "learning_rate": 1.6724705882352943e-05, "loss": 0.3072, "step": 5220 }, { "epoch": 0.82, "learning_rate": 1.671843137254902e-05, "loss": 0.5927, "step": 5230 }, { "epoch": 0.82, "learning_rate": 1.67121568627451e-05, "loss": 0.447, "step": 5240 }, { "epoch": 0.82, "learning_rate": 1.670588235294118e-05, "loss": 0.3549, "step": 5250 }, { "epoch": 0.83, "learning_rate": 1.6699607843137254e-05, "loss": 0.3566, "step": 5260 }, { "epoch": 0.83, "learning_rate": 1.6693333333333335e-05, "loss": 0.3702, "step": 5270 }, { "epoch": 0.83, "learning_rate": 1.6687058823529413e-05, "loss": 0.3575, "step": 5280 }, { "epoch": 0.83, "learning_rate": 1.668078431372549e-05, "loss": 0.4595, "step": 5290 }, { "epoch": 0.83, "learning_rate": 1.667450980392157e-05, "loss": 0.4245, "step": 5300 }, { "epoch": 0.83, "learning_rate": 1.666823529411765e-05, "loss": 0.4454, "step": 5310 }, { "epoch": 0.83, "learning_rate": 1.6661960784313727e-05, "loss": 0.4345, "step": 5320 }, { "epoch": 0.84, "learning_rate": 1.6655686274509804e-05, "loss": 0.4295, "step": 5330 }, { "epoch": 0.84, "learning_rate": 1.6649411764705882e-05, "loss": 0.3648, "step": 5340 }, { "epoch": 0.84, "learning_rate": 1.6643137254901963e-05, "loss": 0.3556, "step": 5350 }, { "epoch": 0.84, "learning_rate": 1.663686274509804e-05, "loss": 0.4208, "step": 5360 }, { "epoch": 0.84, "learning_rate": 1.6630588235294118e-05, "loss": 0.5361, "step": 5370 }, { "epoch": 0.84, "learning_rate": 1.66243137254902e-05, "loss": 0.6536, "step": 5380 }, { "epoch": 0.85, "learning_rate": 1.6618039215686277e-05, "loss": 0.3513, "step": 5390 }, { "epoch": 0.85, "learning_rate": 1.6611764705882354e-05, "loss": 0.4162, "step": 5400 }, { "epoch": 0.85, "learning_rate": 1.6605490196078432e-05, "loss": 0.336, "step": 5410 }, { "epoch": 0.85, "learning_rate": 1.659921568627451e-05, "loss": 0.5179, "step": 5420 }, { "epoch": 0.85, "learning_rate": 1.659294117647059e-05, "loss": 0.401, "step": 5430 }, { "epoch": 0.85, "learning_rate": 1.6586666666666668e-05, "loss": 0.4557, "step": 5440 }, { "epoch": 0.85, "learning_rate": 1.6580392156862746e-05, "loss": 0.4171, "step": 5450 }, { "epoch": 0.86, "learning_rate": 1.6574117647058823e-05, "loss": 0.5159, "step": 5460 }, { "epoch": 0.86, "learning_rate": 1.6567843137254905e-05, "loss": 0.3702, "step": 5470 }, { "epoch": 0.86, "learning_rate": 1.6561568627450982e-05, "loss": 0.4609, "step": 5480 }, { "epoch": 0.86, "learning_rate": 1.655529411764706e-05, "loss": 0.2787, "step": 5490 }, { "epoch": 0.86, "learning_rate": 1.6549019607843137e-05, "loss": 0.3137, "step": 5500 }, { "epoch": 0.86, "learning_rate": 1.654274509803922e-05, "loss": 0.5684, "step": 5510 }, { "epoch": 0.87, "learning_rate": 1.6536470588235296e-05, "loss": 0.2612, "step": 5520 }, { "epoch": 0.87, "learning_rate": 1.6530196078431374e-05, "loss": 0.4977, "step": 5530 }, { "epoch": 0.87, "learning_rate": 1.652392156862745e-05, "loss": 0.4448, "step": 5540 }, { "epoch": 0.87, "learning_rate": 1.6517647058823532e-05, "loss": 0.4651, "step": 5550 }, { "epoch": 0.87, "learning_rate": 1.651137254901961e-05, "loss": 0.3953, "step": 5560 }, { "epoch": 0.87, "learning_rate": 1.6505098039215688e-05, "loss": 0.2852, "step": 5570 }, { "epoch": 0.88, "learning_rate": 1.649882352941177e-05, "loss": 0.1827, "step": 5580 }, { "epoch": 0.88, "learning_rate": 1.6492549019607843e-05, "loss": 0.6388, "step": 5590 }, { "epoch": 0.88, "learning_rate": 1.6486274509803924e-05, "loss": 0.3965, "step": 5600 }, { "epoch": 0.88, "learning_rate": 1.648e-05, "loss": 0.6014, "step": 5610 }, { "epoch": 0.88, "learning_rate": 1.647372549019608e-05, "loss": 0.4963, "step": 5620 }, { "epoch": 0.88, "learning_rate": 1.646745098039216e-05, "loss": 0.5413, "step": 5630 }, { "epoch": 0.88, "learning_rate": 1.6461176470588234e-05, "loss": 0.2688, "step": 5640 }, { "epoch": 0.89, "learning_rate": 1.6454901960784315e-05, "loss": 0.6082, "step": 5650 }, { "epoch": 0.89, "learning_rate": 1.6448627450980393e-05, "loss": 0.2162, "step": 5660 }, { "epoch": 0.89, "learning_rate": 1.644235294117647e-05, "loss": 0.7025, "step": 5670 }, { "epoch": 0.89, "learning_rate": 1.643607843137255e-05, "loss": 0.3871, "step": 5680 }, { "epoch": 0.89, "learning_rate": 1.642980392156863e-05, "loss": 0.4952, "step": 5690 }, { "epoch": 0.89, "learning_rate": 1.6423529411764707e-05, "loss": 0.3591, "step": 5700 }, { "epoch": 0.9, "learning_rate": 1.6417254901960788e-05, "loss": 0.3933, "step": 5710 }, { "epoch": 0.9, "learning_rate": 1.6410980392156862e-05, "loss": 0.3182, "step": 5720 }, { "epoch": 0.9, "learning_rate": 1.6404705882352943e-05, "loss": 0.4916, "step": 5730 }, { "epoch": 0.9, "learning_rate": 1.639843137254902e-05, "loss": 0.4413, "step": 5740 }, { "epoch": 0.9, "learning_rate": 1.63921568627451e-05, "loss": 0.3632, "step": 5750 }, { "epoch": 0.9, "learning_rate": 1.638588235294118e-05, "loss": 0.5596, "step": 5760 }, { "epoch": 0.91, "learning_rate": 1.6379607843137254e-05, "loss": 0.443, "step": 5770 }, { "epoch": 0.91, "learning_rate": 1.6373333333333335e-05, "loss": 0.8104, "step": 5780 }, { "epoch": 0.91, "learning_rate": 1.6367058823529412e-05, "loss": 0.4071, "step": 5790 }, { "epoch": 0.91, "learning_rate": 1.636078431372549e-05, "loss": 0.3295, "step": 5800 }, { "epoch": 0.91, "learning_rate": 1.635450980392157e-05, "loss": 0.4213, "step": 5810 }, { "epoch": 0.91, "learning_rate": 1.634823529411765e-05, "loss": 0.3215, "step": 5820 }, { "epoch": 0.91, "learning_rate": 1.6341960784313726e-05, "loss": 0.2585, "step": 5830 }, { "epoch": 0.92, "learning_rate": 1.6335686274509804e-05, "loss": 0.3271, "step": 5840 }, { "epoch": 0.92, "learning_rate": 1.6329411764705885e-05, "loss": 0.4879, "step": 5850 }, { "epoch": 0.92, "learning_rate": 1.6323137254901962e-05, "loss": 0.4661, "step": 5860 }, { "epoch": 0.92, "learning_rate": 1.631686274509804e-05, "loss": 0.5202, "step": 5870 }, { "epoch": 0.92, "learning_rate": 1.6310588235294118e-05, "loss": 0.5641, "step": 5880 }, { "epoch": 0.92, "learning_rate": 1.63043137254902e-05, "loss": 0.5118, "step": 5890 }, { "epoch": 0.93, "learning_rate": 1.6298039215686276e-05, "loss": 0.1979, "step": 5900 }, { "epoch": 0.93, "learning_rate": 1.6291764705882354e-05, "loss": 0.5421, "step": 5910 }, { "epoch": 0.93, "learning_rate": 1.628549019607843e-05, "loss": 0.4006, "step": 5920 }, { "epoch": 0.93, "learning_rate": 1.6279215686274513e-05, "loss": 0.4823, "step": 5930 }, { "epoch": 0.93, "learning_rate": 1.627294117647059e-05, "loss": 0.3473, "step": 5940 }, { "epoch": 0.93, "learning_rate": 1.6266666666666668e-05, "loss": 0.4043, "step": 5950 }, { "epoch": 0.93, "learning_rate": 1.6260392156862746e-05, "loss": 0.7877, "step": 5960 }, { "epoch": 0.94, "learning_rate": 1.6254117647058823e-05, "loss": 0.3899, "step": 5970 }, { "epoch": 0.94, "learning_rate": 1.6247843137254904e-05, "loss": 0.4688, "step": 5980 }, { "epoch": 0.94, "learning_rate": 1.6241568627450982e-05, "loss": 0.4568, "step": 5990 }, { "epoch": 0.94, "learning_rate": 1.623529411764706e-05, "loss": 0.5864, "step": 6000 }, { "epoch": 0.94, "learning_rate": 1.622901960784314e-05, "loss": 0.4957, "step": 6010 }, { "epoch": 0.94, "learning_rate": 1.6222745098039218e-05, "loss": 0.4766, "step": 6020 }, { "epoch": 0.95, "learning_rate": 1.6216470588235296e-05, "loss": 0.2929, "step": 6030 }, { "epoch": 0.95, "learning_rate": 1.6210196078431373e-05, "loss": 0.3014, "step": 6040 }, { "epoch": 0.95, "learning_rate": 1.620392156862745e-05, "loss": 0.2467, "step": 6050 }, { "epoch": 0.95, "learning_rate": 1.6197647058823532e-05, "loss": 0.4922, "step": 6060 }, { "epoch": 0.95, "learning_rate": 1.619137254901961e-05, "loss": 0.4507, "step": 6070 }, { "epoch": 0.95, "learning_rate": 1.6185098039215687e-05, "loss": 0.5948, "step": 6080 }, { "epoch": 0.96, "learning_rate": 1.6178823529411768e-05, "loss": 0.5046, "step": 6090 }, { "epoch": 0.96, "learning_rate": 1.6172549019607842e-05, "loss": 0.3209, "step": 6100 }, { "epoch": 0.96, "learning_rate": 1.6166274509803924e-05, "loss": 0.4981, "step": 6110 }, { "epoch": 0.96, "learning_rate": 1.616e-05, "loss": 0.4509, "step": 6120 }, { "epoch": 0.96, "learning_rate": 1.615372549019608e-05, "loss": 0.3948, "step": 6130 }, { "epoch": 0.96, "learning_rate": 1.614745098039216e-05, "loss": 0.3119, "step": 6140 }, { "epoch": 0.96, "learning_rate": 1.6141176470588234e-05, "loss": 0.3784, "step": 6150 }, { "epoch": 0.97, "learning_rate": 1.6134901960784315e-05, "loss": 0.522, "step": 6160 }, { "epoch": 0.97, "learning_rate": 1.6128627450980393e-05, "loss": 0.5347, "step": 6170 }, { "epoch": 0.97, "learning_rate": 1.612235294117647e-05, "loss": 0.3549, "step": 6180 }, { "epoch": 0.97, "learning_rate": 1.611607843137255e-05, "loss": 0.2802, "step": 6190 }, { "epoch": 0.97, "learning_rate": 1.610980392156863e-05, "loss": 0.2526, "step": 6200 }, { "epoch": 0.97, "learning_rate": 1.6103529411764707e-05, "loss": 0.5391, "step": 6210 }, { "epoch": 0.98, "learning_rate": 1.6097254901960788e-05, "loss": 0.2805, "step": 6220 }, { "epoch": 0.98, "learning_rate": 1.6090980392156862e-05, "loss": 0.2797, "step": 6230 }, { "epoch": 0.98, "learning_rate": 1.6084705882352943e-05, "loss": 0.4103, "step": 6240 }, { "epoch": 0.98, "learning_rate": 1.607843137254902e-05, "loss": 0.2979, "step": 6250 }, { "epoch": 0.98, "learning_rate": 1.6072156862745098e-05, "loss": 0.4344, "step": 6260 }, { "epoch": 0.98, "learning_rate": 1.606588235294118e-05, "loss": 0.3211, "step": 6270 }, { "epoch": 0.99, "learning_rate": 1.6059607843137257e-05, "loss": 0.3504, "step": 6280 }, { "epoch": 0.99, "learning_rate": 1.6053333333333334e-05, "loss": 0.2593, "step": 6290 }, { "epoch": 0.99, "learning_rate": 1.6047058823529412e-05, "loss": 0.3661, "step": 6300 }, { "epoch": 0.99, "learning_rate": 1.6040784313725493e-05, "loss": 0.467, "step": 6310 }, { "epoch": 0.99, "learning_rate": 1.603450980392157e-05, "loss": 0.4271, "step": 6320 }, { "epoch": 0.99, "learning_rate": 1.6028235294117648e-05, "loss": 0.1932, "step": 6330 }, { "epoch": 0.99, "learning_rate": 1.6021960784313726e-05, "loss": 0.2795, "step": 6340 }, { "epoch": 1.0, "learning_rate": 1.6015686274509804e-05, "loss": 0.3387, "step": 6350 }, { "epoch": 1.0, "learning_rate": 1.6009411764705885e-05, "loss": 0.4764, "step": 6360 }, { "epoch": 1.0, "learning_rate": 1.6003137254901962e-05, "loss": 0.3215, "step": 6370 }, { "epoch": 1.0, "eval_accuracy": 0.9855555555555555, "eval_loss": 0.0630427896976471, "eval_runtime": 195.1788, "eval_samples_per_second": 46.112, "eval_steps_per_second": 5.764, "step": 6375 }, { "epoch": 1.0, "learning_rate": 1.599686274509804e-05, "loss": 0.3041, "step": 6380 }, { "epoch": 1.0, "learning_rate": 1.599058823529412e-05, "loss": 0.5059, "step": 6390 }, { "epoch": 1.0, "learning_rate": 1.59843137254902e-05, "loss": 0.4157, "step": 6400 }, { "epoch": 1.01, "learning_rate": 1.5978039215686276e-05, "loss": 0.5571, "step": 6410 }, { "epoch": 1.01, "learning_rate": 1.5971764705882354e-05, "loss": 0.2575, "step": 6420 }, { "epoch": 1.01, "learning_rate": 1.596549019607843e-05, "loss": 0.535, "step": 6430 }, { "epoch": 1.01, "learning_rate": 1.5959215686274512e-05, "loss": 0.3776, "step": 6440 }, { "epoch": 1.01, "learning_rate": 1.595294117647059e-05, "loss": 0.4385, "step": 6450 }, { "epoch": 1.01, "learning_rate": 1.5946666666666668e-05, "loss": 0.3886, "step": 6460 }, { "epoch": 1.01, "learning_rate": 1.594039215686275e-05, "loss": 0.3589, "step": 6470 }, { "epoch": 1.02, "learning_rate": 1.5934117647058823e-05, "loss": 0.4019, "step": 6480 }, { "epoch": 1.02, "learning_rate": 1.5927843137254904e-05, "loss": 0.4696, "step": 6490 }, { "epoch": 1.02, "learning_rate": 1.592156862745098e-05, "loss": 0.4431, "step": 6500 }, { "epoch": 1.02, "learning_rate": 1.591529411764706e-05, "loss": 0.3013, "step": 6510 }, { "epoch": 1.02, "learning_rate": 1.590901960784314e-05, "loss": 0.173, "step": 6520 }, { "epoch": 1.02, "learning_rate": 1.5902745098039218e-05, "loss": 0.4556, "step": 6530 }, { "epoch": 1.03, "learning_rate": 1.5896470588235295e-05, "loss": 0.4613, "step": 6540 }, { "epoch": 1.03, "learning_rate": 1.5890196078431373e-05, "loss": 0.3366, "step": 6550 }, { "epoch": 1.03, "learning_rate": 1.588392156862745e-05, "loss": 0.5419, "step": 6560 }, { "epoch": 1.03, "learning_rate": 1.587764705882353e-05, "loss": 0.4957, "step": 6570 }, { "epoch": 1.03, "learning_rate": 1.587137254901961e-05, "loss": 0.5506, "step": 6580 }, { "epoch": 1.03, "learning_rate": 1.5865098039215687e-05, "loss": 0.5984, "step": 6590 }, { "epoch": 1.04, "learning_rate": 1.5858823529411768e-05, "loss": 0.3397, "step": 6600 }, { "epoch": 1.04, "learning_rate": 1.5852549019607842e-05, "loss": 0.4229, "step": 6610 }, { "epoch": 1.04, "learning_rate": 1.5846274509803923e-05, "loss": 0.345, "step": 6620 }, { "epoch": 1.04, "learning_rate": 1.584e-05, "loss": 0.3988, "step": 6630 }, { "epoch": 1.04, "learning_rate": 1.583372549019608e-05, "loss": 0.3384, "step": 6640 }, { "epoch": 1.04, "learning_rate": 1.582745098039216e-05, "loss": 0.359, "step": 6650 }, { "epoch": 1.04, "learning_rate": 1.5821176470588237e-05, "loss": 0.3325, "step": 6660 }, { "epoch": 1.05, "learning_rate": 1.5814901960784315e-05, "loss": 0.314, "step": 6670 }, { "epoch": 1.05, "learning_rate": 1.5808627450980392e-05, "loss": 0.3733, "step": 6680 }, { "epoch": 1.05, "learning_rate": 1.580235294117647e-05, "loss": 0.3034, "step": 6690 }, { "epoch": 1.05, "learning_rate": 1.579607843137255e-05, "loss": 0.3849, "step": 6700 }, { "epoch": 1.05, "learning_rate": 1.578980392156863e-05, "loss": 0.4868, "step": 6710 }, { "epoch": 1.05, "learning_rate": 1.5783529411764706e-05, "loss": 0.441, "step": 6720 }, { "epoch": 1.06, "learning_rate": 1.5777254901960787e-05, "loss": 0.5153, "step": 6730 }, { "epoch": 1.06, "learning_rate": 1.5770980392156865e-05, "loss": 0.3287, "step": 6740 }, { "epoch": 1.06, "learning_rate": 1.5764705882352943e-05, "loss": 0.5351, "step": 6750 }, { "epoch": 1.06, "learning_rate": 1.575843137254902e-05, "loss": 0.3428, "step": 6760 }, { "epoch": 1.06, "learning_rate": 1.5752156862745098e-05, "loss": 0.3315, "step": 6770 }, { "epoch": 1.06, "learning_rate": 1.574588235294118e-05, "loss": 0.3392, "step": 6780 }, { "epoch": 1.07, "learning_rate": 1.5739607843137256e-05, "loss": 0.4041, "step": 6790 }, { "epoch": 1.07, "learning_rate": 1.5733333333333334e-05, "loss": 0.4932, "step": 6800 }, { "epoch": 1.07, "learning_rate": 1.572705882352941e-05, "loss": 0.3889, "step": 6810 }, { "epoch": 1.07, "learning_rate": 1.5720784313725493e-05, "loss": 0.6592, "step": 6820 }, { "epoch": 1.07, "learning_rate": 1.571450980392157e-05, "loss": 0.2555, "step": 6830 }, { "epoch": 1.07, "learning_rate": 1.5708235294117648e-05, "loss": 0.3131, "step": 6840 }, { "epoch": 1.07, "learning_rate": 1.570196078431373e-05, "loss": 0.427, "step": 6850 }, { "epoch": 1.08, "learning_rate": 1.5695686274509803e-05, "loss": 0.3293, "step": 6860 }, { "epoch": 1.08, "learning_rate": 1.5689411764705884e-05, "loss": 0.2734, "step": 6870 }, { "epoch": 1.08, "learning_rate": 1.5683137254901962e-05, "loss": 0.5089, "step": 6880 }, { "epoch": 1.08, "learning_rate": 1.567686274509804e-05, "loss": 0.3836, "step": 6890 }, { "epoch": 1.08, "learning_rate": 1.567058823529412e-05, "loss": 0.4281, "step": 6900 }, { "epoch": 1.08, "learning_rate": 1.5664313725490198e-05, "loss": 0.4322, "step": 6910 }, { "epoch": 1.09, "learning_rate": 1.5658039215686276e-05, "loss": 0.6048, "step": 6920 }, { "epoch": 1.09, "learning_rate": 1.5651764705882357e-05, "loss": 0.2903, "step": 6930 }, { "epoch": 1.09, "learning_rate": 1.564549019607843e-05, "loss": 0.2281, "step": 6940 }, { "epoch": 1.09, "learning_rate": 1.5639215686274512e-05, "loss": 0.5197, "step": 6950 }, { "epoch": 1.09, "learning_rate": 1.563294117647059e-05, "loss": 0.4961, "step": 6960 }, { "epoch": 1.09, "learning_rate": 1.5626666666666667e-05, "loss": 0.3651, "step": 6970 }, { "epoch": 1.09, "learning_rate": 1.5620392156862748e-05, "loss": 0.3695, "step": 6980 }, { "epoch": 1.1, "learning_rate": 1.5614117647058823e-05, "loss": 0.3021, "step": 6990 }, { "epoch": 1.1, "learning_rate": 1.5607843137254904e-05, "loss": 0.3306, "step": 7000 }, { "epoch": 1.1, "learning_rate": 1.560156862745098e-05, "loss": 0.3024, "step": 7010 }, { "epoch": 1.1, "learning_rate": 1.559529411764706e-05, "loss": 0.4151, "step": 7020 }, { "epoch": 1.1, "learning_rate": 1.558901960784314e-05, "loss": 0.1288, "step": 7030 }, { "epoch": 1.1, "learning_rate": 1.5582745098039217e-05, "loss": 0.4721, "step": 7040 }, { "epoch": 1.11, "learning_rate": 1.5576470588235295e-05, "loss": 0.4818, "step": 7050 }, { "epoch": 1.11, "learning_rate": 1.5570196078431373e-05, "loss": 0.3279, "step": 7060 }, { "epoch": 1.11, "learning_rate": 1.556392156862745e-05, "loss": 0.3983, "step": 7070 }, { "epoch": 1.11, "learning_rate": 1.555764705882353e-05, "loss": 0.3446, "step": 7080 }, { "epoch": 1.11, "learning_rate": 1.555137254901961e-05, "loss": 0.2286, "step": 7090 }, { "epoch": 1.11, "learning_rate": 1.5545098039215687e-05, "loss": 0.2466, "step": 7100 }, { "epoch": 1.12, "learning_rate": 1.5538823529411768e-05, "loss": 0.7458, "step": 7110 }, { "epoch": 1.12, "learning_rate": 1.5532549019607845e-05, "loss": 0.3632, "step": 7120 }, { "epoch": 1.12, "learning_rate": 1.5526274509803923e-05, "loss": 0.1974, "step": 7130 }, { "epoch": 1.12, "learning_rate": 1.552e-05, "loss": 0.5814, "step": 7140 }, { "epoch": 1.12, "learning_rate": 1.5513725490196078e-05, "loss": 0.5306, "step": 7150 }, { "epoch": 1.12, "learning_rate": 1.550745098039216e-05, "loss": 0.3181, "step": 7160 }, { "epoch": 1.12, "learning_rate": 1.5501176470588237e-05, "loss": 0.3923, "step": 7170 }, { "epoch": 1.13, "learning_rate": 1.5494901960784314e-05, "loss": 0.3909, "step": 7180 }, { "epoch": 1.13, "learning_rate": 1.5488627450980392e-05, "loss": 0.3897, "step": 7190 }, { "epoch": 1.13, "learning_rate": 1.5482352941176473e-05, "loss": 0.4475, "step": 7200 }, { "epoch": 1.13, "learning_rate": 1.547607843137255e-05, "loss": 0.5448, "step": 7210 }, { "epoch": 1.13, "learning_rate": 1.5469803921568628e-05, "loss": 0.4473, "step": 7220 }, { "epoch": 1.13, "learning_rate": 1.5463529411764706e-05, "loss": 0.406, "step": 7230 }, { "epoch": 1.14, "learning_rate": 1.5457254901960787e-05, "loss": 0.2976, "step": 7240 }, { "epoch": 1.14, "learning_rate": 1.5450980392156865e-05, "loss": 0.3248, "step": 7250 }, { "epoch": 1.14, "learning_rate": 1.5444705882352942e-05, "loss": 0.2822, "step": 7260 }, { "epoch": 1.14, "learning_rate": 1.543843137254902e-05, "loss": 0.4786, "step": 7270 }, { "epoch": 1.14, "learning_rate": 1.54321568627451e-05, "loss": 0.4543, "step": 7280 }, { "epoch": 1.14, "learning_rate": 1.542588235294118e-05, "loss": 0.4149, "step": 7290 }, { "epoch": 1.15, "learning_rate": 1.5419607843137256e-05, "loss": 0.2612, "step": 7300 }, { "epoch": 1.15, "learning_rate": 1.5413333333333337e-05, "loss": 0.3975, "step": 7310 }, { "epoch": 1.15, "learning_rate": 1.540705882352941e-05, "loss": 0.4275, "step": 7320 }, { "epoch": 1.15, "learning_rate": 1.5400784313725492e-05, "loss": 0.2669, "step": 7330 }, { "epoch": 1.15, "learning_rate": 1.539450980392157e-05, "loss": 0.3457, "step": 7340 }, { "epoch": 1.15, "learning_rate": 1.5388235294117648e-05, "loss": 0.2881, "step": 7350 }, { "epoch": 1.15, "learning_rate": 1.538196078431373e-05, "loss": 0.5251, "step": 7360 }, { "epoch": 1.16, "learning_rate": 1.5375686274509803e-05, "loss": 0.4257, "step": 7370 }, { "epoch": 1.16, "learning_rate": 1.5369411764705884e-05, "loss": 0.557, "step": 7380 }, { "epoch": 1.16, "learning_rate": 1.536313725490196e-05, "loss": 0.3768, "step": 7390 }, { "epoch": 1.16, "learning_rate": 1.535686274509804e-05, "loss": 0.3895, "step": 7400 }, { "epoch": 1.16, "learning_rate": 1.535058823529412e-05, "loss": 0.324, "step": 7410 }, { "epoch": 1.16, "learning_rate": 1.5344313725490198e-05, "loss": 0.3413, "step": 7420 }, { "epoch": 1.17, "learning_rate": 1.5338039215686275e-05, "loss": 0.5086, "step": 7430 }, { "epoch": 1.17, "learning_rate": 1.5331764705882356e-05, "loss": 0.4101, "step": 7440 }, { "epoch": 1.17, "learning_rate": 1.532549019607843e-05, "loss": 0.243, "step": 7450 }, { "epoch": 1.17, "learning_rate": 1.531921568627451e-05, "loss": 0.38, "step": 7460 }, { "epoch": 1.17, "learning_rate": 1.531294117647059e-05, "loss": 0.311, "step": 7470 }, { "epoch": 1.17, "learning_rate": 1.5306666666666667e-05, "loss": 0.4682, "step": 7480 }, { "epoch": 1.17, "learning_rate": 1.5300392156862748e-05, "loss": 0.6407, "step": 7490 }, { "epoch": 1.18, "learning_rate": 1.5294117647058822e-05, "loss": 0.4661, "step": 7500 }, { "epoch": 1.18, "learning_rate": 1.5287843137254903e-05, "loss": 0.3103, "step": 7510 }, { "epoch": 1.18, "learning_rate": 1.528156862745098e-05, "loss": 0.2927, "step": 7520 }, { "epoch": 1.18, "learning_rate": 1.527529411764706e-05, "loss": 0.3163, "step": 7530 }, { "epoch": 1.18, "learning_rate": 1.526901960784314e-05, "loss": 0.2846, "step": 7540 }, { "epoch": 1.18, "learning_rate": 1.5262745098039217e-05, "loss": 0.4317, "step": 7550 }, { "epoch": 1.19, "learning_rate": 1.5256470588235295e-05, "loss": 0.3564, "step": 7560 }, { "epoch": 1.19, "learning_rate": 1.5250196078431374e-05, "loss": 0.3321, "step": 7570 }, { "epoch": 1.19, "learning_rate": 1.5243921568627453e-05, "loss": 0.4838, "step": 7580 }, { "epoch": 1.19, "learning_rate": 1.5237647058823531e-05, "loss": 0.6571, "step": 7590 }, { "epoch": 1.19, "learning_rate": 1.523137254901961e-05, "loss": 0.5685, "step": 7600 }, { "epoch": 1.19, "learning_rate": 1.5225098039215686e-05, "loss": 0.4965, "step": 7610 }, { "epoch": 1.2, "learning_rate": 1.5218823529411766e-05, "loss": 0.4268, "step": 7620 }, { "epoch": 1.2, "learning_rate": 1.5212549019607845e-05, "loss": 0.4265, "step": 7630 }, { "epoch": 1.2, "learning_rate": 1.5206274509803923e-05, "loss": 0.4997, "step": 7640 }, { "epoch": 1.2, "learning_rate": 1.5200000000000002e-05, "loss": 0.287, "step": 7650 }, { "epoch": 1.2, "learning_rate": 1.5193725490196081e-05, "loss": 0.1912, "step": 7660 }, { "epoch": 1.2, "learning_rate": 1.5187450980392157e-05, "loss": 0.5219, "step": 7670 }, { "epoch": 1.2, "learning_rate": 1.5181176470588236e-05, "loss": 0.4263, "step": 7680 }, { "epoch": 1.21, "learning_rate": 1.5174901960784314e-05, "loss": 0.4353, "step": 7690 }, { "epoch": 1.21, "learning_rate": 1.5168627450980393e-05, "loss": 0.3218, "step": 7700 }, { "epoch": 1.21, "learning_rate": 1.5162352941176473e-05, "loss": 0.3203, "step": 7710 }, { "epoch": 1.21, "learning_rate": 1.515607843137255e-05, "loss": 0.2629, "step": 7720 }, { "epoch": 1.21, "learning_rate": 1.5149803921568628e-05, "loss": 0.468, "step": 7730 }, { "epoch": 1.21, "learning_rate": 1.5143529411764707e-05, "loss": 0.3244, "step": 7740 }, { "epoch": 1.22, "learning_rate": 1.5137254901960785e-05, "loss": 0.4206, "step": 7750 }, { "epoch": 1.22, "learning_rate": 1.5130980392156864e-05, "loss": 0.434, "step": 7760 }, { "epoch": 1.22, "learning_rate": 1.5124705882352942e-05, "loss": 0.3678, "step": 7770 }, { "epoch": 1.22, "learning_rate": 1.5118431372549021e-05, "loss": 0.6026, "step": 7780 }, { "epoch": 1.22, "learning_rate": 1.51121568627451e-05, "loss": 0.5508, "step": 7790 }, { "epoch": 1.22, "learning_rate": 1.5105882352941176e-05, "loss": 0.4021, "step": 7800 }, { "epoch": 1.23, "learning_rate": 1.5099607843137256e-05, "loss": 0.2723, "step": 7810 }, { "epoch": 1.23, "learning_rate": 1.5093333333333335e-05, "loss": 0.3753, "step": 7820 }, { "epoch": 1.23, "learning_rate": 1.5087058823529413e-05, "loss": 0.4063, "step": 7830 }, { "epoch": 1.23, "learning_rate": 1.5080784313725492e-05, "loss": 0.3289, "step": 7840 }, { "epoch": 1.23, "learning_rate": 1.5074509803921571e-05, "loss": 0.3467, "step": 7850 }, { "epoch": 1.23, "learning_rate": 1.5068235294117647e-05, "loss": 0.4548, "step": 7860 }, { "epoch": 1.23, "learning_rate": 1.5061960784313727e-05, "loss": 0.4324, "step": 7870 }, { "epoch": 1.24, "learning_rate": 1.5055686274509804e-05, "loss": 0.2672, "step": 7880 }, { "epoch": 1.24, "learning_rate": 1.5049411764705884e-05, "loss": 0.4862, "step": 7890 }, { "epoch": 1.24, "learning_rate": 1.5043137254901963e-05, "loss": 0.2526, "step": 7900 }, { "epoch": 1.24, "learning_rate": 1.503686274509804e-05, "loss": 0.5486, "step": 7910 }, { "epoch": 1.24, "learning_rate": 1.503058823529412e-05, "loss": 0.4179, "step": 7920 }, { "epoch": 1.24, "learning_rate": 1.5024313725490197e-05, "loss": 0.4983, "step": 7930 }, { "epoch": 1.25, "learning_rate": 1.5018039215686275e-05, "loss": 0.2445, "step": 7940 }, { "epoch": 1.25, "learning_rate": 1.5011764705882354e-05, "loss": 0.373, "step": 7950 }, { "epoch": 1.25, "learning_rate": 1.5005490196078432e-05, "loss": 0.2499, "step": 7960 }, { "epoch": 1.25, "learning_rate": 1.4999215686274511e-05, "loss": 0.3575, "step": 7970 }, { "epoch": 1.25, "learning_rate": 1.499294117647059e-05, "loss": 0.4701, "step": 7980 }, { "epoch": 1.25, "learning_rate": 1.4986666666666667e-05, "loss": 0.5049, "step": 7990 }, { "epoch": 1.25, "learning_rate": 1.4980392156862746e-05, "loss": 0.4469, "step": 8000 }, { "epoch": 1.26, "learning_rate": 1.4974117647058825e-05, "loss": 0.3214, "step": 8010 }, { "epoch": 1.26, "learning_rate": 1.4967843137254903e-05, "loss": 0.2661, "step": 8020 }, { "epoch": 1.26, "learning_rate": 1.4961568627450982e-05, "loss": 0.3075, "step": 8030 }, { "epoch": 1.26, "learning_rate": 1.4955294117647058e-05, "loss": 0.3521, "step": 8040 }, { "epoch": 1.26, "learning_rate": 1.4949019607843137e-05, "loss": 0.3413, "step": 8050 }, { "epoch": 1.26, "learning_rate": 1.4942745098039217e-05, "loss": 0.504, "step": 8060 }, { "epoch": 1.27, "learning_rate": 1.4936470588235294e-05, "loss": 0.2539, "step": 8070 }, { "epoch": 1.27, "learning_rate": 1.4930196078431374e-05, "loss": 0.2342, "step": 8080 }, { "epoch": 1.27, "learning_rate": 1.4923921568627453e-05, "loss": 0.5391, "step": 8090 }, { "epoch": 1.27, "learning_rate": 1.491764705882353e-05, "loss": 0.3094, "step": 8100 }, { "epoch": 1.27, "learning_rate": 1.491137254901961e-05, "loss": 0.4582, "step": 8110 }, { "epoch": 1.27, "learning_rate": 1.490509803921569e-05, "loss": 0.5282, "step": 8120 }, { "epoch": 1.28, "learning_rate": 1.4898823529411765e-05, "loss": 0.3356, "step": 8130 }, { "epoch": 1.28, "learning_rate": 1.4892549019607845e-05, "loss": 0.4054, "step": 8140 }, { "epoch": 1.28, "learning_rate": 1.4886274509803922e-05, "loss": 0.3075, "step": 8150 }, { "epoch": 1.28, "learning_rate": 1.4880000000000002e-05, "loss": 0.2435, "step": 8160 }, { "epoch": 1.28, "learning_rate": 1.4873725490196081e-05, "loss": 0.3527, "step": 8170 }, { "epoch": 1.28, "learning_rate": 1.4867450980392157e-05, "loss": 0.4457, "step": 8180 }, { "epoch": 1.28, "learning_rate": 1.4861176470588236e-05, "loss": 0.4885, "step": 8190 }, { "epoch": 1.29, "learning_rate": 1.4854901960784315e-05, "loss": 0.3724, "step": 8200 }, { "epoch": 1.29, "learning_rate": 1.4848627450980393e-05, "loss": 0.3174, "step": 8210 }, { "epoch": 1.29, "learning_rate": 1.4842352941176472e-05, "loss": 0.255, "step": 8220 }, { "epoch": 1.29, "learning_rate": 1.483607843137255e-05, "loss": 0.3772, "step": 8230 }, { "epoch": 1.29, "learning_rate": 1.4829803921568628e-05, "loss": 0.3247, "step": 8240 }, { "epoch": 1.29, "learning_rate": 1.4823529411764707e-05, "loss": 0.4215, "step": 8250 }, { "epoch": 1.3, "learning_rate": 1.4817254901960785e-05, "loss": 0.3202, "step": 8260 }, { "epoch": 1.3, "learning_rate": 1.4810980392156864e-05, "loss": 0.3516, "step": 8270 }, { "epoch": 1.3, "learning_rate": 1.4804705882352943e-05, "loss": 0.3415, "step": 8280 }, { "epoch": 1.3, "learning_rate": 1.4798431372549021e-05, "loss": 0.2825, "step": 8290 }, { "epoch": 1.3, "learning_rate": 1.47921568627451e-05, "loss": 0.5494, "step": 8300 }, { "epoch": 1.3, "learning_rate": 1.4785882352941176e-05, "loss": 0.461, "step": 8310 }, { "epoch": 1.31, "learning_rate": 1.4779607843137255e-05, "loss": 0.2832, "step": 8320 }, { "epoch": 1.31, "learning_rate": 1.4773333333333335e-05, "loss": 0.3305, "step": 8330 }, { "epoch": 1.31, "learning_rate": 1.4767058823529412e-05, "loss": 0.2603, "step": 8340 }, { "epoch": 1.31, "learning_rate": 1.4760784313725492e-05, "loss": 0.5993, "step": 8350 }, { "epoch": 1.31, "learning_rate": 1.4754509803921571e-05, "loss": 0.3527, "step": 8360 }, { "epoch": 1.31, "learning_rate": 1.4748235294117647e-05, "loss": 0.3487, "step": 8370 }, { "epoch": 1.31, "learning_rate": 1.4741960784313726e-05, "loss": 0.3064, "step": 8380 }, { "epoch": 1.32, "learning_rate": 1.4735686274509806e-05, "loss": 0.4511, "step": 8390 }, { "epoch": 1.32, "learning_rate": 1.4729411764705883e-05, "loss": 0.3199, "step": 8400 }, { "epoch": 1.32, "learning_rate": 1.4723137254901963e-05, "loss": 0.2742, "step": 8410 }, { "epoch": 1.32, "learning_rate": 1.471686274509804e-05, "loss": 0.6535, "step": 8420 }, { "epoch": 1.32, "learning_rate": 1.471058823529412e-05, "loss": 0.5921, "step": 8430 }, { "epoch": 1.32, "learning_rate": 1.4704313725490197e-05, "loss": 0.3706, "step": 8440 }, { "epoch": 1.33, "learning_rate": 1.4698039215686275e-05, "loss": 0.3022, "step": 8450 }, { "epoch": 1.33, "learning_rate": 1.4691764705882354e-05, "loss": 0.2821, "step": 8460 }, { "epoch": 1.33, "learning_rate": 1.4685490196078433e-05, "loss": 0.2855, "step": 8470 }, { "epoch": 1.33, "learning_rate": 1.4679215686274511e-05, "loss": 0.2559, "step": 8480 }, { "epoch": 1.33, "learning_rate": 1.467294117647059e-05, "loss": 0.4715, "step": 8490 }, { "epoch": 1.33, "learning_rate": 1.4666666666666666e-05, "loss": 0.4952, "step": 8500 }, { "epoch": 1.33, "learning_rate": 1.4660392156862746e-05, "loss": 0.2954, "step": 8510 }, { "epoch": 1.34, "learning_rate": 1.4654117647058825e-05, "loss": 0.4162, "step": 8520 }, { "epoch": 1.34, "learning_rate": 1.4647843137254903e-05, "loss": 0.4224, "step": 8530 }, { "epoch": 1.34, "learning_rate": 1.4641568627450982e-05, "loss": 0.3469, "step": 8540 }, { "epoch": 1.34, "learning_rate": 1.4635294117647061e-05, "loss": 0.2902, "step": 8550 }, { "epoch": 1.34, "learning_rate": 1.4629019607843137e-05, "loss": 0.216, "step": 8560 }, { "epoch": 1.34, "learning_rate": 1.4622745098039216e-05, "loss": 0.4603, "step": 8570 }, { "epoch": 1.35, "learning_rate": 1.4616470588235296e-05, "loss": 0.4493, "step": 8580 }, { "epoch": 1.35, "learning_rate": 1.4610196078431373e-05, "loss": 0.2641, "step": 8590 }, { "epoch": 1.35, "learning_rate": 1.4603921568627453e-05, "loss": 0.4157, "step": 8600 }, { "epoch": 1.35, "learning_rate": 1.459764705882353e-05, "loss": 0.4641, "step": 8610 }, { "epoch": 1.35, "learning_rate": 1.459137254901961e-05, "loss": 0.4579, "step": 8620 }, { "epoch": 1.35, "learning_rate": 1.4585098039215689e-05, "loss": 0.3132, "step": 8630 }, { "epoch": 1.36, "learning_rate": 1.4578823529411765e-05, "loss": 0.3968, "step": 8640 }, { "epoch": 1.36, "learning_rate": 1.4572549019607844e-05, "loss": 0.4259, "step": 8650 }, { "epoch": 1.36, "learning_rate": 1.4566274509803924e-05, "loss": 0.3659, "step": 8660 }, { "epoch": 1.36, "learning_rate": 1.4560000000000001e-05, "loss": 0.2324, "step": 8670 }, { "epoch": 1.36, "learning_rate": 1.455372549019608e-05, "loss": 0.4214, "step": 8680 }, { "epoch": 1.36, "learning_rate": 1.4547450980392156e-05, "loss": 0.6062, "step": 8690 }, { "epoch": 1.36, "learning_rate": 1.4541176470588236e-05, "loss": 0.2915, "step": 8700 }, { "epoch": 1.37, "learning_rate": 1.4534901960784315e-05, "loss": 0.4581, "step": 8710 }, { "epoch": 1.37, "learning_rate": 1.4528627450980393e-05, "loss": 0.2756, "step": 8720 }, { "epoch": 1.37, "learning_rate": 1.4522352941176472e-05, "loss": 0.4001, "step": 8730 }, { "epoch": 1.37, "learning_rate": 1.4516078431372551e-05, "loss": 0.1802, "step": 8740 }, { "epoch": 1.37, "learning_rate": 1.4509803921568629e-05, "loss": 0.4303, "step": 8750 }, { "epoch": 1.37, "learning_rate": 1.4503529411764707e-05, "loss": 0.3707, "step": 8760 }, { "epoch": 1.38, "learning_rate": 1.4497254901960784e-05, "loss": 0.516, "step": 8770 }, { "epoch": 1.38, "learning_rate": 1.4490980392156864e-05, "loss": 0.3068, "step": 8780 }, { "epoch": 1.38, "learning_rate": 1.4484705882352943e-05, "loss": 0.2312, "step": 8790 }, { "epoch": 1.38, "learning_rate": 1.447843137254902e-05, "loss": 0.4658, "step": 8800 }, { "epoch": 1.38, "learning_rate": 1.44721568627451e-05, "loss": 0.3876, "step": 8810 }, { "epoch": 1.38, "learning_rate": 1.446588235294118e-05, "loss": 0.3885, "step": 8820 }, { "epoch": 1.39, "learning_rate": 1.4459607843137255e-05, "loss": 0.3186, "step": 8830 }, { "epoch": 1.39, "learning_rate": 1.4453333333333334e-05, "loss": 0.3225, "step": 8840 }, { "epoch": 1.39, "learning_rate": 1.4447058823529414e-05, "loss": 0.3831, "step": 8850 }, { "epoch": 1.39, "learning_rate": 1.4440784313725491e-05, "loss": 0.3142, "step": 8860 }, { "epoch": 1.39, "learning_rate": 1.443450980392157e-05, "loss": 0.4597, "step": 8870 }, { "epoch": 1.39, "learning_rate": 1.4428235294117647e-05, "loss": 0.5696, "step": 8880 }, { "epoch": 1.39, "learning_rate": 1.4421960784313726e-05, "loss": 0.3284, "step": 8890 }, { "epoch": 1.4, "learning_rate": 1.4415686274509805e-05, "loss": 0.1671, "step": 8900 }, { "epoch": 1.4, "learning_rate": 1.4409411764705883e-05, "loss": 0.4223, "step": 8910 }, { "epoch": 1.4, "learning_rate": 1.4403137254901962e-05, "loss": 0.5975, "step": 8920 }, { "epoch": 1.4, "learning_rate": 1.4396862745098042e-05, "loss": 0.3239, "step": 8930 }, { "epoch": 1.4, "learning_rate": 1.439058823529412e-05, "loss": 0.3355, "step": 8940 }, { "epoch": 1.4, "learning_rate": 1.4384313725490199e-05, "loss": 0.216, "step": 8950 }, { "epoch": 1.41, "learning_rate": 1.4378039215686274e-05, "loss": 0.2673, "step": 8960 }, { "epoch": 1.41, "learning_rate": 1.4371764705882354e-05, "loss": 0.4958, "step": 8970 }, { "epoch": 1.41, "learning_rate": 1.4365490196078433e-05, "loss": 0.4574, "step": 8980 }, { "epoch": 1.41, "learning_rate": 1.435921568627451e-05, "loss": 0.3497, "step": 8990 }, { "epoch": 1.41, "learning_rate": 1.435294117647059e-05, "loss": 0.3084, "step": 9000 }, { "epoch": 1.41, "learning_rate": 1.434666666666667e-05, "loss": 0.2804, "step": 9010 }, { "epoch": 1.41, "learning_rate": 1.4340392156862745e-05, "loss": 0.5458, "step": 9020 }, { "epoch": 1.42, "learning_rate": 1.4334117647058825e-05, "loss": 0.3094, "step": 9030 }, { "epoch": 1.42, "learning_rate": 1.4327843137254902e-05, "loss": 0.5567, "step": 9040 }, { "epoch": 1.42, "learning_rate": 1.4321568627450982e-05, "loss": 0.5726, "step": 9050 }, { "epoch": 1.42, "learning_rate": 1.4315294117647061e-05, "loss": 0.3276, "step": 9060 }, { "epoch": 1.42, "learning_rate": 1.4309019607843137e-05, "loss": 0.2647, "step": 9070 }, { "epoch": 1.42, "learning_rate": 1.4302745098039216e-05, "loss": 0.394, "step": 9080 }, { "epoch": 1.43, "learning_rate": 1.4296470588235295e-05, "loss": 0.1932, "step": 9090 }, { "epoch": 1.43, "learning_rate": 1.4290196078431373e-05, "loss": 0.4212, "step": 9100 }, { "epoch": 1.43, "learning_rate": 1.4283921568627452e-05, "loss": 0.3135, "step": 9110 }, { "epoch": 1.43, "learning_rate": 1.4277647058823532e-05, "loss": 0.3921, "step": 9120 }, { "epoch": 1.43, "learning_rate": 1.427137254901961e-05, "loss": 0.3855, "step": 9130 }, { "epoch": 1.43, "learning_rate": 1.4265098039215689e-05, "loss": 0.5767, "step": 9140 }, { "epoch": 1.44, "learning_rate": 1.4258823529411765e-05, "loss": 0.1343, "step": 9150 }, { "epoch": 1.44, "learning_rate": 1.4252549019607844e-05, "loss": 0.3209, "step": 9160 }, { "epoch": 1.44, "learning_rate": 1.4246274509803923e-05, "loss": 0.3968, "step": 9170 }, { "epoch": 1.44, "learning_rate": 1.4240000000000001e-05, "loss": 0.2874, "step": 9180 }, { "epoch": 1.44, "learning_rate": 1.423372549019608e-05, "loss": 0.3934, "step": 9190 }, { "epoch": 1.44, "learning_rate": 1.422745098039216e-05, "loss": 0.5119, "step": 9200 }, { "epoch": 1.44, "learning_rate": 1.4221176470588235e-05, "loss": 0.3295, "step": 9210 }, { "epoch": 1.45, "learning_rate": 1.4214901960784315e-05, "loss": 0.598, "step": 9220 }, { "epoch": 1.45, "learning_rate": 1.4208627450980392e-05, "loss": 0.3395, "step": 9230 }, { "epoch": 1.45, "learning_rate": 1.4202352941176472e-05, "loss": 0.4401, "step": 9240 }, { "epoch": 1.45, "learning_rate": 1.4196078431372551e-05, "loss": 0.3498, "step": 9250 }, { "epoch": 1.45, "learning_rate": 1.4189803921568629e-05, "loss": 0.3493, "step": 9260 }, { "epoch": 1.45, "learning_rate": 1.4183529411764706e-05, "loss": 0.3605, "step": 9270 }, { "epoch": 1.46, "learning_rate": 1.4177254901960786e-05, "loss": 0.2195, "step": 9280 }, { "epoch": 1.46, "learning_rate": 1.4170980392156863e-05, "loss": 0.3451, "step": 9290 }, { "epoch": 1.46, "learning_rate": 1.4164705882352943e-05, "loss": 0.4205, "step": 9300 }, { "epoch": 1.46, "learning_rate": 1.415843137254902e-05, "loss": 0.4011, "step": 9310 }, { "epoch": 1.46, "learning_rate": 1.41521568627451e-05, "loss": 0.5428, "step": 9320 }, { "epoch": 1.46, "learning_rate": 1.4145882352941179e-05, "loss": 0.2527, "step": 9330 }, { "epoch": 1.47, "learning_rate": 1.4139607843137255e-05, "loss": 0.289, "step": 9340 }, { "epoch": 1.47, "learning_rate": 1.4133333333333334e-05, "loss": 0.1735, "step": 9350 }, { "epoch": 1.47, "learning_rate": 1.4127058823529413e-05, "loss": 0.539, "step": 9360 }, { "epoch": 1.47, "learning_rate": 1.4120784313725491e-05, "loss": 0.4972, "step": 9370 }, { "epoch": 1.47, "learning_rate": 1.411450980392157e-05, "loss": 0.3301, "step": 9380 }, { "epoch": 1.47, "learning_rate": 1.410823529411765e-05, "loss": 0.3951, "step": 9390 }, { "epoch": 1.47, "learning_rate": 1.4101960784313726e-05, "loss": 0.4577, "step": 9400 }, { "epoch": 1.48, "learning_rate": 1.4095686274509805e-05, "loss": 0.2461, "step": 9410 }, { "epoch": 1.48, "learning_rate": 1.4089411764705883e-05, "loss": 0.2672, "step": 9420 }, { "epoch": 1.48, "learning_rate": 1.4083137254901962e-05, "loss": 0.5548, "step": 9430 }, { "epoch": 1.48, "learning_rate": 1.4076862745098041e-05, "loss": 0.356, "step": 9440 }, { "epoch": 1.48, "learning_rate": 1.4070588235294119e-05, "loss": 0.4268, "step": 9450 }, { "epoch": 1.48, "learning_rate": 1.4064313725490198e-05, "loss": 0.1845, "step": 9460 }, { "epoch": 1.49, "learning_rate": 1.4058039215686276e-05, "loss": 0.4279, "step": 9470 }, { "epoch": 1.49, "learning_rate": 1.4051764705882353e-05, "loss": 0.4218, "step": 9480 }, { "epoch": 1.49, "learning_rate": 1.4045490196078433e-05, "loss": 0.2997, "step": 9490 }, { "epoch": 1.49, "learning_rate": 1.403921568627451e-05, "loss": 0.3511, "step": 9500 }, { "epoch": 1.49, "learning_rate": 1.403294117647059e-05, "loss": 0.2978, "step": 9510 }, { "epoch": 1.49, "learning_rate": 1.4026666666666669e-05, "loss": 0.6571, "step": 9520 }, { "epoch": 1.49, "learning_rate": 1.4020392156862745e-05, "loss": 0.2559, "step": 9530 }, { "epoch": 1.5, "learning_rate": 1.4014117647058824e-05, "loss": 0.45, "step": 9540 }, { "epoch": 1.5, "learning_rate": 1.4007843137254904e-05, "loss": 0.3305, "step": 9550 }, { "epoch": 1.5, "learning_rate": 1.4001568627450981e-05, "loss": 0.2222, "step": 9560 }, { "epoch": 1.5, "learning_rate": 1.399529411764706e-05, "loss": 0.255, "step": 9570 }, { "epoch": 1.5, "learning_rate": 1.398901960784314e-05, "loss": 0.5271, "step": 9580 }, { "epoch": 1.5, "learning_rate": 1.3982745098039216e-05, "loss": 0.2346, "step": 9590 }, { "epoch": 1.51, "learning_rate": 1.3976470588235295e-05, "loss": 0.5366, "step": 9600 }, { "epoch": 1.51, "learning_rate": 1.3970196078431373e-05, "loss": 0.2788, "step": 9610 }, { "epoch": 1.51, "learning_rate": 1.3963921568627452e-05, "loss": 0.2279, "step": 9620 }, { "epoch": 1.51, "learning_rate": 1.3957647058823531e-05, "loss": 0.3354, "step": 9630 }, { "epoch": 1.51, "learning_rate": 1.3951372549019609e-05, "loss": 0.5706, "step": 9640 }, { "epoch": 1.51, "learning_rate": 1.3945098039215688e-05, "loss": 0.4966, "step": 9650 }, { "epoch": 1.52, "learning_rate": 1.3938823529411768e-05, "loss": 0.3457, "step": 9660 }, { "epoch": 1.52, "learning_rate": 1.3932549019607844e-05, "loss": 0.4558, "step": 9670 }, { "epoch": 1.52, "learning_rate": 1.3926274509803923e-05, "loss": 0.4184, "step": 9680 }, { "epoch": 1.52, "learning_rate": 1.392e-05, "loss": 0.599, "step": 9690 }, { "epoch": 1.52, "learning_rate": 1.391372549019608e-05, "loss": 0.2719, "step": 9700 }, { "epoch": 1.52, "learning_rate": 1.390745098039216e-05, "loss": 0.3155, "step": 9710 }, { "epoch": 1.52, "learning_rate": 1.3901176470588235e-05, "loss": 0.4897, "step": 9720 }, { "epoch": 1.53, "learning_rate": 1.3894901960784314e-05, "loss": 0.5144, "step": 9730 }, { "epoch": 1.53, "learning_rate": 1.3888627450980394e-05, "loss": 0.3718, "step": 9740 }, { "epoch": 1.53, "learning_rate": 1.3882352941176471e-05, "loss": 0.3472, "step": 9750 }, { "epoch": 1.53, "learning_rate": 1.387607843137255e-05, "loss": 0.3077, "step": 9760 }, { "epoch": 1.53, "learning_rate": 1.3869803921568628e-05, "loss": 0.2836, "step": 9770 }, { "epoch": 1.53, "learning_rate": 1.3863529411764706e-05, "loss": 0.3464, "step": 9780 }, { "epoch": 1.54, "learning_rate": 1.3857254901960785e-05, "loss": 0.4095, "step": 9790 }, { "epoch": 1.54, "learning_rate": 1.3850980392156863e-05, "loss": 0.4772, "step": 9800 }, { "epoch": 1.54, "learning_rate": 1.3844705882352942e-05, "loss": 0.2961, "step": 9810 }, { "epoch": 1.54, "learning_rate": 1.3838431372549022e-05, "loss": 0.5481, "step": 9820 }, { "epoch": 1.54, "learning_rate": 1.38321568627451e-05, "loss": 0.2357, "step": 9830 }, { "epoch": 1.54, "learning_rate": 1.3825882352941179e-05, "loss": 0.6238, "step": 9840 }, { "epoch": 1.55, "learning_rate": 1.3819607843137258e-05, "loss": 0.2643, "step": 9850 }, { "epoch": 1.55, "learning_rate": 1.3813333333333334e-05, "loss": 0.4827, "step": 9860 }, { "epoch": 1.55, "learning_rate": 1.3807058823529413e-05, "loss": 0.3979, "step": 9870 }, { "epoch": 1.55, "learning_rate": 1.380078431372549e-05, "loss": 0.2617, "step": 9880 }, { "epoch": 1.55, "learning_rate": 1.379450980392157e-05, "loss": 0.3529, "step": 9890 }, { "epoch": 1.55, "learning_rate": 1.378823529411765e-05, "loss": 0.2172, "step": 9900 }, { "epoch": 1.55, "learning_rate": 1.3781960784313725e-05, "loss": 0.5613, "step": 9910 }, { "epoch": 1.56, "learning_rate": 1.3775686274509805e-05, "loss": 0.5647, "step": 9920 }, { "epoch": 1.56, "learning_rate": 1.3769411764705884e-05, "loss": 0.558, "step": 9930 }, { "epoch": 1.56, "learning_rate": 1.3763137254901962e-05, "loss": 0.4158, "step": 9940 }, { "epoch": 1.56, "learning_rate": 1.3756862745098041e-05, "loss": 0.4336, "step": 9950 }, { "epoch": 1.56, "learning_rate": 1.3750588235294119e-05, "loss": 0.3426, "step": 9960 }, { "epoch": 1.56, "learning_rate": 1.3744313725490198e-05, "loss": 0.6537, "step": 9970 }, { "epoch": 1.57, "learning_rate": 1.3738039215686275e-05, "loss": 0.359, "step": 9980 }, { "epoch": 1.57, "learning_rate": 1.3731764705882353e-05, "loss": 0.2782, "step": 9990 }, { "epoch": 1.57, "learning_rate": 1.3725490196078432e-05, "loss": 0.268, "step": 10000 }, { "epoch": 1.57, "learning_rate": 1.3719215686274512e-05, "loss": 0.464, "step": 10010 }, { "epoch": 1.57, "learning_rate": 1.371294117647059e-05, "loss": 0.4159, "step": 10020 }, { "epoch": 1.57, "learning_rate": 1.3706666666666669e-05, "loss": 0.3393, "step": 10030 }, { "epoch": 1.57, "learning_rate": 1.3700392156862745e-05, "loss": 0.2547, "step": 10040 }, { "epoch": 1.58, "learning_rate": 1.3694117647058824e-05, "loss": 0.3339, "step": 10050 }, { "epoch": 1.58, "learning_rate": 1.3687843137254903e-05, "loss": 0.1709, "step": 10060 }, { "epoch": 1.58, "learning_rate": 1.3681568627450981e-05, "loss": 0.5462, "step": 10070 }, { "epoch": 1.58, "learning_rate": 1.367529411764706e-05, "loss": 0.3424, "step": 10080 }, { "epoch": 1.58, "learning_rate": 1.366901960784314e-05, "loss": 0.249, "step": 10090 }, { "epoch": 1.58, "learning_rate": 1.3662745098039215e-05, "loss": 0.5113, "step": 10100 }, { "epoch": 1.59, "learning_rate": 1.3656470588235295e-05, "loss": 0.3698, "step": 10110 }, { "epoch": 1.59, "learning_rate": 1.3650196078431374e-05, "loss": 0.2984, "step": 10120 }, { "epoch": 1.59, "learning_rate": 1.3643921568627452e-05, "loss": 0.4613, "step": 10130 }, { "epoch": 1.59, "learning_rate": 1.3637647058823531e-05, "loss": 0.3706, "step": 10140 }, { "epoch": 1.59, "learning_rate": 1.3631372549019609e-05, "loss": 0.4068, "step": 10150 }, { "epoch": 1.59, "learning_rate": 1.3625098039215688e-05, "loss": 0.2641, "step": 10160 }, { "epoch": 1.6, "learning_rate": 1.3618823529411767e-05, "loss": 0.4207, "step": 10170 }, { "epoch": 1.6, "learning_rate": 1.3612549019607843e-05, "loss": 0.4925, "step": 10180 }, { "epoch": 1.6, "learning_rate": 1.3606274509803923e-05, "loss": 0.327, "step": 10190 }, { "epoch": 1.6, "learning_rate": 1.3600000000000002e-05, "loss": 0.3224, "step": 10200 }, { "epoch": 1.6, "learning_rate": 1.359372549019608e-05, "loss": 0.1843, "step": 10210 }, { "epoch": 1.6, "learning_rate": 1.3587450980392159e-05, "loss": 0.2856, "step": 10220 }, { "epoch": 1.6, "learning_rate": 1.3581176470588235e-05, "loss": 0.3014, "step": 10230 }, { "epoch": 1.61, "learning_rate": 1.3574901960784314e-05, "loss": 0.5219, "step": 10240 }, { "epoch": 1.61, "learning_rate": 1.3568627450980393e-05, "loss": 0.3454, "step": 10250 }, { "epoch": 1.61, "learning_rate": 1.3562352941176471e-05, "loss": 0.4247, "step": 10260 }, { "epoch": 1.61, "learning_rate": 1.355607843137255e-05, "loss": 0.399, "step": 10270 }, { "epoch": 1.61, "learning_rate": 1.354980392156863e-05, "loss": 0.4148, "step": 10280 }, { "epoch": 1.61, "learning_rate": 1.3543529411764706e-05, "loss": 0.4155, "step": 10290 }, { "epoch": 1.62, "learning_rate": 1.3537254901960785e-05, "loss": 0.4085, "step": 10300 }, { "epoch": 1.62, "learning_rate": 1.3530980392156863e-05, "loss": 0.4144, "step": 10310 }, { "epoch": 1.62, "learning_rate": 1.3524705882352942e-05, "loss": 0.3597, "step": 10320 }, { "epoch": 1.62, "learning_rate": 1.3518431372549021e-05, "loss": 0.4445, "step": 10330 }, { "epoch": 1.62, "learning_rate": 1.3512156862745099e-05, "loss": 0.3207, "step": 10340 }, { "epoch": 1.62, "learning_rate": 1.3505882352941178e-05, "loss": 0.1896, "step": 10350 }, { "epoch": 1.63, "learning_rate": 1.3499607843137258e-05, "loss": 0.4039, "step": 10360 }, { "epoch": 1.63, "learning_rate": 1.3493333333333333e-05, "loss": 0.2932, "step": 10370 }, { "epoch": 1.63, "learning_rate": 1.3487058823529413e-05, "loss": 0.2875, "step": 10380 }, { "epoch": 1.63, "learning_rate": 1.3480784313725492e-05, "loss": 0.414, "step": 10390 }, { "epoch": 1.63, "learning_rate": 1.347450980392157e-05, "loss": 0.1822, "step": 10400 }, { "epoch": 1.63, "learning_rate": 1.3468235294117649e-05, "loss": 0.5252, "step": 10410 }, { "epoch": 1.63, "learning_rate": 1.3461960784313725e-05, "loss": 0.2688, "step": 10420 }, { "epoch": 1.64, "learning_rate": 1.3455686274509804e-05, "loss": 0.5229, "step": 10430 }, { "epoch": 1.64, "learning_rate": 1.3449411764705884e-05, "loss": 0.4627, "step": 10440 }, { "epoch": 1.64, "learning_rate": 1.3443137254901961e-05, "loss": 0.4972, "step": 10450 }, { "epoch": 1.64, "learning_rate": 1.343686274509804e-05, "loss": 0.3324, "step": 10460 }, { "epoch": 1.64, "learning_rate": 1.343058823529412e-05, "loss": 0.3484, "step": 10470 }, { "epoch": 1.64, "learning_rate": 1.3424313725490198e-05, "loss": 0.451, "step": 10480 }, { "epoch": 1.65, "learning_rate": 1.3418039215686275e-05, "loss": 0.2483, "step": 10490 }, { "epoch": 1.65, "learning_rate": 1.3411764705882353e-05, "loss": 0.2454, "step": 10500 }, { "epoch": 1.65, "learning_rate": 1.3405490196078432e-05, "loss": 0.5722, "step": 10510 }, { "epoch": 1.65, "learning_rate": 1.3399215686274511e-05, "loss": 0.221, "step": 10520 }, { "epoch": 1.65, "learning_rate": 1.3392941176470589e-05, "loss": 0.3422, "step": 10530 }, { "epoch": 1.65, "learning_rate": 1.3386666666666668e-05, "loss": 0.3867, "step": 10540 }, { "epoch": 1.65, "learning_rate": 1.3380392156862748e-05, "loss": 0.3328, "step": 10550 }, { "epoch": 1.66, "learning_rate": 1.3374117647058824e-05, "loss": 0.346, "step": 10560 }, { "epoch": 1.66, "learning_rate": 1.3367843137254903e-05, "loss": 0.2724, "step": 10570 }, { "epoch": 1.66, "learning_rate": 1.336156862745098e-05, "loss": 0.2522, "step": 10580 }, { "epoch": 1.66, "learning_rate": 1.335529411764706e-05, "loss": 0.4238, "step": 10590 }, { "epoch": 1.66, "learning_rate": 1.334901960784314e-05, "loss": 0.2705, "step": 10600 }, { "epoch": 1.66, "learning_rate": 1.3342745098039215e-05, "loss": 0.4663, "step": 10610 }, { "epoch": 1.67, "learning_rate": 1.3336470588235294e-05, "loss": 0.391, "step": 10620 }, { "epoch": 1.67, "learning_rate": 1.3330196078431374e-05, "loss": 0.3345, "step": 10630 }, { "epoch": 1.67, "learning_rate": 1.3323921568627451e-05, "loss": 0.3156, "step": 10640 }, { "epoch": 1.67, "learning_rate": 1.331764705882353e-05, "loss": 0.3943, "step": 10650 }, { "epoch": 1.67, "learning_rate": 1.331137254901961e-05, "loss": 0.2667, "step": 10660 }, { "epoch": 1.67, "learning_rate": 1.3305098039215688e-05, "loss": 0.4984, "step": 10670 }, { "epoch": 1.68, "learning_rate": 1.3298823529411767e-05, "loss": 0.5166, "step": 10680 }, { "epoch": 1.68, "learning_rate": 1.3292549019607843e-05, "loss": 0.4107, "step": 10690 }, { "epoch": 1.68, "learning_rate": 1.3286274509803922e-05, "loss": 0.2452, "step": 10700 }, { "epoch": 1.68, "learning_rate": 1.3280000000000002e-05, "loss": 0.5441, "step": 10710 }, { "epoch": 1.68, "learning_rate": 1.327372549019608e-05, "loss": 0.2916, "step": 10720 }, { "epoch": 1.68, "learning_rate": 1.3267450980392159e-05, "loss": 0.5439, "step": 10730 }, { "epoch": 1.68, "learning_rate": 1.3261176470588238e-05, "loss": 0.722, "step": 10740 }, { "epoch": 1.69, "learning_rate": 1.3254901960784314e-05, "loss": 0.3391, "step": 10750 }, { "epoch": 1.69, "learning_rate": 1.3248627450980393e-05, "loss": 0.2018, "step": 10760 }, { "epoch": 1.69, "learning_rate": 1.324235294117647e-05, "loss": 0.5892, "step": 10770 }, { "epoch": 1.69, "learning_rate": 1.323607843137255e-05, "loss": 0.36, "step": 10780 }, { "epoch": 1.69, "learning_rate": 1.322980392156863e-05, "loss": 0.3912, "step": 10790 }, { "epoch": 1.69, "learning_rate": 1.3223529411764705e-05, "loss": 0.4338, "step": 10800 }, { "epoch": 1.7, "learning_rate": 1.3217254901960785e-05, "loss": 0.5696, "step": 10810 }, { "epoch": 1.7, "learning_rate": 1.3210980392156864e-05, "loss": 0.2492, "step": 10820 }, { "epoch": 1.7, "learning_rate": 1.3204705882352942e-05, "loss": 0.3229, "step": 10830 }, { "epoch": 1.7, "learning_rate": 1.3198431372549021e-05, "loss": 0.4964, "step": 10840 }, { "epoch": 1.7, "learning_rate": 1.31921568627451e-05, "loss": 0.4414, "step": 10850 }, { "epoch": 1.7, "learning_rate": 1.3185882352941178e-05, "loss": 0.1655, "step": 10860 }, { "epoch": 1.71, "learning_rate": 1.3179607843137257e-05, "loss": 0.4725, "step": 10870 }, { "epoch": 1.71, "learning_rate": 1.3173333333333333e-05, "loss": 0.3141, "step": 10880 }, { "epoch": 1.71, "learning_rate": 1.3167058823529412e-05, "loss": 0.5239, "step": 10890 }, { "epoch": 1.71, "learning_rate": 1.3160784313725492e-05, "loss": 0.3468, "step": 10900 }, { "epoch": 1.71, "learning_rate": 1.315450980392157e-05, "loss": 0.2106, "step": 10910 }, { "epoch": 1.71, "learning_rate": 1.3148235294117649e-05, "loss": 0.1843, "step": 10920 }, { "epoch": 1.71, "learning_rate": 1.3141960784313728e-05, "loss": 0.1685, "step": 10930 }, { "epoch": 1.72, "learning_rate": 1.3135686274509804e-05, "loss": 0.2727, "step": 10940 }, { "epoch": 1.72, "learning_rate": 1.3129411764705883e-05, "loss": 0.2142, "step": 10950 }, { "epoch": 1.72, "learning_rate": 1.3123137254901961e-05, "loss": 0.2367, "step": 10960 }, { "epoch": 1.72, "learning_rate": 1.311686274509804e-05, "loss": 0.2884, "step": 10970 }, { "epoch": 1.72, "learning_rate": 1.311058823529412e-05, "loss": 0.2898, "step": 10980 }, { "epoch": 1.72, "learning_rate": 1.3104313725490197e-05, "loss": 0.2133, "step": 10990 }, { "epoch": 1.73, "learning_rate": 1.3098039215686275e-05, "loss": 0.4117, "step": 11000 }, { "epoch": 1.73, "learning_rate": 1.3091764705882354e-05, "loss": 0.4761, "step": 11010 }, { "epoch": 1.73, "learning_rate": 1.3085490196078432e-05, "loss": 0.5273, "step": 11020 }, { "epoch": 1.73, "learning_rate": 1.3079215686274511e-05, "loss": 0.3177, "step": 11030 }, { "epoch": 1.73, "learning_rate": 1.3072941176470589e-05, "loss": 0.3578, "step": 11040 }, { "epoch": 1.73, "learning_rate": 1.3066666666666668e-05, "loss": 0.5648, "step": 11050 }, { "epoch": 1.73, "learning_rate": 1.3060392156862747e-05, "loss": 0.3829, "step": 11060 }, { "epoch": 1.74, "learning_rate": 1.3054117647058823e-05, "loss": 0.3417, "step": 11070 }, { "epoch": 1.74, "learning_rate": 1.3047843137254903e-05, "loss": 0.4169, "step": 11080 }, { "epoch": 1.74, "learning_rate": 1.3041568627450982e-05, "loss": 0.3338, "step": 11090 }, { "epoch": 1.74, "learning_rate": 1.303529411764706e-05, "loss": 0.3193, "step": 11100 }, { "epoch": 1.74, "learning_rate": 1.3029019607843139e-05, "loss": 0.4829, "step": 11110 }, { "epoch": 1.74, "learning_rate": 1.3022745098039218e-05, "loss": 0.3575, "step": 11120 }, { "epoch": 1.75, "learning_rate": 1.3016470588235294e-05, "loss": 0.2474, "step": 11130 }, { "epoch": 1.75, "learning_rate": 1.3010196078431373e-05, "loss": 0.2213, "step": 11140 }, { "epoch": 1.75, "learning_rate": 1.3003921568627451e-05, "loss": 0.6516, "step": 11150 }, { "epoch": 1.75, "learning_rate": 1.299764705882353e-05, "loss": 0.4498, "step": 11160 }, { "epoch": 1.75, "learning_rate": 1.299137254901961e-05, "loss": 0.5632, "step": 11170 }, { "epoch": 1.75, "learning_rate": 1.2985098039215687e-05, "loss": 0.2757, "step": 11180 }, { "epoch": 1.76, "learning_rate": 1.2978823529411767e-05, "loss": 0.4219, "step": 11190 }, { "epoch": 1.76, "learning_rate": 1.2972549019607846e-05, "loss": 0.4714, "step": 11200 }, { "epoch": 1.76, "learning_rate": 1.2966274509803922e-05, "loss": 0.4, "step": 11210 }, { "epoch": 1.76, "learning_rate": 1.2960000000000001e-05, "loss": 0.4498, "step": 11220 }, { "epoch": 1.76, "learning_rate": 1.2953725490196079e-05, "loss": 0.4098, "step": 11230 }, { "epoch": 1.76, "learning_rate": 1.2947450980392158e-05, "loss": 0.2643, "step": 11240 }, { "epoch": 1.76, "learning_rate": 1.2941176470588238e-05, "loss": 0.135, "step": 11250 }, { "epoch": 1.77, "learning_rate": 1.2934901960784313e-05, "loss": 0.3841, "step": 11260 }, { "epoch": 1.77, "learning_rate": 1.2928627450980393e-05, "loss": 0.4091, "step": 11270 }, { "epoch": 1.77, "learning_rate": 1.2922352941176472e-05, "loss": 0.3861, "step": 11280 }, { "epoch": 1.77, "learning_rate": 1.291607843137255e-05, "loss": 0.5825, "step": 11290 }, { "epoch": 1.77, "learning_rate": 1.2909803921568629e-05, "loss": 0.2816, "step": 11300 }, { "epoch": 1.77, "learning_rate": 1.2903529411764707e-05, "loss": 0.5509, "step": 11310 }, { "epoch": 1.78, "learning_rate": 1.2897254901960784e-05, "loss": 0.3297, "step": 11320 }, { "epoch": 1.78, "learning_rate": 1.2890980392156864e-05, "loss": 0.5811, "step": 11330 }, { "epoch": 1.78, "learning_rate": 1.2884705882352941e-05, "loss": 0.462, "step": 11340 }, { "epoch": 1.78, "learning_rate": 1.287843137254902e-05, "loss": 0.2338, "step": 11350 }, { "epoch": 1.78, "learning_rate": 1.28721568627451e-05, "loss": 0.3535, "step": 11360 }, { "epoch": 1.78, "learning_rate": 1.2865882352941178e-05, "loss": 0.3749, "step": 11370 }, { "epoch": 1.79, "learning_rate": 1.2859607843137257e-05, "loss": 0.3956, "step": 11380 }, { "epoch": 1.79, "learning_rate": 1.2853333333333336e-05, "loss": 0.376, "step": 11390 }, { "epoch": 1.79, "learning_rate": 1.2847058823529412e-05, "loss": 0.3961, "step": 11400 }, { "epoch": 1.79, "learning_rate": 1.2840784313725491e-05, "loss": 0.3428, "step": 11410 }, { "epoch": 1.79, "learning_rate": 1.2834509803921569e-05, "loss": 0.4303, "step": 11420 }, { "epoch": 1.79, "learning_rate": 1.2828235294117648e-05, "loss": 0.3831, "step": 11430 }, { "epoch": 1.79, "learning_rate": 1.2821960784313728e-05, "loss": 0.4349, "step": 11440 }, { "epoch": 1.8, "learning_rate": 1.2815686274509804e-05, "loss": 0.4024, "step": 11450 }, { "epoch": 1.8, "learning_rate": 1.2809411764705883e-05, "loss": 0.2848, "step": 11460 }, { "epoch": 1.8, "learning_rate": 1.2803137254901962e-05, "loss": 0.4243, "step": 11470 }, { "epoch": 1.8, "learning_rate": 1.279686274509804e-05, "loss": 0.249, "step": 11480 }, { "epoch": 1.8, "learning_rate": 1.279058823529412e-05, "loss": 0.4759, "step": 11490 }, { "epoch": 1.8, "learning_rate": 1.2784313725490197e-05, "loss": 0.5418, "step": 11500 }, { "epoch": 1.81, "learning_rate": 1.2778039215686276e-05, "loss": 0.3161, "step": 11510 }, { "epoch": 1.81, "learning_rate": 1.2771764705882354e-05, "loss": 0.4522, "step": 11520 }, { "epoch": 1.81, "learning_rate": 1.2765490196078431e-05, "loss": 0.4239, "step": 11530 }, { "epoch": 1.81, "learning_rate": 1.275921568627451e-05, "loss": 0.4725, "step": 11540 }, { "epoch": 1.81, "learning_rate": 1.275294117647059e-05, "loss": 0.4988, "step": 11550 }, { "epoch": 1.81, "learning_rate": 1.2746666666666668e-05, "loss": 0.2355, "step": 11560 }, { "epoch": 1.81, "learning_rate": 1.2740392156862747e-05, "loss": 0.4977, "step": 11570 }, { "epoch": 1.82, "learning_rate": 1.2734117647058823e-05, "loss": 0.3372, "step": 11580 }, { "epoch": 1.82, "learning_rate": 1.2727843137254902e-05, "loss": 0.3054, "step": 11590 }, { "epoch": 1.82, "learning_rate": 1.2721568627450982e-05, "loss": 0.4848, "step": 11600 }, { "epoch": 1.82, "learning_rate": 1.271529411764706e-05, "loss": 0.2698, "step": 11610 }, { "epoch": 1.82, "learning_rate": 1.2709019607843139e-05, "loss": 0.3586, "step": 11620 }, { "epoch": 1.82, "learning_rate": 1.2702745098039218e-05, "loss": 0.4535, "step": 11630 }, { "epoch": 1.83, "learning_rate": 1.2696470588235294e-05, "loss": 0.4109, "step": 11640 }, { "epoch": 1.83, "learning_rate": 1.2690196078431373e-05, "loss": 0.3265, "step": 11650 }, { "epoch": 1.83, "learning_rate": 1.2683921568627452e-05, "loss": 0.4608, "step": 11660 }, { "epoch": 1.83, "learning_rate": 1.267764705882353e-05, "loss": 0.3205, "step": 11670 }, { "epoch": 1.83, "learning_rate": 1.267137254901961e-05, "loss": 0.3797, "step": 11680 }, { "epoch": 1.83, "learning_rate": 1.2665098039215687e-05, "loss": 0.1754, "step": 11690 }, { "epoch": 1.84, "learning_rate": 1.2658823529411766e-05, "loss": 0.3429, "step": 11700 }, { "epoch": 1.84, "learning_rate": 1.2652549019607846e-05, "loss": 0.3937, "step": 11710 }, { "epoch": 1.84, "learning_rate": 1.2646274509803922e-05, "loss": 0.2729, "step": 11720 }, { "epoch": 1.84, "learning_rate": 1.2640000000000001e-05, "loss": 0.3745, "step": 11730 }, { "epoch": 1.84, "learning_rate": 1.263372549019608e-05, "loss": 0.358, "step": 11740 }, { "epoch": 1.84, "learning_rate": 1.2627450980392158e-05, "loss": 0.4804, "step": 11750 }, { "epoch": 1.84, "learning_rate": 1.2621176470588237e-05, "loss": 0.4504, "step": 11760 }, { "epoch": 1.85, "learning_rate": 1.2614901960784313e-05, "loss": 0.4287, "step": 11770 }, { "epoch": 1.85, "learning_rate": 1.2608627450980392e-05, "loss": 0.5594, "step": 11780 }, { "epoch": 1.85, "learning_rate": 1.2602352941176472e-05, "loss": 0.4308, "step": 11790 }, { "epoch": 1.85, "learning_rate": 1.259607843137255e-05, "loss": 0.4993, "step": 11800 }, { "epoch": 1.85, "learning_rate": 1.2589803921568629e-05, "loss": 0.1402, "step": 11810 }, { "epoch": 1.85, "learning_rate": 1.2583529411764708e-05, "loss": 0.5005, "step": 11820 }, { "epoch": 1.86, "learning_rate": 1.2577254901960784e-05, "loss": 0.4816, "step": 11830 }, { "epoch": 1.86, "learning_rate": 1.2570980392156863e-05, "loss": 0.4048, "step": 11840 }, { "epoch": 1.86, "learning_rate": 1.2564705882352943e-05, "loss": 0.105, "step": 11850 }, { "epoch": 1.86, "learning_rate": 1.255843137254902e-05, "loss": 0.5929, "step": 11860 }, { "epoch": 1.86, "learning_rate": 1.25521568627451e-05, "loss": 0.246, "step": 11870 }, { "epoch": 1.86, "learning_rate": 1.2545882352941177e-05, "loss": 0.2345, "step": 11880 }, { "epoch": 1.87, "learning_rate": 1.2539607843137257e-05, "loss": 0.2364, "step": 11890 }, { "epoch": 1.87, "learning_rate": 1.2533333333333336e-05, "loss": 0.4001, "step": 11900 }, { "epoch": 1.87, "learning_rate": 1.2527058823529412e-05, "loss": 0.2811, "step": 11910 }, { "epoch": 1.87, "learning_rate": 1.2520784313725491e-05, "loss": 0.5179, "step": 11920 }, { "epoch": 1.87, "learning_rate": 1.251450980392157e-05, "loss": 0.2272, "step": 11930 }, { "epoch": 1.87, "learning_rate": 1.2508235294117648e-05, "loss": 0.1938, "step": 11940 }, { "epoch": 1.87, "learning_rate": 1.2501960784313727e-05, "loss": 0.2894, "step": 11950 }, { "epoch": 1.88, "learning_rate": 1.2495686274509803e-05, "loss": 0.352, "step": 11960 }, { "epoch": 1.88, "learning_rate": 1.2489411764705883e-05, "loss": 0.544, "step": 11970 }, { "epoch": 1.88, "learning_rate": 1.2483137254901962e-05, "loss": 0.2947, "step": 11980 }, { "epoch": 1.88, "learning_rate": 1.247686274509804e-05, "loss": 0.1881, "step": 11990 }, { "epoch": 1.88, "learning_rate": 1.2470588235294119e-05, "loss": 0.3673, "step": 12000 }, { "epoch": 1.88, "learning_rate": 1.2464313725490198e-05, "loss": 0.3678, "step": 12010 }, { "epoch": 1.89, "learning_rate": 1.2458039215686276e-05, "loss": 0.0969, "step": 12020 }, { "epoch": 1.89, "learning_rate": 1.2451764705882354e-05, "loss": 0.2062, "step": 12030 }, { "epoch": 1.89, "learning_rate": 1.2445490196078431e-05, "loss": 0.5101, "step": 12040 }, { "epoch": 1.89, "learning_rate": 1.243921568627451e-05, "loss": 0.4532, "step": 12050 }, { "epoch": 1.89, "learning_rate": 1.243294117647059e-05, "loss": 0.4473, "step": 12060 }, { "epoch": 1.89, "learning_rate": 1.2426666666666667e-05, "loss": 0.2824, "step": 12070 }, { "epoch": 1.89, "learning_rate": 1.2420392156862747e-05, "loss": 0.375, "step": 12080 }, { "epoch": 1.9, "learning_rate": 1.2414117647058826e-05, "loss": 0.3707, "step": 12090 }, { "epoch": 1.9, "learning_rate": 1.2407843137254902e-05, "loss": 0.441, "step": 12100 }, { "epoch": 1.9, "learning_rate": 1.2401568627450981e-05, "loss": 0.1625, "step": 12110 }, { "epoch": 1.9, "learning_rate": 1.239529411764706e-05, "loss": 0.522, "step": 12120 }, { "epoch": 1.9, "learning_rate": 1.2389019607843138e-05, "loss": 0.3642, "step": 12130 }, { "epoch": 1.9, "learning_rate": 1.2382745098039218e-05, "loss": 0.3622, "step": 12140 }, { "epoch": 1.91, "learning_rate": 1.2376470588235294e-05, "loss": 0.3136, "step": 12150 }, { "epoch": 1.91, "learning_rate": 1.2370196078431373e-05, "loss": 0.2841, "step": 12160 }, { "epoch": 1.91, "learning_rate": 1.2363921568627452e-05, "loss": 0.3143, "step": 12170 }, { "epoch": 1.91, "learning_rate": 1.235764705882353e-05, "loss": 0.3429, "step": 12180 }, { "epoch": 1.91, "learning_rate": 1.2351372549019609e-05, "loss": 0.3694, "step": 12190 }, { "epoch": 1.91, "learning_rate": 1.2345098039215688e-05, "loss": 0.3474, "step": 12200 }, { "epoch": 1.92, "learning_rate": 1.2338823529411766e-05, "loss": 0.3627, "step": 12210 }, { "epoch": 1.92, "learning_rate": 1.2332549019607845e-05, "loss": 0.2392, "step": 12220 }, { "epoch": 1.92, "learning_rate": 1.2326274509803921e-05, "loss": 0.2377, "step": 12230 }, { "epoch": 1.92, "learning_rate": 1.232e-05, "loss": 0.4246, "step": 12240 }, { "epoch": 1.92, "learning_rate": 1.231372549019608e-05, "loss": 0.2513, "step": 12250 }, { "epoch": 1.92, "learning_rate": 1.2307450980392158e-05, "loss": 0.4154, "step": 12260 }, { "epoch": 1.92, "learning_rate": 1.2301176470588237e-05, "loss": 0.3518, "step": 12270 }, { "epoch": 1.93, "learning_rate": 1.2294901960784316e-05, "loss": 0.504, "step": 12280 }, { "epoch": 1.93, "learning_rate": 1.2288627450980392e-05, "loss": 0.5369, "step": 12290 }, { "epoch": 1.93, "learning_rate": 1.2282352941176471e-05, "loss": 0.3242, "step": 12300 }, { "epoch": 1.93, "learning_rate": 1.2276078431372549e-05, "loss": 0.3073, "step": 12310 }, { "epoch": 1.93, "learning_rate": 1.2269803921568628e-05, "loss": 0.389, "step": 12320 }, { "epoch": 1.93, "learning_rate": 1.2263529411764708e-05, "loss": 0.3297, "step": 12330 }, { "epoch": 1.94, "learning_rate": 1.2257254901960784e-05, "loss": 0.5076, "step": 12340 }, { "epoch": 1.94, "learning_rate": 1.2250980392156863e-05, "loss": 0.227, "step": 12350 }, { "epoch": 1.94, "learning_rate": 1.2244705882352942e-05, "loss": 0.2997, "step": 12360 }, { "epoch": 1.94, "learning_rate": 1.223843137254902e-05, "loss": 0.4103, "step": 12370 }, { "epoch": 1.94, "learning_rate": 1.22321568627451e-05, "loss": 0.451, "step": 12380 }, { "epoch": 1.94, "learning_rate": 1.2225882352941179e-05, "loss": 0.2284, "step": 12390 }, { "epoch": 1.95, "learning_rate": 1.2219607843137256e-05, "loss": 0.3711, "step": 12400 }, { "epoch": 1.95, "learning_rate": 1.2213333333333336e-05, "loss": 0.2899, "step": 12410 }, { "epoch": 1.95, "learning_rate": 1.2207058823529411e-05, "loss": 0.4127, "step": 12420 }, { "epoch": 1.95, "learning_rate": 1.220078431372549e-05, "loss": 0.3065, "step": 12430 }, { "epoch": 1.95, "learning_rate": 1.219450980392157e-05, "loss": 0.5871, "step": 12440 }, { "epoch": 1.95, "learning_rate": 1.2188235294117648e-05, "loss": 0.4413, "step": 12450 }, { "epoch": 1.95, "learning_rate": 1.2181960784313727e-05, "loss": 0.361, "step": 12460 }, { "epoch": 1.96, "learning_rate": 1.2175686274509806e-05, "loss": 0.3683, "step": 12470 }, { "epoch": 1.96, "learning_rate": 1.2169411764705882e-05, "loss": 0.4111, "step": 12480 }, { "epoch": 1.96, "learning_rate": 1.2163137254901962e-05, "loss": 0.3771, "step": 12490 }, { "epoch": 1.96, "learning_rate": 1.215686274509804e-05, "loss": 0.4342, "step": 12500 }, { "epoch": 1.96, "learning_rate": 1.2150588235294119e-05, "loss": 0.3457, "step": 12510 }, { "epoch": 1.96, "learning_rate": 1.2144313725490198e-05, "loss": 0.2906, "step": 12520 }, { "epoch": 1.97, "learning_rate": 1.2138039215686276e-05, "loss": 0.3554, "step": 12530 }, { "epoch": 1.97, "learning_rate": 1.2131764705882353e-05, "loss": 0.3912, "step": 12540 }, { "epoch": 1.97, "learning_rate": 1.2125490196078433e-05, "loss": 0.3306, "step": 12550 }, { "epoch": 1.97, "learning_rate": 1.211921568627451e-05, "loss": 0.3368, "step": 12560 }, { "epoch": 1.97, "learning_rate": 1.211294117647059e-05, "loss": 0.5452, "step": 12570 }, { "epoch": 1.97, "learning_rate": 1.2106666666666667e-05, "loss": 0.4398, "step": 12580 }, { "epoch": 1.97, "learning_rate": 1.2100392156862746e-05, "loss": 0.4171, "step": 12590 }, { "epoch": 1.98, "learning_rate": 1.2094117647058826e-05, "loss": 0.2516, "step": 12600 }, { "epoch": 1.98, "learning_rate": 1.2087843137254902e-05, "loss": 0.3611, "step": 12610 }, { "epoch": 1.98, "learning_rate": 1.2081568627450981e-05, "loss": 0.2926, "step": 12620 }, { "epoch": 1.98, "learning_rate": 1.207529411764706e-05, "loss": 0.5042, "step": 12630 }, { "epoch": 1.98, "learning_rate": 1.2069019607843138e-05, "loss": 0.2658, "step": 12640 }, { "epoch": 1.98, "learning_rate": 1.2062745098039217e-05, "loss": 0.3337, "step": 12650 }, { "epoch": 1.99, "learning_rate": 1.2056470588235297e-05, "loss": 0.4474, "step": 12660 }, { "epoch": 1.99, "learning_rate": 1.2050196078431373e-05, "loss": 0.4588, "step": 12670 }, { "epoch": 1.99, "learning_rate": 1.2043921568627452e-05, "loss": 0.4396, "step": 12680 }, { "epoch": 1.99, "learning_rate": 1.203764705882353e-05, "loss": 0.4365, "step": 12690 }, { "epoch": 1.99, "learning_rate": 1.2031372549019609e-05, "loss": 0.3329, "step": 12700 }, { "epoch": 1.99, "learning_rate": 1.2025098039215688e-05, "loss": 0.339, "step": 12710 }, { "epoch": 2.0, "learning_rate": 1.2018823529411766e-05, "loss": 0.2518, "step": 12720 }, { "epoch": 2.0, "learning_rate": 1.2012549019607845e-05, "loss": 0.3736, "step": 12730 }, { "epoch": 2.0, "learning_rate": 1.2006274509803923e-05, "loss": 0.2254, "step": 12740 }, { "epoch": 2.0, "learning_rate": 1.2e-05, "loss": 0.4689, "step": 12750 }, { "epoch": 2.0, "eval_accuracy": 0.9905555555555555, "eval_loss": 0.03773761913180351, "eval_runtime": 182.2547, "eval_samples_per_second": 49.381, "eval_steps_per_second": 6.173, "step": 12750 }, { "epoch": 2.0, "learning_rate": 1.199372549019608e-05, "loss": 0.2406, "step": 12760 }, { "epoch": 2.0, "learning_rate": 1.1987450980392157e-05, "loss": 0.2403, "step": 12770 }, { "epoch": 2.0, "learning_rate": 1.1981176470588237e-05, "loss": 0.4193, "step": 12780 }, { "epoch": 2.01, "learning_rate": 1.1974901960784316e-05, "loss": 0.5526, "step": 12790 }, { "epoch": 2.01, "learning_rate": 1.1968627450980392e-05, "loss": 0.574, "step": 12800 }, { "epoch": 2.01, "learning_rate": 1.1962352941176471e-05, "loss": 0.3188, "step": 12810 }, { "epoch": 2.01, "learning_rate": 1.195607843137255e-05, "loss": 0.3882, "step": 12820 }, { "epoch": 2.01, "learning_rate": 1.1949803921568628e-05, "loss": 0.3393, "step": 12830 }, { "epoch": 2.01, "learning_rate": 1.1943529411764707e-05, "loss": 0.4666, "step": 12840 }, { "epoch": 2.02, "learning_rate": 1.1937254901960783e-05, "loss": 0.3478, "step": 12850 }, { "epoch": 2.02, "learning_rate": 1.1930980392156863e-05, "loss": 0.1423, "step": 12860 }, { "epoch": 2.02, "learning_rate": 1.1924705882352942e-05, "loss": 0.3163, "step": 12870 }, { "epoch": 2.02, "learning_rate": 1.191843137254902e-05, "loss": 0.528, "step": 12880 }, { "epoch": 2.02, "learning_rate": 1.1912156862745099e-05, "loss": 0.2968, "step": 12890 }, { "epoch": 2.02, "learning_rate": 1.1905882352941178e-05, "loss": 0.4141, "step": 12900 }, { "epoch": 2.03, "learning_rate": 1.1899607843137256e-05, "loss": 0.5459, "step": 12910 }, { "epoch": 2.03, "learning_rate": 1.1893333333333335e-05, "loss": 0.3089, "step": 12920 }, { "epoch": 2.03, "learning_rate": 1.1887058823529415e-05, "loss": 0.3373, "step": 12930 }, { "epoch": 2.03, "learning_rate": 1.188078431372549e-05, "loss": 0.2781, "step": 12940 }, { "epoch": 2.03, "learning_rate": 1.187450980392157e-05, "loss": 0.4583, "step": 12950 }, { "epoch": 2.03, "learning_rate": 1.1868235294117647e-05, "loss": 0.2337, "step": 12960 }, { "epoch": 2.03, "learning_rate": 1.1861960784313727e-05, "loss": 0.395, "step": 12970 }, { "epoch": 2.04, "learning_rate": 1.1855686274509806e-05, "loss": 0.306, "step": 12980 }, { "epoch": 2.04, "learning_rate": 1.1849411764705882e-05, "loss": 0.467, "step": 12990 }, { "epoch": 2.04, "learning_rate": 1.1843137254901961e-05, "loss": 0.3191, "step": 13000 }, { "epoch": 2.04, "learning_rate": 1.183686274509804e-05, "loss": 0.3429, "step": 13010 }, { "epoch": 2.04, "learning_rate": 1.1830588235294118e-05, "loss": 0.4106, "step": 13020 }, { "epoch": 2.04, "learning_rate": 1.1824313725490198e-05, "loss": 0.334, "step": 13030 }, { "epoch": 2.05, "learning_rate": 1.1818039215686275e-05, "loss": 0.3218, "step": 13040 }, { "epoch": 2.05, "learning_rate": 1.1811764705882353e-05, "loss": 0.2783, "step": 13050 }, { "epoch": 2.05, "learning_rate": 1.1805490196078432e-05, "loss": 0.4557, "step": 13060 }, { "epoch": 2.05, "learning_rate": 1.179921568627451e-05, "loss": 0.3817, "step": 13070 }, { "epoch": 2.05, "learning_rate": 1.1792941176470589e-05, "loss": 0.2742, "step": 13080 }, { "epoch": 2.05, "learning_rate": 1.1786666666666668e-05, "loss": 0.3523, "step": 13090 }, { "epoch": 2.05, "learning_rate": 1.1780392156862746e-05, "loss": 0.2954, "step": 13100 }, { "epoch": 2.06, "learning_rate": 1.1774117647058825e-05, "loss": 0.2097, "step": 13110 }, { "epoch": 2.06, "learning_rate": 1.1767843137254905e-05, "loss": 0.4294, "step": 13120 }, { "epoch": 2.06, "learning_rate": 1.176156862745098e-05, "loss": 0.5554, "step": 13130 }, { "epoch": 2.06, "learning_rate": 1.175529411764706e-05, "loss": 0.1561, "step": 13140 }, { "epoch": 2.06, "learning_rate": 1.1749019607843138e-05, "loss": 0.4669, "step": 13150 }, { "epoch": 2.06, "learning_rate": 1.1742745098039217e-05, "loss": 0.4059, "step": 13160 }, { "epoch": 2.07, "learning_rate": 1.1736470588235296e-05, "loss": 0.3406, "step": 13170 }, { "epoch": 2.07, "learning_rate": 1.1730196078431372e-05, "loss": 0.4243, "step": 13180 }, { "epoch": 2.07, "learning_rate": 1.1723921568627452e-05, "loss": 0.4783, "step": 13190 }, { "epoch": 2.07, "learning_rate": 1.171764705882353e-05, "loss": 0.2876, "step": 13200 }, { "epoch": 2.07, "learning_rate": 1.1711372549019608e-05, "loss": 0.5244, "step": 13210 }, { "epoch": 2.07, "learning_rate": 1.1705098039215688e-05, "loss": 0.2988, "step": 13220 }, { "epoch": 2.08, "learning_rate": 1.1698823529411765e-05, "loss": 0.3589, "step": 13230 }, { "epoch": 2.08, "learning_rate": 1.1692549019607845e-05, "loss": 0.5624, "step": 13240 }, { "epoch": 2.08, "learning_rate": 1.1686274509803922e-05, "loss": 0.497, "step": 13250 }, { "epoch": 2.08, "learning_rate": 1.168e-05, "loss": 0.4, "step": 13260 }, { "epoch": 2.08, "learning_rate": 1.167372549019608e-05, "loss": 0.3746, "step": 13270 }, { "epoch": 2.08, "learning_rate": 1.1667450980392159e-05, "loss": 0.5904, "step": 13280 }, { "epoch": 2.08, "learning_rate": 1.1661176470588236e-05, "loss": 0.4584, "step": 13290 }, { "epoch": 2.09, "learning_rate": 1.1654901960784316e-05, "loss": 0.3455, "step": 13300 }, { "epoch": 2.09, "learning_rate": 1.1648627450980392e-05, "loss": 0.3335, "step": 13310 }, { "epoch": 2.09, "learning_rate": 1.164235294117647e-05, "loss": 0.2819, "step": 13320 }, { "epoch": 2.09, "learning_rate": 1.163607843137255e-05, "loss": 0.3565, "step": 13330 }, { "epoch": 2.09, "learning_rate": 1.1629803921568628e-05, "loss": 0.6155, "step": 13340 }, { "epoch": 2.09, "learning_rate": 1.1623529411764707e-05, "loss": 0.3931, "step": 13350 }, { "epoch": 2.1, "learning_rate": 1.1617254901960786e-05, "loss": 0.3741, "step": 13360 }, { "epoch": 2.1, "learning_rate": 1.1610980392156862e-05, "loss": 0.401, "step": 13370 }, { "epoch": 2.1, "learning_rate": 1.1604705882352942e-05, "loss": 0.3142, "step": 13380 }, { "epoch": 2.1, "learning_rate": 1.1598431372549021e-05, "loss": 0.2936, "step": 13390 }, { "epoch": 2.1, "learning_rate": 1.1592156862745099e-05, "loss": 0.325, "step": 13400 }, { "epoch": 2.1, "learning_rate": 1.1585882352941178e-05, "loss": 0.54, "step": 13410 }, { "epoch": 2.11, "learning_rate": 1.1579607843137256e-05, "loss": 0.2909, "step": 13420 }, { "epoch": 2.11, "learning_rate": 1.1573333333333335e-05, "loss": 0.39, "step": 13430 }, { "epoch": 2.11, "learning_rate": 1.1567058823529414e-05, "loss": 0.4796, "step": 13440 }, { "epoch": 2.11, "learning_rate": 1.156078431372549e-05, "loss": 0.2456, "step": 13450 }, { "epoch": 2.11, "learning_rate": 1.155450980392157e-05, "loss": 0.4366, "step": 13460 }, { "epoch": 2.11, "learning_rate": 1.1548235294117649e-05, "loss": 0.3972, "step": 13470 }, { "epoch": 2.11, "learning_rate": 1.1541960784313726e-05, "loss": 0.468, "step": 13480 }, { "epoch": 2.12, "learning_rate": 1.1535686274509806e-05, "loss": 0.3053, "step": 13490 }, { "epoch": 2.12, "learning_rate": 1.1529411764705882e-05, "loss": 0.2489, "step": 13500 }, { "epoch": 2.12, "learning_rate": 1.1523137254901961e-05, "loss": 0.4202, "step": 13510 }, { "epoch": 2.12, "learning_rate": 1.151686274509804e-05, "loss": 0.4136, "step": 13520 }, { "epoch": 2.12, "learning_rate": 1.1510588235294118e-05, "loss": 0.3156, "step": 13530 }, { "epoch": 2.12, "learning_rate": 1.1504313725490197e-05, "loss": 0.3633, "step": 13540 }, { "epoch": 2.13, "learning_rate": 1.1498039215686277e-05, "loss": 0.2734, "step": 13550 }, { "epoch": 2.13, "learning_rate": 1.1491764705882353e-05, "loss": 0.3648, "step": 13560 }, { "epoch": 2.13, "learning_rate": 1.1485490196078432e-05, "loss": 0.2979, "step": 13570 }, { "epoch": 2.13, "learning_rate": 1.147921568627451e-05, "loss": 0.5242, "step": 13580 }, { "epoch": 2.13, "learning_rate": 1.1472941176470589e-05, "loss": 0.3455, "step": 13590 }, { "epoch": 2.13, "learning_rate": 1.1466666666666668e-05, "loss": 0.3223, "step": 13600 }, { "epoch": 2.13, "learning_rate": 1.1460392156862746e-05, "loss": 0.3784, "step": 13610 }, { "epoch": 2.14, "learning_rate": 1.1454117647058825e-05, "loss": 0.2115, "step": 13620 }, { "epoch": 2.14, "learning_rate": 1.1447843137254904e-05, "loss": 0.3401, "step": 13630 }, { "epoch": 2.14, "learning_rate": 1.144156862745098e-05, "loss": 0.3074, "step": 13640 }, { "epoch": 2.14, "learning_rate": 1.143529411764706e-05, "loss": 0.312, "step": 13650 }, { "epoch": 2.14, "learning_rate": 1.1429019607843139e-05, "loss": 0.2944, "step": 13660 }, { "epoch": 2.14, "learning_rate": 1.1422745098039217e-05, "loss": 0.1851, "step": 13670 }, { "epoch": 2.15, "learning_rate": 1.1416470588235296e-05, "loss": 0.3709, "step": 13680 }, { "epoch": 2.15, "learning_rate": 1.1410196078431372e-05, "loss": 0.3869, "step": 13690 }, { "epoch": 2.15, "learning_rate": 1.1403921568627451e-05, "loss": 0.5517, "step": 13700 }, { "epoch": 2.15, "learning_rate": 1.139764705882353e-05, "loss": 0.5065, "step": 13710 }, { "epoch": 2.15, "learning_rate": 1.1391372549019608e-05, "loss": 0.3628, "step": 13720 }, { "epoch": 2.15, "learning_rate": 1.1385098039215687e-05, "loss": 0.4072, "step": 13730 }, { "epoch": 2.16, "learning_rate": 1.1378823529411767e-05, "loss": 0.192, "step": 13740 }, { "epoch": 2.16, "learning_rate": 1.1372549019607844e-05, "loss": 0.2416, "step": 13750 }, { "epoch": 2.16, "learning_rate": 1.1366274509803924e-05, "loss": 0.1988, "step": 13760 }, { "epoch": 2.16, "learning_rate": 1.136e-05, "loss": 0.4047, "step": 13770 }, { "epoch": 2.16, "learning_rate": 1.1353725490196079e-05, "loss": 0.2843, "step": 13780 }, { "epoch": 2.16, "learning_rate": 1.1347450980392158e-05, "loss": 0.3676, "step": 13790 }, { "epoch": 2.16, "learning_rate": 1.1341176470588236e-05, "loss": 0.3623, "step": 13800 }, { "epoch": 2.17, "learning_rate": 1.1334901960784315e-05, "loss": 0.3358, "step": 13810 }, { "epoch": 2.17, "learning_rate": 1.1328627450980395e-05, "loss": 0.769, "step": 13820 }, { "epoch": 2.17, "learning_rate": 1.132235294117647e-05, "loss": 0.2401, "step": 13830 }, { "epoch": 2.17, "learning_rate": 1.131607843137255e-05, "loss": 0.3726, "step": 13840 }, { "epoch": 2.17, "learning_rate": 1.1309803921568627e-05, "loss": 0.3603, "step": 13850 }, { "epoch": 2.17, "learning_rate": 1.1303529411764707e-05, "loss": 0.4292, "step": 13860 }, { "epoch": 2.18, "learning_rate": 1.1297254901960786e-05, "loss": 0.3748, "step": 13870 }, { "epoch": 2.18, "learning_rate": 1.1290980392156862e-05, "loss": 0.3486, "step": 13880 }, { "epoch": 2.18, "learning_rate": 1.1284705882352941e-05, "loss": 0.2039, "step": 13890 }, { "epoch": 2.18, "learning_rate": 1.127843137254902e-05, "loss": 0.2692, "step": 13900 }, { "epoch": 2.18, "learning_rate": 1.1272156862745098e-05, "loss": 0.3658, "step": 13910 }, { "epoch": 2.18, "learning_rate": 1.1265882352941178e-05, "loss": 0.3857, "step": 13920 }, { "epoch": 2.19, "learning_rate": 1.1259607843137257e-05, "loss": 0.2601, "step": 13930 }, { "epoch": 2.19, "learning_rate": 1.1253333333333335e-05, "loss": 0.2853, "step": 13940 }, { "epoch": 2.19, "learning_rate": 1.1247058823529414e-05, "loss": 0.3517, "step": 13950 }, { "epoch": 2.19, "learning_rate": 1.124078431372549e-05, "loss": 0.1509, "step": 13960 }, { "epoch": 2.19, "learning_rate": 1.1234509803921569e-05, "loss": 0.3203, "step": 13970 }, { "epoch": 2.19, "learning_rate": 1.1228235294117648e-05, "loss": 0.4246, "step": 13980 }, { "epoch": 2.19, "learning_rate": 1.1221960784313726e-05, "loss": 0.4574, "step": 13990 }, { "epoch": 2.2, "learning_rate": 1.1215686274509805e-05, "loss": 0.2144, "step": 14000 }, { "epoch": 2.2, "learning_rate": 1.1209411764705885e-05, "loss": 0.2308, "step": 14010 }, { "epoch": 2.2, "learning_rate": 1.120313725490196e-05, "loss": 0.6192, "step": 14020 }, { "epoch": 2.2, "learning_rate": 1.119686274509804e-05, "loss": 0.3516, "step": 14030 }, { "epoch": 2.2, "learning_rate": 1.1190588235294118e-05, "loss": 0.4589, "step": 14040 }, { "epoch": 2.2, "learning_rate": 1.1184313725490197e-05, "loss": 0.2692, "step": 14050 }, { "epoch": 2.21, "learning_rate": 1.1178039215686276e-05, "loss": 0.3792, "step": 14060 }, { "epoch": 2.21, "learning_rate": 1.1171764705882354e-05, "loss": 0.5552, "step": 14070 }, { "epoch": 2.21, "learning_rate": 1.1165490196078432e-05, "loss": 0.1431, "step": 14080 }, { "epoch": 2.21, "learning_rate": 1.1159215686274511e-05, "loss": 0.2926, "step": 14090 }, { "epoch": 2.21, "learning_rate": 1.1152941176470588e-05, "loss": 0.1476, "step": 14100 }, { "epoch": 2.21, "learning_rate": 1.1146666666666668e-05, "loss": 0.3713, "step": 14110 }, { "epoch": 2.21, "learning_rate": 1.1140392156862747e-05, "loss": 0.2045, "step": 14120 }, { "epoch": 2.22, "learning_rate": 1.1134117647058825e-05, "loss": 0.2488, "step": 14130 }, { "epoch": 2.22, "learning_rate": 1.1127843137254904e-05, "loss": 0.4876, "step": 14140 }, { "epoch": 2.22, "learning_rate": 1.112156862745098e-05, "loss": 0.525, "step": 14150 }, { "epoch": 2.22, "learning_rate": 1.111529411764706e-05, "loss": 0.2542, "step": 14160 }, { "epoch": 2.22, "learning_rate": 1.1109019607843139e-05, "loss": 0.2802, "step": 14170 }, { "epoch": 2.22, "learning_rate": 1.1102745098039216e-05, "loss": 0.2729, "step": 14180 }, { "epoch": 2.23, "learning_rate": 1.1096470588235296e-05, "loss": 0.2511, "step": 14190 }, { "epoch": 2.23, "learning_rate": 1.1090196078431375e-05, "loss": 0.4058, "step": 14200 }, { "epoch": 2.23, "learning_rate": 1.1083921568627451e-05, "loss": 0.5665, "step": 14210 }, { "epoch": 2.23, "learning_rate": 1.107764705882353e-05, "loss": 0.5177, "step": 14220 }, { "epoch": 2.23, "learning_rate": 1.1071372549019608e-05, "loss": 0.2111, "step": 14230 }, { "epoch": 2.23, "learning_rate": 1.1065098039215687e-05, "loss": 0.1996, "step": 14240 }, { "epoch": 2.24, "learning_rate": 1.1058823529411766e-05, "loss": 0.5164, "step": 14250 }, { "epoch": 2.24, "learning_rate": 1.1052549019607844e-05, "loss": 0.286, "step": 14260 }, { "epoch": 2.24, "learning_rate": 1.1046274509803923e-05, "loss": 0.1127, "step": 14270 }, { "epoch": 2.24, "learning_rate": 1.1040000000000001e-05, "loss": 0.4524, "step": 14280 }, { "epoch": 2.24, "learning_rate": 1.1033725490196079e-05, "loss": 0.1723, "step": 14290 }, { "epoch": 2.24, "learning_rate": 1.1027450980392158e-05, "loss": 0.396, "step": 14300 }, { "epoch": 2.24, "learning_rate": 1.1021176470588236e-05, "loss": 0.3337, "step": 14310 }, { "epoch": 2.25, "learning_rate": 1.1014901960784315e-05, "loss": 0.2836, "step": 14320 }, { "epoch": 2.25, "learning_rate": 1.1008627450980394e-05, "loss": 0.2952, "step": 14330 }, { "epoch": 2.25, "learning_rate": 1.100235294117647e-05, "loss": 0.3604, "step": 14340 }, { "epoch": 2.25, "learning_rate": 1.099607843137255e-05, "loss": 0.3171, "step": 14350 }, { "epoch": 2.25, "learning_rate": 1.0989803921568629e-05, "loss": 0.3354, "step": 14360 }, { "epoch": 2.25, "learning_rate": 1.0983529411764706e-05, "loss": 0.3074, "step": 14370 }, { "epoch": 2.26, "learning_rate": 1.0977254901960786e-05, "loss": 0.3708, "step": 14380 }, { "epoch": 2.26, "learning_rate": 1.0970980392156865e-05, "loss": 0.2815, "step": 14390 }, { "epoch": 2.26, "learning_rate": 1.0964705882352941e-05, "loss": 0.4066, "step": 14400 }, { "epoch": 2.26, "learning_rate": 1.095843137254902e-05, "loss": 0.2553, "step": 14410 }, { "epoch": 2.26, "learning_rate": 1.0952156862745098e-05, "loss": 0.4058, "step": 14420 }, { "epoch": 2.26, "learning_rate": 1.0945882352941177e-05, "loss": 0.2896, "step": 14430 }, { "epoch": 2.27, "learning_rate": 1.0939607843137257e-05, "loss": 0.4801, "step": 14440 }, { "epoch": 2.27, "learning_rate": 1.0933333333333334e-05, "loss": 0.4578, "step": 14450 }, { "epoch": 2.27, "learning_rate": 1.0927058823529414e-05, "loss": 0.2777, "step": 14460 }, { "epoch": 2.27, "learning_rate": 1.0920784313725493e-05, "loss": 0.4983, "step": 14470 }, { "epoch": 2.27, "learning_rate": 1.0914509803921569e-05, "loss": 0.7186, "step": 14480 }, { "epoch": 2.27, "learning_rate": 1.0908235294117648e-05, "loss": 0.4323, "step": 14490 }, { "epoch": 2.27, "learning_rate": 1.0901960784313726e-05, "loss": 0.3291, "step": 14500 }, { "epoch": 2.28, "learning_rate": 1.0895686274509805e-05, "loss": 0.3589, "step": 14510 }, { "epoch": 2.28, "learning_rate": 1.0889411764705884e-05, "loss": 0.3561, "step": 14520 }, { "epoch": 2.28, "learning_rate": 1.088313725490196e-05, "loss": 0.3464, "step": 14530 }, { "epoch": 2.28, "learning_rate": 1.087686274509804e-05, "loss": 0.2743, "step": 14540 }, { "epoch": 2.28, "learning_rate": 1.0870588235294119e-05, "loss": 0.4668, "step": 14550 }, { "epoch": 2.28, "learning_rate": 1.0864313725490197e-05, "loss": 0.4439, "step": 14560 }, { "epoch": 2.29, "learning_rate": 1.0858039215686276e-05, "loss": 0.4195, "step": 14570 }, { "epoch": 2.29, "learning_rate": 1.0851764705882354e-05, "loss": 0.3618, "step": 14580 }, { "epoch": 2.29, "learning_rate": 1.0845490196078431e-05, "loss": 0.3317, "step": 14590 }, { "epoch": 2.29, "learning_rate": 1.083921568627451e-05, "loss": 0.278, "step": 14600 }, { "epoch": 2.29, "learning_rate": 1.0832941176470588e-05, "loss": 0.577, "step": 14610 }, { "epoch": 2.29, "learning_rate": 1.0826666666666667e-05, "loss": 0.3288, "step": 14620 }, { "epoch": 2.29, "learning_rate": 1.0820392156862747e-05, "loss": 0.2406, "step": 14630 }, { "epoch": 2.3, "learning_rate": 1.0814117647058824e-05, "loss": 0.2032, "step": 14640 }, { "epoch": 2.3, "learning_rate": 1.0807843137254904e-05, "loss": 0.6748, "step": 14650 }, { "epoch": 2.3, "learning_rate": 1.0801568627450983e-05, "loss": 0.1761, "step": 14660 }, { "epoch": 2.3, "learning_rate": 1.0795294117647059e-05, "loss": 0.2052, "step": 14670 }, { "epoch": 2.3, "learning_rate": 1.0789019607843138e-05, "loss": 0.2884, "step": 14680 }, { "epoch": 2.3, "learning_rate": 1.0782745098039216e-05, "loss": 0.2332, "step": 14690 }, { "epoch": 2.31, "learning_rate": 1.0776470588235295e-05, "loss": 0.2615, "step": 14700 }, { "epoch": 2.31, "learning_rate": 1.0770196078431375e-05, "loss": 0.5669, "step": 14710 }, { "epoch": 2.31, "learning_rate": 1.076392156862745e-05, "loss": 0.4049, "step": 14720 }, { "epoch": 2.31, "learning_rate": 1.075764705882353e-05, "loss": 0.2578, "step": 14730 }, { "epoch": 2.31, "learning_rate": 1.075137254901961e-05, "loss": 0.2976, "step": 14740 }, { "epoch": 2.31, "learning_rate": 1.0745098039215687e-05, "loss": 0.555, "step": 14750 }, { "epoch": 2.32, "learning_rate": 1.0738823529411766e-05, "loss": 0.3478, "step": 14760 }, { "epoch": 2.32, "learning_rate": 1.0732549019607844e-05, "loss": 0.3096, "step": 14770 }, { "epoch": 2.32, "learning_rate": 1.0726274509803923e-05, "loss": 0.4688, "step": 14780 }, { "epoch": 2.32, "learning_rate": 1.072e-05, "loss": 0.258, "step": 14790 }, { "epoch": 2.32, "learning_rate": 1.0713725490196078e-05, "loss": 0.2816, "step": 14800 }, { "epoch": 2.32, "learning_rate": 1.0707450980392158e-05, "loss": 0.2776, "step": 14810 }, { "epoch": 2.32, "learning_rate": 1.0701176470588237e-05, "loss": 0.3605, "step": 14820 }, { "epoch": 2.33, "learning_rate": 1.0694901960784315e-05, "loss": 0.4212, "step": 14830 }, { "epoch": 2.33, "learning_rate": 1.0688627450980394e-05, "loss": 0.2498, "step": 14840 }, { "epoch": 2.33, "learning_rate": 1.068235294117647e-05, "loss": 0.4571, "step": 14850 }, { "epoch": 2.33, "learning_rate": 1.067607843137255e-05, "loss": 0.2332, "step": 14860 }, { "epoch": 2.33, "learning_rate": 1.0669803921568628e-05, "loss": 0.3177, "step": 14870 }, { "epoch": 2.33, "learning_rate": 1.0663529411764706e-05, "loss": 0.5041, "step": 14880 }, { "epoch": 2.34, "learning_rate": 1.0657254901960785e-05, "loss": 0.2441, "step": 14890 }, { "epoch": 2.34, "learning_rate": 1.0650980392156865e-05, "loss": 0.1064, "step": 14900 }, { "epoch": 2.34, "learning_rate": 1.064470588235294e-05, "loss": 0.4557, "step": 14910 }, { "epoch": 2.34, "learning_rate": 1.063843137254902e-05, "loss": 0.3112, "step": 14920 }, { "epoch": 2.34, "learning_rate": 1.06321568627451e-05, "loss": 0.3331, "step": 14930 }, { "epoch": 2.34, "learning_rate": 1.0625882352941177e-05, "loss": 0.3432, "step": 14940 }, { "epoch": 2.35, "learning_rate": 1.0619607843137256e-05, "loss": 0.3318, "step": 14950 }, { "epoch": 2.35, "learning_rate": 1.0613333333333334e-05, "loss": 0.6346, "step": 14960 }, { "epoch": 2.35, "learning_rate": 1.0607058823529413e-05, "loss": 0.2444, "step": 14970 }, { "epoch": 2.35, "learning_rate": 1.0600784313725493e-05, "loss": 0.1291, "step": 14980 }, { "epoch": 2.35, "learning_rate": 1.0594509803921569e-05, "loss": 0.2714, "step": 14990 }, { "epoch": 2.35, "learning_rate": 1.0588235294117648e-05, "loss": 0.4792, "step": 15000 }, { "epoch": 2.35, "learning_rate": 1.0581960784313727e-05, "loss": 0.4351, "step": 15010 }, { "epoch": 2.36, "learning_rate": 1.0575686274509805e-05, "loss": 0.628, "step": 15020 }, { "epoch": 2.36, "learning_rate": 1.0569411764705884e-05, "loss": 0.2169, "step": 15030 }, { "epoch": 2.36, "learning_rate": 1.056313725490196e-05, "loss": 0.1387, "step": 15040 }, { "epoch": 2.36, "learning_rate": 1.055686274509804e-05, "loss": 0.3273, "step": 15050 }, { "epoch": 2.36, "learning_rate": 1.0550588235294119e-05, "loss": 0.6752, "step": 15060 }, { "epoch": 2.36, "learning_rate": 1.0544313725490196e-05, "loss": 0.3007, "step": 15070 }, { "epoch": 2.37, "learning_rate": 1.0538039215686276e-05, "loss": 0.2582, "step": 15080 }, { "epoch": 2.37, "learning_rate": 1.0531764705882355e-05, "loss": 0.3072, "step": 15090 }, { "epoch": 2.37, "learning_rate": 1.0525490196078431e-05, "loss": 0.3586, "step": 15100 }, { "epoch": 2.37, "learning_rate": 1.051921568627451e-05, "loss": 0.1393, "step": 15110 }, { "epoch": 2.37, "learning_rate": 1.051294117647059e-05, "loss": 0.3309, "step": 15120 }, { "epoch": 2.37, "learning_rate": 1.0506666666666667e-05, "loss": 0.3292, "step": 15130 }, { "epoch": 2.37, "learning_rate": 1.0500392156862746e-05, "loss": 0.3782, "step": 15140 }, { "epoch": 2.38, "learning_rate": 1.0494117647058824e-05, "loss": 0.1466, "step": 15150 }, { "epoch": 2.38, "learning_rate": 1.0487843137254903e-05, "loss": 0.356, "step": 15160 }, { "epoch": 2.38, "learning_rate": 1.0481568627450983e-05, "loss": 0.27, "step": 15170 }, { "epoch": 2.38, "learning_rate": 1.0475294117647059e-05, "loss": 0.4566, "step": 15180 }, { "epoch": 2.38, "learning_rate": 1.0469019607843138e-05, "loss": 0.2929, "step": 15190 }, { "epoch": 2.38, "learning_rate": 1.0462745098039217e-05, "loss": 0.2115, "step": 15200 }, { "epoch": 2.39, "learning_rate": 1.0456470588235295e-05, "loss": 0.323, "step": 15210 }, { "epoch": 2.39, "learning_rate": 1.0450196078431374e-05, "loss": 0.393, "step": 15220 }, { "epoch": 2.39, "learning_rate": 1.044392156862745e-05, "loss": 0.4537, "step": 15230 }, { "epoch": 2.39, "learning_rate": 1.043764705882353e-05, "loss": 0.3423, "step": 15240 }, { "epoch": 2.39, "learning_rate": 1.0431372549019609e-05, "loss": 0.197, "step": 15250 }, { "epoch": 2.39, "learning_rate": 1.0425098039215686e-05, "loss": 0.2092, "step": 15260 }, { "epoch": 2.4, "learning_rate": 1.0418823529411766e-05, "loss": 0.2906, "step": 15270 }, { "epoch": 2.4, "learning_rate": 1.0412549019607845e-05, "loss": 0.1557, "step": 15280 }, { "epoch": 2.4, "learning_rate": 1.0406274509803923e-05, "loss": 0.269, "step": 15290 }, { "epoch": 2.4, "learning_rate": 1.04e-05, "loss": 0.3151, "step": 15300 }, { "epoch": 2.4, "learning_rate": 1.0393725490196078e-05, "loss": 0.4001, "step": 15310 }, { "epoch": 2.4, "learning_rate": 1.0387450980392157e-05, "loss": 0.1382, "step": 15320 }, { "epoch": 2.4, "learning_rate": 1.0381176470588237e-05, "loss": 0.3138, "step": 15330 }, { "epoch": 2.41, "learning_rate": 1.0374901960784314e-05, "loss": 0.5191, "step": 15340 }, { "epoch": 2.41, "learning_rate": 1.0368627450980394e-05, "loss": 0.551, "step": 15350 }, { "epoch": 2.41, "learning_rate": 1.0362352941176473e-05, "loss": 0.521, "step": 15360 }, { "epoch": 2.41, "learning_rate": 1.0356078431372549e-05, "loss": 0.2413, "step": 15370 }, { "epoch": 2.41, "learning_rate": 1.0349803921568628e-05, "loss": 0.2602, "step": 15380 }, { "epoch": 2.41, "learning_rate": 1.0343529411764707e-05, "loss": 0.2863, "step": 15390 }, { "epoch": 2.42, "learning_rate": 1.0337254901960785e-05, "loss": 0.3858, "step": 15400 }, { "epoch": 2.42, "learning_rate": 1.0330980392156864e-05, "loss": 0.3681, "step": 15410 }, { "epoch": 2.42, "learning_rate": 1.032470588235294e-05, "loss": 0.3421, "step": 15420 }, { "epoch": 2.42, "learning_rate": 1.031843137254902e-05, "loss": 0.204, "step": 15430 }, { "epoch": 2.42, "learning_rate": 1.0312156862745099e-05, "loss": 0.39, "step": 15440 }, { "epoch": 2.42, "learning_rate": 1.0305882352941177e-05, "loss": 0.2958, "step": 15450 }, { "epoch": 2.43, "learning_rate": 1.0299607843137256e-05, "loss": 0.5264, "step": 15460 }, { "epoch": 2.43, "learning_rate": 1.0293333333333335e-05, "loss": 0.3299, "step": 15470 }, { "epoch": 2.43, "learning_rate": 1.0287058823529413e-05, "loss": 0.4705, "step": 15480 }, { "epoch": 2.43, "learning_rate": 1.0280784313725492e-05, "loss": 0.4241, "step": 15490 }, { "epoch": 2.43, "learning_rate": 1.0274509803921568e-05, "loss": 0.3819, "step": 15500 }, { "epoch": 2.43, "learning_rate": 1.0268235294117648e-05, "loss": 0.3528, "step": 15510 }, { "epoch": 2.43, "learning_rate": 1.0261960784313727e-05, "loss": 0.2655, "step": 15520 }, { "epoch": 2.44, "learning_rate": 1.0255686274509804e-05, "loss": 0.3596, "step": 15530 }, { "epoch": 2.44, "learning_rate": 1.0249411764705884e-05, "loss": 0.3838, "step": 15540 }, { "epoch": 2.44, "learning_rate": 1.0243137254901963e-05, "loss": 0.4005, "step": 15550 }, { "epoch": 2.44, "learning_rate": 1.0236862745098039e-05, "loss": 0.2307, "step": 15560 }, { "epoch": 2.44, "learning_rate": 1.0230588235294118e-05, "loss": 0.23, "step": 15570 }, { "epoch": 2.44, "learning_rate": 1.0224313725490196e-05, "loss": 0.5121, "step": 15580 }, { "epoch": 2.45, "learning_rate": 1.0218039215686275e-05, "loss": 0.3481, "step": 15590 }, { "epoch": 2.45, "learning_rate": 1.0211764705882355e-05, "loss": 0.3429, "step": 15600 }, { "epoch": 2.45, "learning_rate": 1.020549019607843e-05, "loss": 0.1136, "step": 15610 }, { "epoch": 2.45, "learning_rate": 1.019921568627451e-05, "loss": 0.3211, "step": 15620 }, { "epoch": 2.45, "learning_rate": 1.019294117647059e-05, "loss": 0.361, "step": 15630 }, { "epoch": 2.45, "learning_rate": 1.0186666666666667e-05, "loss": 0.5195, "step": 15640 }, { "epoch": 2.45, "learning_rate": 1.0180392156862746e-05, "loss": 0.2425, "step": 15650 }, { "epoch": 2.46, "learning_rate": 1.0174117647058825e-05, "loss": 0.3621, "step": 15660 }, { "epoch": 2.46, "learning_rate": 1.0167843137254903e-05, "loss": 0.1971, "step": 15670 }, { "epoch": 2.46, "learning_rate": 1.0161568627450982e-05, "loss": 0.2483, "step": 15680 }, { "epoch": 2.46, "learning_rate": 1.0155294117647058e-05, "loss": 0.5579, "step": 15690 }, { "epoch": 2.46, "learning_rate": 1.0149019607843138e-05, "loss": 0.4062, "step": 15700 }, { "epoch": 2.46, "learning_rate": 1.0142745098039217e-05, "loss": 0.3023, "step": 15710 }, { "epoch": 2.47, "learning_rate": 1.0136470588235295e-05, "loss": 0.4117, "step": 15720 }, { "epoch": 2.47, "learning_rate": 1.0130196078431374e-05, "loss": 0.2716, "step": 15730 }, { "epoch": 2.47, "learning_rate": 1.0123921568627453e-05, "loss": 0.2503, "step": 15740 }, { "epoch": 2.47, "learning_rate": 1.011764705882353e-05, "loss": 0.4701, "step": 15750 }, { "epoch": 2.47, "learning_rate": 1.0111372549019609e-05, "loss": 0.4577, "step": 15760 }, { "epoch": 2.47, "learning_rate": 1.0105098039215686e-05, "loss": 0.4197, "step": 15770 }, { "epoch": 2.48, "learning_rate": 1.0098823529411765e-05, "loss": 0.2147, "step": 15780 }, { "epoch": 2.48, "learning_rate": 1.0092549019607845e-05, "loss": 0.4119, "step": 15790 }, { "epoch": 2.48, "learning_rate": 1.0086274509803922e-05, "loss": 0.1328, "step": 15800 }, { "epoch": 2.48, "learning_rate": 1.008e-05, "loss": 0.488, "step": 15810 }, { "epoch": 2.48, "learning_rate": 1.007372549019608e-05, "loss": 0.4325, "step": 15820 }, { "epoch": 2.48, "learning_rate": 1.0067450980392157e-05, "loss": 0.4427, "step": 15830 }, { "epoch": 2.48, "learning_rate": 1.0061176470588236e-05, "loss": 0.47, "step": 15840 }, { "epoch": 2.49, "learning_rate": 1.0054901960784314e-05, "loss": 0.3384, "step": 15850 }, { "epoch": 2.49, "learning_rate": 1.0048627450980393e-05, "loss": 0.3875, "step": 15860 }, { "epoch": 2.49, "learning_rate": 1.0042352941176473e-05, "loss": 0.5334, "step": 15870 }, { "epoch": 2.49, "learning_rate": 1.0036078431372549e-05, "loss": 0.7144, "step": 15880 }, { "epoch": 2.49, "learning_rate": 1.0029803921568628e-05, "loss": 0.3149, "step": 15890 }, { "epoch": 2.49, "learning_rate": 1.0023529411764707e-05, "loss": 0.3114, "step": 15900 }, { "epoch": 2.5, "learning_rate": 1.0017254901960785e-05, "loss": 0.5961, "step": 15910 }, { "epoch": 2.5, "learning_rate": 1.0010980392156864e-05, "loss": 0.4315, "step": 15920 }, { "epoch": 2.5, "learning_rate": 1.0004705882352943e-05, "loss": 0.2637, "step": 15930 }, { "epoch": 2.5, "learning_rate": 9.99843137254902e-06, "loss": 0.4175, "step": 15940 }, { "epoch": 2.5, "learning_rate": 9.992156862745099e-06, "loss": 0.2068, "step": 15950 }, { "epoch": 2.5, "learning_rate": 9.985882352941178e-06, "loss": 0.3505, "step": 15960 }, { "epoch": 2.51, "learning_rate": 9.979607843137256e-06, "loss": 0.3309, "step": 15970 }, { "epoch": 2.51, "learning_rate": 9.973333333333333e-06, "loss": 0.464, "step": 15980 }, { "epoch": 2.51, "learning_rate": 9.967058823529413e-06, "loss": 0.41, "step": 15990 }, { "epoch": 2.51, "learning_rate": 9.960784313725492e-06, "loss": 0.2453, "step": 16000 }, { "epoch": 2.51, "learning_rate": 9.95450980392157e-06, "loss": 0.4763, "step": 16010 }, { "epoch": 2.51, "learning_rate": 9.948235294117647e-06, "loss": 0.4351, "step": 16020 }, { "epoch": 2.51, "learning_rate": 9.941960784313726e-06, "loss": 0.2585, "step": 16030 }, { "epoch": 2.52, "learning_rate": 9.935686274509804e-06, "loss": 0.3612, "step": 16040 }, { "epoch": 2.52, "learning_rate": 9.929411764705883e-06, "loss": 0.3547, "step": 16050 }, { "epoch": 2.52, "learning_rate": 9.923137254901961e-06, "loss": 0.296, "step": 16060 }, { "epoch": 2.52, "learning_rate": 9.91686274509804e-06, "loss": 0.5302, "step": 16070 }, { "epoch": 2.52, "learning_rate": 9.910588235294118e-06, "loss": 0.4031, "step": 16080 }, { "epoch": 2.52, "learning_rate": 9.904313725490197e-06, "loss": 0.4171, "step": 16090 }, { "epoch": 2.53, "learning_rate": 9.898039215686275e-06, "loss": 0.309, "step": 16100 }, { "epoch": 2.53, "learning_rate": 9.891764705882354e-06, "loss": 0.4506, "step": 16110 }, { "epoch": 2.53, "learning_rate": 9.885490196078432e-06, "loss": 0.276, "step": 16120 }, { "epoch": 2.53, "learning_rate": 9.87921568627451e-06, "loss": 0.2637, "step": 16130 }, { "epoch": 2.53, "learning_rate": 9.872941176470589e-06, "loss": 0.213, "step": 16140 }, { "epoch": 2.53, "learning_rate": 9.866666666666668e-06, "loss": 0.2998, "step": 16150 }, { "epoch": 2.53, "learning_rate": 9.860392156862746e-06, "loss": 0.3159, "step": 16160 }, { "epoch": 2.54, "learning_rate": 9.854117647058823e-06, "loss": 0.4187, "step": 16170 }, { "epoch": 2.54, "learning_rate": 9.847843137254903e-06, "loss": 0.3742, "step": 16180 }, { "epoch": 2.54, "learning_rate": 9.841568627450982e-06, "loss": 0.2718, "step": 16190 }, { "epoch": 2.54, "learning_rate": 9.83529411764706e-06, "loss": 0.5147, "step": 16200 }, { "epoch": 2.54, "learning_rate": 9.829019607843137e-06, "loss": 0.4929, "step": 16210 }, { "epoch": 2.54, "learning_rate": 9.822745098039217e-06, "loss": 0.4976, "step": 16220 }, { "epoch": 2.55, "learning_rate": 9.816470588235294e-06, "loss": 0.5006, "step": 16230 }, { "epoch": 2.55, "learning_rate": 9.810196078431374e-06, "loss": 0.3335, "step": 16240 }, { "epoch": 2.55, "learning_rate": 9.803921568627451e-06, "loss": 0.3526, "step": 16250 }, { "epoch": 2.55, "learning_rate": 9.79764705882353e-06, "loss": 0.3012, "step": 16260 }, { "epoch": 2.55, "learning_rate": 9.791372549019608e-06, "loss": 0.2213, "step": 16270 }, { "epoch": 2.55, "learning_rate": 9.785098039215688e-06, "loss": 0.3931, "step": 16280 }, { "epoch": 2.56, "learning_rate": 9.778823529411765e-06, "loss": 0.4125, "step": 16290 }, { "epoch": 2.56, "learning_rate": 9.772549019607844e-06, "loss": 0.4107, "step": 16300 }, { "epoch": 2.56, "learning_rate": 9.766274509803922e-06, "loss": 0.2403, "step": 16310 }, { "epoch": 2.56, "learning_rate": 9.760000000000001e-06, "loss": 0.289, "step": 16320 }, { "epoch": 2.56, "learning_rate": 9.753725490196079e-06, "loss": 0.3813, "step": 16330 }, { "epoch": 2.56, "learning_rate": 9.747450980392158e-06, "loss": 0.3611, "step": 16340 }, { "epoch": 2.56, "learning_rate": 9.741176470588236e-06, "loss": 0.3266, "step": 16350 }, { "epoch": 2.57, "learning_rate": 9.734901960784314e-06, "loss": 0.455, "step": 16360 }, { "epoch": 2.57, "learning_rate": 9.728627450980393e-06, "loss": 0.4048, "step": 16370 }, { "epoch": 2.57, "learning_rate": 9.722352941176472e-06, "loss": 0.4067, "step": 16380 }, { "epoch": 2.57, "learning_rate": 9.71607843137255e-06, "loss": 0.322, "step": 16390 }, { "epoch": 2.57, "learning_rate": 9.709803921568628e-06, "loss": 0.4528, "step": 16400 }, { "epoch": 2.57, "learning_rate": 9.703529411764707e-06, "loss": 0.2736, "step": 16410 }, { "epoch": 2.58, "learning_rate": 9.697254901960786e-06, "loss": 0.2843, "step": 16420 }, { "epoch": 2.58, "learning_rate": 9.690980392156864e-06, "loss": 0.3208, "step": 16430 }, { "epoch": 2.58, "learning_rate": 9.684705882352941e-06, "loss": 0.2969, "step": 16440 }, { "epoch": 2.58, "learning_rate": 9.67843137254902e-06, "loss": 0.2669, "step": 16450 }, { "epoch": 2.58, "learning_rate": 9.672156862745098e-06, "loss": 0.3356, "step": 16460 }, { "epoch": 2.58, "learning_rate": 9.665882352941178e-06, "loss": 0.3641, "step": 16470 }, { "epoch": 2.59, "learning_rate": 9.659607843137255e-06, "loss": 0.2104, "step": 16480 }, { "epoch": 2.59, "learning_rate": 9.653333333333335e-06, "loss": 0.2775, "step": 16490 }, { "epoch": 2.59, "learning_rate": 9.647058823529412e-06, "loss": 0.3053, "step": 16500 }, { "epoch": 2.59, "learning_rate": 9.640784313725492e-06, "loss": 0.3016, "step": 16510 }, { "epoch": 2.59, "learning_rate": 9.63450980392157e-06, "loss": 0.4148, "step": 16520 }, { "epoch": 2.59, "learning_rate": 9.628235294117649e-06, "loss": 0.2485, "step": 16530 }, { "epoch": 2.59, "learning_rate": 9.621960784313726e-06, "loss": 0.3078, "step": 16540 }, { "epoch": 2.6, "learning_rate": 9.615686274509804e-06, "loss": 0.3147, "step": 16550 }, { "epoch": 2.6, "learning_rate": 9.609411764705883e-06, "loss": 0.4572, "step": 16560 }, { "epoch": 2.6, "learning_rate": 9.603137254901962e-06, "loss": 0.2724, "step": 16570 }, { "epoch": 2.6, "learning_rate": 9.59686274509804e-06, "loss": 0.3283, "step": 16580 }, { "epoch": 2.6, "learning_rate": 9.590588235294118e-06, "loss": 0.371, "step": 16590 }, { "epoch": 2.6, "learning_rate": 9.584313725490197e-06, "loss": 0.49, "step": 16600 }, { "epoch": 2.61, "learning_rate": 9.578039215686276e-06, "loss": 0.4166, "step": 16610 }, { "epoch": 2.61, "learning_rate": 9.571764705882354e-06, "loss": 0.2415, "step": 16620 }, { "epoch": 2.61, "learning_rate": 9.565490196078432e-06, "loss": 0.382, "step": 16630 }, { "epoch": 2.61, "learning_rate": 9.559215686274511e-06, "loss": 0.3109, "step": 16640 }, { "epoch": 2.61, "learning_rate": 9.552941176470589e-06, "loss": 0.3005, "step": 16650 }, { "epoch": 2.61, "learning_rate": 9.546666666666668e-06, "loss": 0.4489, "step": 16660 }, { "epoch": 2.61, "learning_rate": 9.540392156862746e-06, "loss": 0.2348, "step": 16670 }, { "epoch": 2.62, "learning_rate": 9.534117647058825e-06, "loss": 0.3815, "step": 16680 }, { "epoch": 2.62, "learning_rate": 9.527843137254902e-06, "loss": 0.2291, "step": 16690 }, { "epoch": 2.62, "learning_rate": 9.521568627450982e-06, "loss": 0.4113, "step": 16700 }, { "epoch": 2.62, "learning_rate": 9.51529411764706e-06, "loss": 0.1742, "step": 16710 }, { "epoch": 2.62, "learning_rate": 9.509019607843139e-06, "loss": 0.3357, "step": 16720 }, { "epoch": 2.62, "learning_rate": 9.502745098039216e-06, "loss": 0.1955, "step": 16730 }, { "epoch": 2.63, "learning_rate": 9.496470588235294e-06, "loss": 0.2106, "step": 16740 }, { "epoch": 2.63, "learning_rate": 9.490196078431373e-06, "loss": 0.3124, "step": 16750 }, { "epoch": 2.63, "learning_rate": 9.483921568627453e-06, "loss": 0.1505, "step": 16760 }, { "epoch": 2.63, "learning_rate": 9.47764705882353e-06, "loss": 0.3812, "step": 16770 }, { "epoch": 2.63, "learning_rate": 9.471372549019608e-06, "loss": 0.2272, "step": 16780 }, { "epoch": 2.63, "learning_rate": 9.465098039215687e-06, "loss": 0.5445, "step": 16790 }, { "epoch": 2.64, "learning_rate": 9.458823529411767e-06, "loss": 0.3409, "step": 16800 }, { "epoch": 2.64, "learning_rate": 9.452549019607844e-06, "loss": 0.3686, "step": 16810 }, { "epoch": 2.64, "learning_rate": 9.446274509803922e-06, "loss": 0.2379, "step": 16820 }, { "epoch": 2.64, "learning_rate": 9.440000000000001e-06, "loss": 0.2819, "step": 16830 }, { "epoch": 2.64, "learning_rate": 9.433725490196079e-06, "loss": 0.2623, "step": 16840 }, { "epoch": 2.64, "learning_rate": 9.427450980392158e-06, "loss": 0.3107, "step": 16850 }, { "epoch": 2.64, "learning_rate": 9.421176470588236e-06, "loss": 0.2158, "step": 16860 }, { "epoch": 2.65, "learning_rate": 9.414901960784313e-06, "loss": 0.5024, "step": 16870 }, { "epoch": 2.65, "learning_rate": 9.408627450980393e-06, "loss": 0.3543, "step": 16880 }, { "epoch": 2.65, "learning_rate": 9.402352941176472e-06, "loss": 0.5014, "step": 16890 }, { "epoch": 2.65, "learning_rate": 9.39607843137255e-06, "loss": 0.3221, "step": 16900 }, { "epoch": 2.65, "learning_rate": 9.389803921568629e-06, "loss": 0.1757, "step": 16910 }, { "epoch": 2.65, "learning_rate": 9.383529411764707e-06, "loss": 0.2237, "step": 16920 }, { "epoch": 2.66, "learning_rate": 9.377254901960786e-06, "loss": 0.301, "step": 16930 }, { "epoch": 2.66, "learning_rate": 9.370980392156863e-06, "loss": 0.1763, "step": 16940 }, { "epoch": 2.66, "learning_rate": 9.364705882352943e-06, "loss": 0.1997, "step": 16950 }, { "epoch": 2.66, "learning_rate": 9.35843137254902e-06, "loss": 0.1699, "step": 16960 }, { "epoch": 2.66, "learning_rate": 9.352156862745098e-06, "loss": 0.2393, "step": 16970 }, { "epoch": 2.66, "learning_rate": 9.345882352941177e-06, "loss": 0.3898, "step": 16980 }, { "epoch": 2.67, "learning_rate": 9.339607843137257e-06, "loss": 0.2809, "step": 16990 }, { "epoch": 2.67, "learning_rate": 9.333333333333334e-06, "loss": 0.3397, "step": 17000 }, { "epoch": 2.67, "learning_rate": 9.327058823529412e-06, "loss": 0.4529, "step": 17010 }, { "epoch": 2.67, "learning_rate": 9.320784313725491e-06, "loss": 0.3092, "step": 17020 }, { "epoch": 2.67, "learning_rate": 9.31450980392157e-06, "loss": 0.4513, "step": 17030 }, { "epoch": 2.67, "learning_rate": 9.308235294117648e-06, "loss": 0.2748, "step": 17040 }, { "epoch": 2.67, "learning_rate": 9.301960784313726e-06, "loss": 0.415, "step": 17050 }, { "epoch": 2.68, "learning_rate": 9.295686274509803e-06, "loss": 0.3642, "step": 17060 }, { "epoch": 2.68, "learning_rate": 9.289411764705883e-06, "loss": 0.4075, "step": 17070 }, { "epoch": 2.68, "learning_rate": 9.283137254901962e-06, "loss": 0.3108, "step": 17080 }, { "epoch": 2.68, "learning_rate": 9.27686274509804e-06, "loss": 0.3896, "step": 17090 }, { "epoch": 2.68, "learning_rate": 9.270588235294117e-06, "loss": 0.4337, "step": 17100 }, { "epoch": 2.68, "learning_rate": 9.264313725490197e-06, "loss": 0.1986, "step": 17110 }, { "epoch": 2.69, "learning_rate": 9.258039215686276e-06, "loss": 0.4211, "step": 17120 }, { "epoch": 2.69, "learning_rate": 9.251764705882354e-06, "loss": 0.4288, "step": 17130 }, { "epoch": 2.69, "learning_rate": 9.245490196078433e-06, "loss": 0.4626, "step": 17140 }, { "epoch": 2.69, "learning_rate": 9.23921568627451e-06, "loss": 0.2967, "step": 17150 }, { "epoch": 2.69, "learning_rate": 9.232941176470588e-06, "loss": 0.2951, "step": 17160 }, { "epoch": 2.69, "learning_rate": 9.226666666666668e-06, "loss": 0.2076, "step": 17170 }, { "epoch": 2.69, "learning_rate": 9.220392156862747e-06, "loss": 0.2957, "step": 17180 }, { "epoch": 2.7, "learning_rate": 9.214117647058824e-06, "loss": 0.5631, "step": 17190 }, { "epoch": 2.7, "learning_rate": 9.207843137254902e-06, "loss": 0.6402, "step": 17200 }, { "epoch": 2.7, "learning_rate": 9.201568627450981e-06, "loss": 0.3443, "step": 17210 }, { "epoch": 2.7, "learning_rate": 9.19529411764706e-06, "loss": 0.2822, "step": 17220 }, { "epoch": 2.7, "learning_rate": 9.189019607843138e-06, "loss": 0.3522, "step": 17230 }, { "epoch": 2.7, "learning_rate": 9.182745098039216e-06, "loss": 0.2387, "step": 17240 }, { "epoch": 2.71, "learning_rate": 9.176470588235294e-06, "loss": 0.3309, "step": 17250 }, { "epoch": 2.71, "learning_rate": 9.170196078431373e-06, "loss": 0.3627, "step": 17260 }, { "epoch": 2.71, "learning_rate": 9.163921568627452e-06, "loss": 0.2738, "step": 17270 }, { "epoch": 2.71, "learning_rate": 9.15764705882353e-06, "loss": 0.2509, "step": 17280 }, { "epoch": 2.71, "learning_rate": 9.151372549019608e-06, "loss": 0.2328, "step": 17290 }, { "epoch": 2.71, "learning_rate": 9.145098039215687e-06, "loss": 0.2605, "step": 17300 }, { "epoch": 2.72, "learning_rate": 9.138823529411766e-06, "loss": 0.491, "step": 17310 }, { "epoch": 2.72, "learning_rate": 9.132549019607844e-06, "loss": 0.5577, "step": 17320 }, { "epoch": 2.72, "learning_rate": 9.126274509803921e-06, "loss": 0.2428, "step": 17330 }, { "epoch": 2.72, "learning_rate": 9.12e-06, "loss": 0.3172, "step": 17340 }, { "epoch": 2.72, "learning_rate": 9.113725490196078e-06, "loss": 0.2311, "step": 17350 }, { "epoch": 2.72, "learning_rate": 9.107450980392158e-06, "loss": 0.2775, "step": 17360 }, { "epoch": 2.72, "learning_rate": 9.101176470588235e-06, "loss": 0.3055, "step": 17370 }, { "epoch": 2.73, "learning_rate": 9.094901960784315e-06, "loss": 0.3814, "step": 17380 }, { "epoch": 2.73, "learning_rate": 9.088627450980392e-06, "loss": 0.2801, "step": 17390 }, { "epoch": 2.73, "learning_rate": 9.082352941176472e-06, "loss": 0.3293, "step": 17400 }, { "epoch": 2.73, "learning_rate": 9.076078431372551e-06, "loss": 0.5122, "step": 17410 }, { "epoch": 2.73, "learning_rate": 9.069803921568629e-06, "loss": 0.4207, "step": 17420 }, { "epoch": 2.73, "learning_rate": 9.063529411764706e-06, "loss": 0.3576, "step": 17430 }, { "epoch": 2.74, "learning_rate": 9.057254901960786e-06, "loss": 0.4623, "step": 17440 }, { "epoch": 2.74, "learning_rate": 9.050980392156863e-06, "loss": 0.4395, "step": 17450 }, { "epoch": 2.74, "learning_rate": 9.044705882352942e-06, "loss": 0.2487, "step": 17460 }, { "epoch": 2.74, "learning_rate": 9.03843137254902e-06, "loss": 0.2597, "step": 17470 }, { "epoch": 2.74, "learning_rate": 9.032156862745098e-06, "loss": 0.4361, "step": 17480 }, { "epoch": 2.74, "learning_rate": 9.025882352941177e-06, "loss": 0.3286, "step": 17490 }, { "epoch": 2.75, "learning_rate": 9.019607843137256e-06, "loss": 0.4866, "step": 17500 }, { "epoch": 2.75, "learning_rate": 9.013333333333334e-06, "loss": 0.3998, "step": 17510 }, { "epoch": 2.75, "learning_rate": 9.007058823529412e-06, "loss": 0.4062, "step": 17520 }, { "epoch": 2.75, "learning_rate": 9.000784313725491e-06, "loss": 0.5076, "step": 17530 }, { "epoch": 2.75, "learning_rate": 8.99450980392157e-06, "loss": 0.3624, "step": 17540 }, { "epoch": 2.75, "learning_rate": 8.988235294117648e-06, "loss": 0.1621, "step": 17550 }, { "epoch": 2.75, "learning_rate": 8.981960784313726e-06, "loss": 0.2153, "step": 17560 }, { "epoch": 2.76, "learning_rate": 8.975686274509805e-06, "loss": 0.3741, "step": 17570 }, { "epoch": 2.76, "learning_rate": 8.969411764705882e-06, "loss": 0.3143, "step": 17580 }, { "epoch": 2.76, "learning_rate": 8.963137254901962e-06, "loss": 0.2188, "step": 17590 }, { "epoch": 2.76, "learning_rate": 8.95686274509804e-06, "loss": 0.3407, "step": 17600 }, { "epoch": 2.76, "learning_rate": 8.950588235294119e-06, "loss": 0.4454, "step": 17610 }, { "epoch": 2.76, "learning_rate": 8.944313725490196e-06, "loss": 0.3082, "step": 17620 }, { "epoch": 2.77, "learning_rate": 8.938039215686276e-06, "loss": 0.2336, "step": 17630 }, { "epoch": 2.77, "learning_rate": 8.931764705882355e-06, "loss": 0.2706, "step": 17640 }, { "epoch": 2.77, "learning_rate": 8.925490196078433e-06, "loss": 0.4002, "step": 17650 }, { "epoch": 2.77, "learning_rate": 8.91921568627451e-06, "loss": 0.1667, "step": 17660 }, { "epoch": 2.77, "learning_rate": 8.912941176470588e-06, "loss": 0.3441, "step": 17670 }, { "epoch": 2.77, "learning_rate": 8.906666666666667e-06, "loss": 0.2599, "step": 17680 }, { "epoch": 2.77, "learning_rate": 8.900392156862747e-06, "loss": 0.4733, "step": 17690 }, { "epoch": 2.78, "learning_rate": 8.894117647058824e-06, "loss": 0.2756, "step": 17700 }, { "epoch": 2.78, "learning_rate": 8.887843137254902e-06, "loss": 0.2822, "step": 17710 }, { "epoch": 2.78, "learning_rate": 8.881568627450981e-06, "loss": 0.1401, "step": 17720 }, { "epoch": 2.78, "learning_rate": 8.87529411764706e-06, "loss": 0.3638, "step": 17730 }, { "epoch": 2.78, "learning_rate": 8.869019607843138e-06, "loss": 0.6338, "step": 17740 }, { "epoch": 2.78, "learning_rate": 8.862745098039216e-06, "loss": 0.5741, "step": 17750 }, { "epoch": 2.79, "learning_rate": 8.856470588235295e-06, "loss": 0.3848, "step": 17760 }, { "epoch": 2.79, "learning_rate": 8.850196078431373e-06, "loss": 0.5591, "step": 17770 }, { "epoch": 2.79, "learning_rate": 8.843921568627452e-06, "loss": 0.3041, "step": 17780 }, { "epoch": 2.79, "learning_rate": 8.83764705882353e-06, "loss": 0.3772, "step": 17790 }, { "epoch": 2.79, "learning_rate": 8.831372549019609e-06, "loss": 0.2909, "step": 17800 }, { "epoch": 2.79, "learning_rate": 8.825098039215687e-06, "loss": 0.3164, "step": 17810 }, { "epoch": 2.8, "learning_rate": 8.818823529411766e-06, "loss": 0.4319, "step": 17820 }, { "epoch": 2.8, "learning_rate": 8.812549019607843e-06, "loss": 0.325, "step": 17830 }, { "epoch": 2.8, "learning_rate": 8.806274509803923e-06, "loss": 0.1971, "step": 17840 }, { "epoch": 2.8, "learning_rate": 8.8e-06, "loss": 0.162, "step": 17850 }, { "epoch": 2.8, "learning_rate": 8.793725490196078e-06, "loss": 0.2597, "step": 17860 }, { "epoch": 2.8, "learning_rate": 8.787450980392157e-06, "loss": 0.3993, "step": 17870 }, { "epoch": 2.8, "learning_rate": 8.781176470588237e-06, "loss": 0.3027, "step": 17880 }, { "epoch": 2.81, "learning_rate": 8.774901960784314e-06, "loss": 0.5104, "step": 17890 }, { "epoch": 2.81, "learning_rate": 8.768627450980392e-06, "loss": 0.4581, "step": 17900 }, { "epoch": 2.81, "learning_rate": 8.762352941176471e-06, "loss": 0.4674, "step": 17910 }, { "epoch": 2.81, "learning_rate": 8.75607843137255e-06, "loss": 0.3338, "step": 17920 }, { "epoch": 2.81, "learning_rate": 8.749803921568628e-06, "loss": 0.3931, "step": 17930 }, { "epoch": 2.81, "learning_rate": 8.743529411764706e-06, "loss": 0.308, "step": 17940 }, { "epoch": 2.82, "learning_rate": 8.737254901960785e-06, "loss": 0.2227, "step": 17950 }, { "epoch": 2.82, "learning_rate": 8.730980392156863e-06, "loss": 0.358, "step": 17960 }, { "epoch": 2.82, "learning_rate": 8.724705882352942e-06, "loss": 0.425, "step": 17970 }, { "epoch": 2.82, "learning_rate": 8.71843137254902e-06, "loss": 0.4379, "step": 17980 }, { "epoch": 2.82, "learning_rate": 8.712156862745099e-06, "loss": 0.4078, "step": 17990 }, { "epoch": 2.82, "learning_rate": 8.705882352941177e-06, "loss": 0.3025, "step": 18000 }, { "epoch": 2.83, "learning_rate": 8.699607843137256e-06, "loss": 0.3551, "step": 18010 }, { "epoch": 2.83, "learning_rate": 8.693333333333334e-06, "loss": 0.5342, "step": 18020 }, { "epoch": 2.83, "learning_rate": 8.687058823529413e-06, "loss": 0.4157, "step": 18030 }, { "epoch": 2.83, "learning_rate": 8.68078431372549e-06, "loss": 0.3125, "step": 18040 }, { "epoch": 2.83, "learning_rate": 8.67450980392157e-06, "loss": 0.5572, "step": 18050 }, { "epoch": 2.83, "learning_rate": 8.668235294117648e-06, "loss": 0.3792, "step": 18060 }, { "epoch": 2.83, "learning_rate": 8.661960784313727e-06, "loss": 0.3301, "step": 18070 }, { "epoch": 2.84, "learning_rate": 8.655686274509805e-06, "loss": 0.1222, "step": 18080 }, { "epoch": 2.84, "learning_rate": 8.649411764705882e-06, "loss": 0.1756, "step": 18090 }, { "epoch": 2.84, "learning_rate": 8.643137254901961e-06, "loss": 0.4489, "step": 18100 }, { "epoch": 2.84, "learning_rate": 8.63686274509804e-06, "loss": 0.2971, "step": 18110 }, { "epoch": 2.84, "learning_rate": 8.630588235294118e-06, "loss": 0.3564, "step": 18120 }, { "epoch": 2.84, "learning_rate": 8.624313725490196e-06, "loss": 0.4169, "step": 18130 }, { "epoch": 2.85, "learning_rate": 8.618039215686275e-06, "loss": 0.3553, "step": 18140 }, { "epoch": 2.85, "learning_rate": 8.611764705882355e-06, "loss": 0.502, "step": 18150 }, { "epoch": 2.85, "learning_rate": 8.605490196078432e-06, "loss": 0.3605, "step": 18160 }, { "epoch": 2.85, "learning_rate": 8.59921568627451e-06, "loss": 0.2922, "step": 18170 }, { "epoch": 2.85, "learning_rate": 8.59294117647059e-06, "loss": 0.3779, "step": 18180 }, { "epoch": 2.85, "learning_rate": 8.586666666666667e-06, "loss": 0.2982, "step": 18190 }, { "epoch": 2.85, "learning_rate": 8.580392156862746e-06, "loss": 0.2616, "step": 18200 }, { "epoch": 2.86, "learning_rate": 8.574117647058824e-06, "loss": 0.3698, "step": 18210 }, { "epoch": 2.86, "learning_rate": 8.567843137254903e-06, "loss": 0.3343, "step": 18220 }, { "epoch": 2.86, "learning_rate": 8.56156862745098e-06, "loss": 0.249, "step": 18230 }, { "epoch": 2.86, "learning_rate": 8.55529411764706e-06, "loss": 0.518, "step": 18240 }, { "epoch": 2.86, "learning_rate": 8.549019607843138e-06, "loss": 0.1583, "step": 18250 }, { "epoch": 2.86, "learning_rate": 8.542745098039217e-06, "loss": 0.5207, "step": 18260 }, { "epoch": 2.87, "learning_rate": 8.536470588235295e-06, "loss": 0.1492, "step": 18270 }, { "epoch": 2.87, "learning_rate": 8.530196078431372e-06, "loss": 0.2311, "step": 18280 }, { "epoch": 2.87, "learning_rate": 8.523921568627452e-06, "loss": 0.4858, "step": 18290 }, { "epoch": 2.87, "learning_rate": 8.517647058823531e-06, "loss": 0.2784, "step": 18300 }, { "epoch": 2.87, "learning_rate": 8.511372549019609e-06, "loss": 0.3158, "step": 18310 }, { "epoch": 2.87, "learning_rate": 8.505098039215686e-06, "loss": 0.2644, "step": 18320 }, { "epoch": 2.88, "learning_rate": 8.498823529411766e-06, "loss": 0.3994, "step": 18330 }, { "epoch": 2.88, "learning_rate": 8.492549019607845e-06, "loss": 0.2194, "step": 18340 }, { "epoch": 2.88, "learning_rate": 8.486274509803922e-06, "loss": 0.1751, "step": 18350 }, { "epoch": 2.88, "learning_rate": 8.48e-06, "loss": 0.1486, "step": 18360 }, { "epoch": 2.88, "learning_rate": 8.473725490196078e-06, "loss": 0.3047, "step": 18370 }, { "epoch": 2.88, "learning_rate": 8.467450980392157e-06, "loss": 0.317, "step": 18380 }, { "epoch": 2.88, "learning_rate": 8.461176470588236e-06, "loss": 0.3215, "step": 18390 }, { "epoch": 2.89, "learning_rate": 8.454901960784314e-06, "loss": 0.3033, "step": 18400 }, { "epoch": 2.89, "learning_rate": 8.448627450980393e-06, "loss": 0.4056, "step": 18410 }, { "epoch": 2.89, "learning_rate": 8.442352941176471e-06, "loss": 0.449, "step": 18420 }, { "epoch": 2.89, "learning_rate": 8.43607843137255e-06, "loss": 0.4118, "step": 18430 }, { "epoch": 2.89, "learning_rate": 8.429803921568628e-06, "loss": 0.2624, "step": 18440 }, { "epoch": 2.89, "learning_rate": 8.423529411764707e-06, "loss": 0.4446, "step": 18450 }, { "epoch": 2.9, "learning_rate": 8.417254901960785e-06, "loss": 0.4061, "step": 18460 }, { "epoch": 2.9, "learning_rate": 8.410980392156864e-06, "loss": 0.4048, "step": 18470 }, { "epoch": 2.9, "learning_rate": 8.404705882352942e-06, "loss": 0.2968, "step": 18480 }, { "epoch": 2.9, "learning_rate": 8.398431372549021e-06, "loss": 0.434, "step": 18490 }, { "epoch": 2.9, "learning_rate": 8.392156862745099e-06, "loss": 0.2889, "step": 18500 }, { "epoch": 2.9, "learning_rate": 8.385882352941176e-06, "loss": 0.3498, "step": 18510 }, { "epoch": 2.91, "learning_rate": 8.379607843137256e-06, "loss": 0.4105, "step": 18520 }, { "epoch": 2.91, "learning_rate": 8.373333333333335e-06, "loss": 0.337, "step": 18530 }, { "epoch": 2.91, "learning_rate": 8.367058823529413e-06, "loss": 0.3993, "step": 18540 }, { "epoch": 2.91, "learning_rate": 8.36078431372549e-06, "loss": 0.3321, "step": 18550 }, { "epoch": 2.91, "learning_rate": 8.35450980392157e-06, "loss": 0.2295, "step": 18560 }, { "epoch": 2.91, "learning_rate": 8.348235294117649e-06, "loss": 0.3856, "step": 18570 }, { "epoch": 2.91, "learning_rate": 8.341960784313727e-06, "loss": 0.0915, "step": 18580 }, { "epoch": 2.92, "learning_rate": 8.335686274509804e-06, "loss": 0.3896, "step": 18590 }, { "epoch": 2.92, "learning_rate": 8.329411764705882e-06, "loss": 0.4995, "step": 18600 }, { "epoch": 2.92, "learning_rate": 8.323137254901961e-06, "loss": 0.5544, "step": 18610 }, { "epoch": 2.92, "learning_rate": 8.31686274509804e-06, "loss": 0.2163, "step": 18620 }, { "epoch": 2.92, "learning_rate": 8.310588235294118e-06, "loss": 0.3222, "step": 18630 }, { "epoch": 2.92, "learning_rate": 8.304313725490196e-06, "loss": 0.579, "step": 18640 }, { "epoch": 2.93, "learning_rate": 8.298039215686275e-06, "loss": 0.388, "step": 18650 }, { "epoch": 2.93, "learning_rate": 8.291764705882354e-06, "loss": 0.3532, "step": 18660 }, { "epoch": 2.93, "learning_rate": 8.285490196078432e-06, "loss": 0.3911, "step": 18670 }, { "epoch": 2.93, "learning_rate": 8.279215686274511e-06, "loss": 0.5194, "step": 18680 }, { "epoch": 2.93, "learning_rate": 8.272941176470589e-06, "loss": 0.2557, "step": 18690 }, { "epoch": 2.93, "learning_rate": 8.266666666666667e-06, "loss": 0.1698, "step": 18700 }, { "epoch": 2.93, "learning_rate": 8.260392156862746e-06, "loss": 0.4303, "step": 18710 }, { "epoch": 2.94, "learning_rate": 8.254117647058825e-06, "loss": 0.4816, "step": 18720 }, { "epoch": 2.94, "learning_rate": 8.247843137254903e-06, "loss": 0.3586, "step": 18730 }, { "epoch": 2.94, "learning_rate": 8.24156862745098e-06, "loss": 0.4685, "step": 18740 }, { "epoch": 2.94, "learning_rate": 8.23529411764706e-06, "loss": 0.41, "step": 18750 }, { "epoch": 2.94, "learning_rate": 8.229019607843139e-06, "loss": 0.3134, "step": 18760 }, { "epoch": 2.94, "learning_rate": 8.222745098039217e-06, "loss": 0.438, "step": 18770 }, { "epoch": 2.95, "learning_rate": 8.216470588235294e-06, "loss": 0.554, "step": 18780 }, { "epoch": 2.95, "learning_rate": 8.210196078431372e-06, "loss": 0.3345, "step": 18790 }, { "epoch": 2.95, "learning_rate": 8.203921568627451e-06, "loss": 0.3442, "step": 18800 }, { "epoch": 2.95, "learning_rate": 8.19764705882353e-06, "loss": 0.312, "step": 18810 }, { "epoch": 2.95, "learning_rate": 8.191372549019608e-06, "loss": 0.4881, "step": 18820 }, { "epoch": 2.95, "learning_rate": 8.185098039215686e-06, "loss": 0.327, "step": 18830 }, { "epoch": 2.96, "learning_rate": 8.178823529411765e-06, "loss": 0.4489, "step": 18840 }, { "epoch": 2.96, "learning_rate": 8.172549019607845e-06, "loss": 0.3385, "step": 18850 }, { "epoch": 2.96, "learning_rate": 8.166274509803922e-06, "loss": 0.2938, "step": 18860 }, { "epoch": 2.96, "learning_rate": 8.16e-06, "loss": 0.295, "step": 18870 }, { "epoch": 2.96, "learning_rate": 8.153725490196079e-06, "loss": 0.3639, "step": 18880 }, { "epoch": 2.96, "learning_rate": 8.147450980392157e-06, "loss": 0.4333, "step": 18890 }, { "epoch": 2.96, "learning_rate": 8.141176470588236e-06, "loss": 0.4526, "step": 18900 }, { "epoch": 2.97, "learning_rate": 8.134901960784315e-06, "loss": 0.1471, "step": 18910 }, { "epoch": 2.97, "learning_rate": 8.128627450980393e-06, "loss": 0.2267, "step": 18920 }, { "epoch": 2.97, "learning_rate": 8.12235294117647e-06, "loss": 0.2743, "step": 18930 }, { "epoch": 2.97, "learning_rate": 8.11607843137255e-06, "loss": 0.459, "step": 18940 }, { "epoch": 2.97, "learning_rate": 8.10980392156863e-06, "loss": 0.3199, "step": 18950 }, { "epoch": 2.97, "learning_rate": 8.103529411764707e-06, "loss": 0.2266, "step": 18960 }, { "epoch": 2.98, "learning_rate": 8.097254901960785e-06, "loss": 0.4463, "step": 18970 }, { "epoch": 2.98, "learning_rate": 8.090980392156864e-06, "loss": 0.4505, "step": 18980 }, { "epoch": 2.98, "learning_rate": 8.084705882352941e-06, "loss": 0.4698, "step": 18990 }, { "epoch": 2.98, "learning_rate": 8.07843137254902e-06, "loss": 0.2585, "step": 19000 }, { "epoch": 2.98, "learning_rate": 8.072156862745098e-06, "loss": 0.3258, "step": 19010 }, { "epoch": 2.98, "learning_rate": 8.065882352941176e-06, "loss": 0.2964, "step": 19020 }, { "epoch": 2.99, "learning_rate": 8.059607843137255e-06, "loss": 0.6006, "step": 19030 }, { "epoch": 2.99, "learning_rate": 8.053333333333335e-06, "loss": 0.3206, "step": 19040 }, { "epoch": 2.99, "learning_rate": 8.047058823529412e-06, "loss": 0.3727, "step": 19050 }, { "epoch": 2.99, "learning_rate": 8.04078431372549e-06, "loss": 0.2278, "step": 19060 }, { "epoch": 2.99, "learning_rate": 8.03450980392157e-06, "loss": 0.4675, "step": 19070 }, { "epoch": 2.99, "learning_rate": 8.028235294117649e-06, "loss": 0.3053, "step": 19080 }, { "epoch": 2.99, "learning_rate": 8.021960784313726e-06, "loss": 0.261, "step": 19090 }, { "epoch": 3.0, "learning_rate": 8.015686274509804e-06, "loss": 0.465, "step": 19100 }, { "epoch": 3.0, "learning_rate": 8.009411764705883e-06, "loss": 0.3827, "step": 19110 }, { "epoch": 3.0, "learning_rate": 8.00313725490196e-06, "loss": 0.3258, "step": 19120 }, { "epoch": 3.0, "eval_accuracy": 0.9907777777777778, "eval_loss": 0.0363878458738327, "eval_runtime": 182.1771, "eval_samples_per_second": 49.402, "eval_steps_per_second": 6.175, "step": 19125 }, { "epoch": 3.0, "learning_rate": 7.99686274509804e-06, "loss": 0.4731, "step": 19130 }, { "epoch": 3.0, "learning_rate": 7.990588235294118e-06, "loss": 0.3229, "step": 19140 }, { "epoch": 3.0, "learning_rate": 7.984313725490197e-06, "loss": 0.4904, "step": 19150 }, { "epoch": 3.01, "learning_rate": 7.978039215686275e-06, "loss": 0.4503, "step": 19160 }, { "epoch": 3.01, "learning_rate": 7.971764705882354e-06, "loss": 0.4685, "step": 19170 }, { "epoch": 3.01, "learning_rate": 7.965490196078433e-06, "loss": 0.3208, "step": 19180 }, { "epoch": 3.01, "learning_rate": 7.959215686274511e-06, "loss": 0.5424, "step": 19190 }, { "epoch": 3.01, "learning_rate": 7.952941176470589e-06, "loss": 0.4642, "step": 19200 }, { "epoch": 3.01, "learning_rate": 7.946666666666666e-06, "loss": 0.3867, "step": 19210 }, { "epoch": 3.01, "learning_rate": 7.940392156862746e-06, "loss": 0.2952, "step": 19220 }, { "epoch": 3.02, "learning_rate": 7.934117647058825e-06, "loss": 0.2858, "step": 19230 }, { "epoch": 3.02, "learning_rate": 7.927843137254903e-06, "loss": 0.4026, "step": 19240 }, { "epoch": 3.02, "learning_rate": 7.92156862745098e-06, "loss": 0.3681, "step": 19250 }, { "epoch": 3.02, "learning_rate": 7.91529411764706e-06, "loss": 0.3217, "step": 19260 }, { "epoch": 3.02, "learning_rate": 7.909019607843139e-06, "loss": 0.2365, "step": 19270 }, { "epoch": 3.02, "learning_rate": 7.902745098039216e-06, "loss": 0.2175, "step": 19280 }, { "epoch": 3.03, "learning_rate": 7.896470588235294e-06, "loss": 0.4554, "step": 19290 }, { "epoch": 3.03, "learning_rate": 7.890196078431373e-06, "loss": 0.2506, "step": 19300 }, { "epoch": 3.03, "learning_rate": 7.883921568627451e-06, "loss": 0.3128, "step": 19310 }, { "epoch": 3.03, "learning_rate": 7.87764705882353e-06, "loss": 0.5141, "step": 19320 }, { "epoch": 3.03, "learning_rate": 7.871372549019608e-06, "loss": 0.3948, "step": 19330 }, { "epoch": 3.03, "learning_rate": 7.865098039215687e-06, "loss": 0.2592, "step": 19340 }, { "epoch": 3.04, "learning_rate": 7.858823529411765e-06, "loss": 0.425, "step": 19350 }, { "epoch": 3.04, "learning_rate": 7.852549019607844e-06, "loss": 0.3417, "step": 19360 }, { "epoch": 3.04, "learning_rate": 7.846274509803922e-06, "loss": 0.3257, "step": 19370 }, { "epoch": 3.04, "learning_rate": 7.840000000000001e-06, "loss": 0.233, "step": 19380 }, { "epoch": 3.04, "learning_rate": 7.833725490196079e-06, "loss": 0.3651, "step": 19390 }, { "epoch": 3.04, "learning_rate": 7.827450980392156e-06, "loss": 0.3103, "step": 19400 }, { "epoch": 3.04, "learning_rate": 7.821176470588236e-06, "loss": 0.1705, "step": 19410 }, { "epoch": 3.05, "learning_rate": 7.814901960784315e-06, "loss": 0.1266, "step": 19420 }, { "epoch": 3.05, "learning_rate": 7.808627450980393e-06, "loss": 0.4769, "step": 19430 }, { "epoch": 3.05, "learning_rate": 7.80235294117647e-06, "loss": 0.239, "step": 19440 }, { "epoch": 3.05, "learning_rate": 7.79607843137255e-06, "loss": 0.4293, "step": 19450 }, { "epoch": 3.05, "learning_rate": 7.789803921568629e-06, "loss": 0.3188, "step": 19460 }, { "epoch": 3.05, "learning_rate": 7.783529411764707e-06, "loss": 0.2302, "step": 19470 }, { "epoch": 3.06, "learning_rate": 7.777254901960784e-06, "loss": 0.2135, "step": 19480 }, { "epoch": 3.06, "learning_rate": 7.770980392156864e-06, "loss": 0.3155, "step": 19490 }, { "epoch": 3.06, "learning_rate": 7.764705882352941e-06, "loss": 0.1963, "step": 19500 }, { "epoch": 3.06, "learning_rate": 7.75843137254902e-06, "loss": 0.3152, "step": 19510 }, { "epoch": 3.06, "learning_rate": 7.752156862745098e-06, "loss": 0.1745, "step": 19520 }, { "epoch": 3.06, "learning_rate": 7.745882352941177e-06, "loss": 0.1329, "step": 19530 }, { "epoch": 3.07, "learning_rate": 7.739607843137255e-06, "loss": 0.3411, "step": 19540 }, { "epoch": 3.07, "learning_rate": 7.733333333333334e-06, "loss": 0.2378, "step": 19550 }, { "epoch": 3.07, "learning_rate": 7.727058823529412e-06, "loss": 0.139, "step": 19560 }, { "epoch": 3.07, "learning_rate": 7.720784313725491e-06, "loss": 0.2839, "step": 19570 }, { "epoch": 3.07, "learning_rate": 7.714509803921569e-06, "loss": 0.4453, "step": 19580 }, { "epoch": 3.07, "learning_rate": 7.708235294117648e-06, "loss": 0.3617, "step": 19590 }, { "epoch": 3.07, "learning_rate": 7.701960784313726e-06, "loss": 0.3364, "step": 19600 }, { "epoch": 3.08, "learning_rate": 7.695686274509805e-06, "loss": 0.395, "step": 19610 }, { "epoch": 3.08, "learning_rate": 7.689411764705883e-06, "loss": 0.3952, "step": 19620 }, { "epoch": 3.08, "learning_rate": 7.68313725490196e-06, "loss": 0.2573, "step": 19630 }, { "epoch": 3.08, "learning_rate": 7.67686274509804e-06, "loss": 0.475, "step": 19640 }, { "epoch": 3.08, "learning_rate": 7.670588235294119e-06, "loss": 0.3967, "step": 19650 }, { "epoch": 3.08, "learning_rate": 7.664313725490197e-06, "loss": 0.667, "step": 19660 }, { "epoch": 3.09, "learning_rate": 7.658039215686274e-06, "loss": 0.3638, "step": 19670 }, { "epoch": 3.09, "learning_rate": 7.651764705882354e-06, "loss": 0.6827, "step": 19680 }, { "epoch": 3.09, "learning_rate": 7.645490196078433e-06, "loss": 0.232, "step": 19690 }, { "epoch": 3.09, "learning_rate": 7.63921568627451e-06, "loss": 0.2673, "step": 19700 }, { "epoch": 3.09, "learning_rate": 7.632941176470588e-06, "loss": 0.2532, "step": 19710 }, { "epoch": 3.09, "learning_rate": 7.626666666666668e-06, "loss": 0.3717, "step": 19720 }, { "epoch": 3.09, "learning_rate": 7.620392156862746e-06, "loss": 0.3205, "step": 19730 }, { "epoch": 3.1, "learning_rate": 7.614117647058824e-06, "loss": 0.4737, "step": 19740 }, { "epoch": 3.1, "learning_rate": 7.607843137254902e-06, "loss": 0.2299, "step": 19750 }, { "epoch": 3.1, "learning_rate": 7.6015686274509815e-06, "loss": 0.3496, "step": 19760 }, { "epoch": 3.1, "learning_rate": 7.59529411764706e-06, "loss": 0.3882, "step": 19770 }, { "epoch": 3.1, "learning_rate": 7.589019607843138e-06, "loss": 0.1965, "step": 19780 }, { "epoch": 3.1, "learning_rate": 7.582745098039216e-06, "loss": 0.3038, "step": 19790 }, { "epoch": 3.11, "learning_rate": 7.576470588235295e-06, "loss": 0.3177, "step": 19800 }, { "epoch": 3.11, "learning_rate": 7.570196078431373e-06, "loss": 0.1441, "step": 19810 }, { "epoch": 3.11, "learning_rate": 7.5639215686274515e-06, "loss": 0.2431, "step": 19820 }, { "epoch": 3.11, "learning_rate": 7.557647058823529e-06, "loss": 0.3756, "step": 19830 }, { "epoch": 3.11, "learning_rate": 7.5513725490196085e-06, "loss": 0.2081, "step": 19840 }, { "epoch": 3.11, "learning_rate": 7.545098039215687e-06, "loss": 0.4178, "step": 19850 }, { "epoch": 3.12, "learning_rate": 7.538823529411765e-06, "loss": 0.3327, "step": 19860 }, { "epoch": 3.12, "learning_rate": 7.532549019607843e-06, "loss": 0.3018, "step": 19870 }, { "epoch": 3.12, "learning_rate": 7.526274509803922e-06, "loss": 0.2699, "step": 19880 }, { "epoch": 3.12, "learning_rate": 7.520000000000001e-06, "loss": 0.2226, "step": 19890 }, { "epoch": 3.12, "learning_rate": 7.5137254901960785e-06, "loss": 0.3493, "step": 19900 }, { "epoch": 3.12, "learning_rate": 7.507450980392158e-06, "loss": 0.2332, "step": 19910 }, { "epoch": 3.12, "learning_rate": 7.501176470588236e-06, "loss": 0.2828, "step": 19920 }, { "epoch": 3.13, "learning_rate": 7.494901960784314e-06, "loss": 0.2694, "step": 19930 }, { "epoch": 3.13, "learning_rate": 7.488627450980392e-06, "loss": 0.3806, "step": 19940 }, { "epoch": 3.13, "learning_rate": 7.482352941176472e-06, "loss": 0.3464, "step": 19950 }, { "epoch": 3.13, "learning_rate": 7.47607843137255e-06, "loss": 0.2655, "step": 19960 }, { "epoch": 3.13, "learning_rate": 7.469803921568628e-06, "loss": 0.1824, "step": 19970 }, { "epoch": 3.13, "learning_rate": 7.463529411764706e-06, "loss": 0.1975, "step": 19980 }, { "epoch": 3.14, "learning_rate": 7.457254901960786e-06, "loss": 0.0615, "step": 19990 }, { "epoch": 3.14, "learning_rate": 7.450980392156863e-06, "loss": 0.2684, "step": 20000 }, { "epoch": 3.14, "learning_rate": 7.444705882352942e-06, "loss": 0.2282, "step": 20010 }, { "epoch": 3.14, "learning_rate": 7.43843137254902e-06, "loss": 0.4422, "step": 20020 }, { "epoch": 3.14, "learning_rate": 7.432156862745099e-06, "loss": 0.2608, "step": 20030 }, { "epoch": 3.14, "learning_rate": 7.425882352941177e-06, "loss": 0.4886, "step": 20040 }, { "epoch": 3.15, "learning_rate": 7.419607843137256e-06, "loss": 0.3025, "step": 20050 }, { "epoch": 3.15, "learning_rate": 7.413333333333333e-06, "loss": 0.2405, "step": 20060 }, { "epoch": 3.15, "learning_rate": 7.4070588235294125e-06, "loss": 0.3864, "step": 20070 }, { "epoch": 3.15, "learning_rate": 7.400784313725491e-06, "loss": 0.2389, "step": 20080 }, { "epoch": 3.15, "learning_rate": 7.394509803921569e-06, "loss": 0.24, "step": 20090 }, { "epoch": 3.15, "learning_rate": 7.388235294117647e-06, "loss": 0.2626, "step": 20100 }, { "epoch": 3.15, "learning_rate": 7.3819607843137264e-06, "loss": 0.2702, "step": 20110 }, { "epoch": 3.16, "learning_rate": 7.375686274509805e-06, "loss": 0.3968, "step": 20120 }, { "epoch": 3.16, "learning_rate": 7.3694117647058825e-06, "loss": 0.4308, "step": 20130 }, { "epoch": 3.16, "learning_rate": 7.363137254901961e-06, "loss": 0.2898, "step": 20140 }, { "epoch": 3.16, "learning_rate": 7.35686274509804e-06, "loss": 0.3434, "step": 20150 }, { "epoch": 3.16, "learning_rate": 7.350588235294118e-06, "loss": 0.373, "step": 20160 }, { "epoch": 3.16, "learning_rate": 7.3443137254901964e-06, "loss": 0.3295, "step": 20170 }, { "epoch": 3.17, "learning_rate": 7.338039215686276e-06, "loss": 0.3631, "step": 20180 }, { "epoch": 3.17, "learning_rate": 7.331764705882353e-06, "loss": 0.2938, "step": 20190 }, { "epoch": 3.17, "learning_rate": 7.325490196078432e-06, "loss": 0.4107, "step": 20200 }, { "epoch": 3.17, "learning_rate": 7.31921568627451e-06, "loss": 0.1986, "step": 20210 }, { "epoch": 3.17, "learning_rate": 7.31294117647059e-06, "loss": 0.2527, "step": 20220 }, { "epoch": 3.17, "learning_rate": 7.306666666666667e-06, "loss": 0.1378, "step": 20230 }, { "epoch": 3.17, "learning_rate": 7.300392156862746e-06, "loss": 0.5631, "step": 20240 }, { "epoch": 3.18, "learning_rate": 7.294117647058823e-06, "loss": 0.3734, "step": 20250 }, { "epoch": 3.18, "learning_rate": 7.287843137254903e-06, "loss": 0.2748, "step": 20260 }, { "epoch": 3.18, "learning_rate": 7.281568627450981e-06, "loss": 0.478, "step": 20270 }, { "epoch": 3.18, "learning_rate": 7.27529411764706e-06, "loss": 0.3338, "step": 20280 }, { "epoch": 3.18, "learning_rate": 7.269019607843137e-06, "loss": 0.434, "step": 20290 }, { "epoch": 3.18, "learning_rate": 7.262745098039217e-06, "loss": 0.4496, "step": 20300 }, { "epoch": 3.19, "learning_rate": 7.256470588235295e-06, "loss": 0.2153, "step": 20310 }, { "epoch": 3.19, "learning_rate": 7.250196078431373e-06, "loss": 0.3476, "step": 20320 }, { "epoch": 3.19, "learning_rate": 7.243921568627451e-06, "loss": 0.3039, "step": 20330 }, { "epoch": 3.19, "learning_rate": 7.2376470588235305e-06, "loss": 0.2146, "step": 20340 }, { "epoch": 3.19, "learning_rate": 7.231372549019608e-06, "loss": 0.4219, "step": 20350 }, { "epoch": 3.19, "learning_rate": 7.225098039215687e-06, "loss": 0.5333, "step": 20360 }, { "epoch": 3.2, "learning_rate": 7.218823529411765e-06, "loss": 0.393, "step": 20370 }, { "epoch": 3.2, "learning_rate": 7.212549019607844e-06, "loss": 0.28, "step": 20380 }, { "epoch": 3.2, "learning_rate": 7.206274509803922e-06, "loss": 0.3916, "step": 20390 }, { "epoch": 3.2, "learning_rate": 7.2000000000000005e-06, "loss": 0.3741, "step": 20400 }, { "epoch": 3.2, "learning_rate": 7.193725490196078e-06, "loss": 0.3447, "step": 20410 }, { "epoch": 3.2, "learning_rate": 7.1874509803921575e-06, "loss": 0.4436, "step": 20420 }, { "epoch": 3.2, "learning_rate": 7.181176470588236e-06, "loss": 0.2574, "step": 20430 }, { "epoch": 3.21, "learning_rate": 7.174901960784314e-06, "loss": 0.196, "step": 20440 }, { "epoch": 3.21, "learning_rate": 7.168627450980393e-06, "loss": 0.3765, "step": 20450 }, { "epoch": 3.21, "learning_rate": 7.162352941176471e-06, "loss": 0.4094, "step": 20460 }, { "epoch": 3.21, "learning_rate": 7.15607843137255e-06, "loss": 0.3931, "step": 20470 }, { "epoch": 3.21, "learning_rate": 7.1498039215686275e-06, "loss": 0.2569, "step": 20480 }, { "epoch": 3.21, "learning_rate": 7.143529411764707e-06, "loss": 0.2151, "step": 20490 }, { "epoch": 3.22, "learning_rate": 7.137254901960785e-06, "loss": 0.1369, "step": 20500 }, { "epoch": 3.22, "learning_rate": 7.130980392156863e-06, "loss": 0.3423, "step": 20510 }, { "epoch": 3.22, "learning_rate": 7.124705882352941e-06, "loss": 0.2859, "step": 20520 }, { "epoch": 3.22, "learning_rate": 7.118431372549021e-06, "loss": 0.4512, "step": 20530 }, { "epoch": 3.22, "learning_rate": 7.112156862745099e-06, "loss": 0.3246, "step": 20540 }, { "epoch": 3.22, "learning_rate": 7.105882352941177e-06, "loss": 0.2924, "step": 20550 }, { "epoch": 3.23, "learning_rate": 7.099607843137255e-06, "loss": 0.3716, "step": 20560 }, { "epoch": 3.23, "learning_rate": 7.093333333333335e-06, "loss": 0.3191, "step": 20570 }, { "epoch": 3.23, "learning_rate": 7.087058823529412e-06, "loss": 0.2671, "step": 20580 }, { "epoch": 3.23, "learning_rate": 7.080784313725491e-06, "loss": 0.4535, "step": 20590 }, { "epoch": 3.23, "learning_rate": 7.074509803921568e-06, "loss": 0.3889, "step": 20600 }, { "epoch": 3.23, "learning_rate": 7.068235294117648e-06, "loss": 0.3776, "step": 20610 }, { "epoch": 3.23, "learning_rate": 7.061960784313726e-06, "loss": 0.2302, "step": 20620 }, { "epoch": 3.24, "learning_rate": 7.055686274509805e-06, "loss": 0.463, "step": 20630 }, { "epoch": 3.24, "learning_rate": 7.049411764705882e-06, "loss": 0.2618, "step": 20640 }, { "epoch": 3.24, "learning_rate": 7.0431372549019615e-06, "loss": 0.2658, "step": 20650 }, { "epoch": 3.24, "learning_rate": 7.03686274509804e-06, "loss": 0.2516, "step": 20660 }, { "epoch": 3.24, "learning_rate": 7.030588235294118e-06, "loss": 0.3984, "step": 20670 }, { "epoch": 3.24, "learning_rate": 7.024313725490197e-06, "loss": 0.1766, "step": 20680 }, { "epoch": 3.25, "learning_rate": 7.0180392156862754e-06, "loss": 0.4152, "step": 20690 }, { "epoch": 3.25, "learning_rate": 7.011764705882353e-06, "loss": 0.373, "step": 20700 }, { "epoch": 3.25, "learning_rate": 7.0054901960784315e-06, "loss": 0.321, "step": 20710 }, { "epoch": 3.25, "learning_rate": 6.999215686274511e-06, "loss": 0.3035, "step": 20720 }, { "epoch": 3.25, "learning_rate": 6.992941176470589e-06, "loss": 0.2156, "step": 20730 }, { "epoch": 3.25, "learning_rate": 6.986666666666667e-06, "loss": 0.2999, "step": 20740 }, { "epoch": 3.25, "learning_rate": 6.9803921568627454e-06, "loss": 0.5007, "step": 20750 }, { "epoch": 3.26, "learning_rate": 6.974117647058825e-06, "loss": 0.6732, "step": 20760 }, { "epoch": 3.26, "learning_rate": 6.967843137254902e-06, "loss": 0.2254, "step": 20770 }, { "epoch": 3.26, "learning_rate": 6.961568627450981e-06, "loss": 0.3905, "step": 20780 }, { "epoch": 3.26, "learning_rate": 6.955294117647059e-06, "loss": 0.1887, "step": 20790 }, { "epoch": 3.26, "learning_rate": 6.949019607843138e-06, "loss": 0.2605, "step": 20800 }, { "epoch": 3.26, "learning_rate": 6.942745098039216e-06, "loss": 0.3464, "step": 20810 }, { "epoch": 3.27, "learning_rate": 6.936470588235295e-06, "loss": 0.3419, "step": 20820 }, { "epoch": 3.27, "learning_rate": 6.930196078431372e-06, "loss": 0.3031, "step": 20830 }, { "epoch": 3.27, "learning_rate": 6.923921568627452e-06, "loss": 0.3469, "step": 20840 }, { "epoch": 3.27, "learning_rate": 6.91764705882353e-06, "loss": 0.1672, "step": 20850 }, { "epoch": 3.27, "learning_rate": 6.911372549019608e-06, "loss": 0.463, "step": 20860 }, { "epoch": 3.27, "learning_rate": 6.905098039215686e-06, "loss": 0.4802, "step": 20870 }, { "epoch": 3.28, "learning_rate": 6.898823529411766e-06, "loss": 0.3593, "step": 20880 }, { "epoch": 3.28, "learning_rate": 6.892549019607844e-06, "loss": 0.3218, "step": 20890 }, { "epoch": 3.28, "learning_rate": 6.886274509803922e-06, "loss": 0.4782, "step": 20900 }, { "epoch": 3.28, "learning_rate": 6.88e-06, "loss": 0.5313, "step": 20910 }, { "epoch": 3.28, "learning_rate": 6.8737254901960795e-06, "loss": 0.1557, "step": 20920 }, { "epoch": 3.28, "learning_rate": 6.867450980392157e-06, "loss": 0.4308, "step": 20930 }, { "epoch": 3.28, "learning_rate": 6.861176470588236e-06, "loss": 0.1753, "step": 20940 }, { "epoch": 3.29, "learning_rate": 6.854901960784315e-06, "loss": 0.339, "step": 20950 }, { "epoch": 3.29, "learning_rate": 6.8486274509803926e-06, "loss": 0.3137, "step": 20960 }, { "epoch": 3.29, "learning_rate": 6.842352941176471e-06, "loss": 0.2151, "step": 20970 }, { "epoch": 3.29, "learning_rate": 6.8360784313725495e-06, "loss": 0.235, "step": 20980 }, { "epoch": 3.29, "learning_rate": 6.829803921568629e-06, "loss": 0.3923, "step": 20990 }, { "epoch": 3.29, "learning_rate": 6.8235294117647065e-06, "loss": 0.4013, "step": 21000 }, { "epoch": 3.3, "learning_rate": 6.817254901960785e-06, "loss": 0.4606, "step": 21010 }, { "epoch": 3.3, "learning_rate": 6.8109803921568626e-06, "loss": 0.4971, "step": 21020 }, { "epoch": 3.3, "learning_rate": 6.804705882352942e-06, "loss": 0.3534, "step": 21030 }, { "epoch": 3.3, "learning_rate": 6.79843137254902e-06, "loss": 0.3288, "step": 21040 }, { "epoch": 3.3, "learning_rate": 6.792156862745099e-06, "loss": 0.3194, "step": 21050 }, { "epoch": 3.3, "learning_rate": 6.7858823529411765e-06, "loss": 0.1455, "step": 21060 }, { "epoch": 3.31, "learning_rate": 6.779607843137256e-06, "loss": 0.3388, "step": 21070 }, { "epoch": 3.31, "learning_rate": 6.773333333333334e-06, "loss": 0.2568, "step": 21080 }, { "epoch": 3.31, "learning_rate": 6.767058823529412e-06, "loss": 0.6231, "step": 21090 }, { "epoch": 3.31, "learning_rate": 6.76078431372549e-06, "loss": 0.1864, "step": 21100 }, { "epoch": 3.31, "learning_rate": 6.75450980392157e-06, "loss": 0.3333, "step": 21110 }, { "epoch": 3.31, "learning_rate": 6.748235294117647e-06, "loss": 0.1873, "step": 21120 }, { "epoch": 3.31, "learning_rate": 6.741960784313726e-06, "loss": 0.3136, "step": 21130 }, { "epoch": 3.32, "learning_rate": 6.735686274509804e-06, "loss": 0.4572, "step": 21140 }, { "epoch": 3.32, "learning_rate": 6.729411764705884e-06, "loss": 0.3208, "step": 21150 }, { "epoch": 3.32, "learning_rate": 6.723137254901961e-06, "loss": 0.3164, "step": 21160 }, { "epoch": 3.32, "learning_rate": 6.71686274509804e-06, "loss": 0.108, "step": 21170 }, { "epoch": 3.32, "learning_rate": 6.710588235294119e-06, "loss": 0.2521, "step": 21180 }, { "epoch": 3.32, "learning_rate": 6.704313725490197e-06, "loss": 0.2937, "step": 21190 }, { "epoch": 3.33, "learning_rate": 6.698039215686275e-06, "loss": 0.2947, "step": 21200 }, { "epoch": 3.33, "learning_rate": 6.691764705882353e-06, "loss": 0.2487, "step": 21210 }, { "epoch": 3.33, "learning_rate": 6.685490196078432e-06, "loss": 0.4345, "step": 21220 }, { "epoch": 3.33, "learning_rate": 6.6792156862745105e-06, "loss": 0.4515, "step": 21230 }, { "epoch": 3.33, "learning_rate": 6.672941176470589e-06, "loss": 0.2615, "step": 21240 }, { "epoch": 3.33, "learning_rate": 6.666666666666667e-06, "loss": 0.257, "step": 21250 }, { "epoch": 3.33, "learning_rate": 6.660392156862746e-06, "loss": 0.2867, "step": 21260 }, { "epoch": 3.34, "learning_rate": 6.6541176470588244e-06, "loss": 0.4876, "step": 21270 }, { "epoch": 3.34, "learning_rate": 6.647843137254902e-06, "loss": 0.2633, "step": 21280 }, { "epoch": 3.34, "learning_rate": 6.6415686274509805e-06, "loss": 0.369, "step": 21290 }, { "epoch": 3.34, "learning_rate": 6.63529411764706e-06, "loss": 0.2826, "step": 21300 }, { "epoch": 3.34, "learning_rate": 6.6290196078431375e-06, "loss": 0.3165, "step": 21310 }, { "epoch": 3.34, "learning_rate": 6.622745098039216e-06, "loss": 0.4126, "step": 21320 }, { "epoch": 3.35, "learning_rate": 6.6164705882352944e-06, "loss": 0.3293, "step": 21330 }, { "epoch": 3.35, "learning_rate": 6.610196078431374e-06, "loss": 0.248, "step": 21340 }, { "epoch": 3.35, "learning_rate": 6.603921568627451e-06, "loss": 0.3007, "step": 21350 }, { "epoch": 3.35, "learning_rate": 6.59764705882353e-06, "loss": 0.401, "step": 21360 }, { "epoch": 3.35, "learning_rate": 6.5913725490196075e-06, "loss": 0.4457, "step": 21370 }, { "epoch": 3.35, "learning_rate": 6.585098039215687e-06, "loss": 0.1998, "step": 21380 }, { "epoch": 3.36, "learning_rate": 6.578823529411765e-06, "loss": 0.1857, "step": 21390 }, { "epoch": 3.36, "learning_rate": 6.572549019607844e-06, "loss": 0.2499, "step": 21400 }, { "epoch": 3.36, "learning_rate": 6.566274509803921e-06, "loss": 0.3, "step": 21410 }, { "epoch": 3.36, "learning_rate": 6.560000000000001e-06, "loss": 0.3338, "step": 21420 }, { "epoch": 3.36, "learning_rate": 6.553725490196079e-06, "loss": 0.2447, "step": 21430 }, { "epoch": 3.36, "learning_rate": 6.547450980392157e-06, "loss": 0.4532, "step": 21440 }, { "epoch": 3.36, "learning_rate": 6.541176470588236e-06, "loss": 0.439, "step": 21450 }, { "epoch": 3.37, "learning_rate": 6.534901960784315e-06, "loss": 0.4341, "step": 21460 }, { "epoch": 3.37, "learning_rate": 6.528627450980392e-06, "loss": 0.3028, "step": 21470 }, { "epoch": 3.37, "learning_rate": 6.522352941176471e-06, "loss": 0.3106, "step": 21480 }, { "epoch": 3.37, "learning_rate": 6.51607843137255e-06, "loss": 0.2245, "step": 21490 }, { "epoch": 3.37, "learning_rate": 6.5098039215686285e-06, "loss": 0.3647, "step": 21500 }, { "epoch": 3.37, "learning_rate": 6.503529411764706e-06, "loss": 0.2352, "step": 21510 }, { "epoch": 3.38, "learning_rate": 6.497254901960785e-06, "loss": 0.1372, "step": 21520 }, { "epoch": 3.38, "learning_rate": 6.490980392156864e-06, "loss": 0.1873, "step": 21530 }, { "epoch": 3.38, "learning_rate": 6.4847058823529416e-06, "loss": 0.4883, "step": 21540 }, { "epoch": 3.38, "learning_rate": 6.47843137254902e-06, "loss": 0.271, "step": 21550 }, { "epoch": 3.38, "learning_rate": 6.4721568627450985e-06, "loss": 0.1791, "step": 21560 }, { "epoch": 3.38, "learning_rate": 6.465882352941177e-06, "loss": 0.288, "step": 21570 }, { "epoch": 3.39, "learning_rate": 6.4596078431372555e-06, "loss": 0.2436, "step": 21580 }, { "epoch": 3.39, "learning_rate": 6.453333333333334e-06, "loss": 0.2503, "step": 21590 }, { "epoch": 3.39, "learning_rate": 6.4470588235294116e-06, "loss": 0.1201, "step": 21600 }, { "epoch": 3.39, "learning_rate": 6.440784313725491e-06, "loss": 0.3332, "step": 21610 }, { "epoch": 3.39, "learning_rate": 6.434509803921569e-06, "loss": 0.2776, "step": 21620 }, { "epoch": 3.39, "learning_rate": 6.428235294117647e-06, "loss": 0.4609, "step": 21630 }, { "epoch": 3.39, "learning_rate": 6.4219607843137255e-06, "loss": 0.2824, "step": 21640 }, { "epoch": 3.4, "learning_rate": 6.415686274509805e-06, "loss": 0.2988, "step": 21650 }, { "epoch": 3.4, "learning_rate": 6.409411764705883e-06, "loss": 0.1067, "step": 21660 }, { "epoch": 3.4, "learning_rate": 6.403137254901961e-06, "loss": 0.2285, "step": 21670 }, { "epoch": 3.4, "learning_rate": 6.39686274509804e-06, "loss": 0.2711, "step": 21680 }, { "epoch": 3.4, "learning_rate": 6.390588235294119e-06, "loss": 0.4535, "step": 21690 }, { "epoch": 3.4, "learning_rate": 6.384313725490196e-06, "loss": 0.3385, "step": 21700 }, { "epoch": 3.41, "learning_rate": 6.378039215686275e-06, "loss": 0.3365, "step": 21710 }, { "epoch": 3.41, "learning_rate": 6.371764705882354e-06, "loss": 0.2344, "step": 21720 }, { "epoch": 3.41, "learning_rate": 6.365490196078432e-06, "loss": 0.507, "step": 21730 }, { "epoch": 3.41, "learning_rate": 6.35921568627451e-06, "loss": 0.4196, "step": 21740 }, { "epoch": 3.41, "learning_rate": 6.352941176470589e-06, "loss": 0.2905, "step": 21750 }, { "epoch": 3.41, "learning_rate": 6.346666666666668e-06, "loss": 0.4668, "step": 21760 }, { "epoch": 3.41, "learning_rate": 6.340392156862746e-06, "loss": 0.4182, "step": 21770 }, { "epoch": 3.42, "learning_rate": 6.334117647058824e-06, "loss": 0.2318, "step": 21780 }, { "epoch": 3.42, "learning_rate": 6.327843137254902e-06, "loss": 0.3779, "step": 21790 }, { "epoch": 3.42, "learning_rate": 6.321568627450981e-06, "loss": 0.4481, "step": 21800 }, { "epoch": 3.42, "learning_rate": 6.3152941176470595e-06, "loss": 0.2573, "step": 21810 }, { "epoch": 3.42, "learning_rate": 6.309019607843138e-06, "loss": 0.4027, "step": 21820 }, { "epoch": 3.42, "learning_rate": 6.302745098039216e-06, "loss": 0.2268, "step": 21830 }, { "epoch": 3.43, "learning_rate": 6.296470588235295e-06, "loss": 0.2652, "step": 21840 }, { "epoch": 3.43, "learning_rate": 6.2901960784313734e-06, "loss": 0.3918, "step": 21850 }, { "epoch": 3.43, "learning_rate": 6.283921568627451e-06, "loss": 0.2705, "step": 21860 }, { "epoch": 3.43, "learning_rate": 6.2776470588235295e-06, "loss": 0.2354, "step": 21870 }, { "epoch": 3.43, "learning_rate": 6.271372549019609e-06, "loss": 0.3956, "step": 21880 }, { "epoch": 3.43, "learning_rate": 6.2650980392156865e-06, "loss": 0.3234, "step": 21890 }, { "epoch": 3.44, "learning_rate": 6.258823529411765e-06, "loss": 0.3407, "step": 21900 }, { "epoch": 3.44, "learning_rate": 6.2525490196078434e-06, "loss": 0.1694, "step": 21910 }, { "epoch": 3.44, "learning_rate": 6.246274509803923e-06, "loss": 0.5101, "step": 21920 }, { "epoch": 3.44, "learning_rate": 6.24e-06, "loss": 0.4408, "step": 21930 }, { "epoch": 3.44, "learning_rate": 6.233725490196079e-06, "loss": 0.3005, "step": 21940 }, { "epoch": 3.44, "learning_rate": 6.227450980392158e-06, "loss": 0.2864, "step": 21950 }, { "epoch": 3.44, "learning_rate": 6.221176470588236e-06, "loss": 0.4679, "step": 21960 }, { "epoch": 3.45, "learning_rate": 6.214901960784314e-06, "loss": 0.2894, "step": 21970 }, { "epoch": 3.45, "learning_rate": 6.208627450980392e-06, "loss": 0.3984, "step": 21980 }, { "epoch": 3.45, "learning_rate": 6.202352941176471e-06, "loss": 0.4162, "step": 21990 }, { "epoch": 3.45, "learning_rate": 6.19607843137255e-06, "loss": 0.4523, "step": 22000 }, { "epoch": 3.45, "learning_rate": 6.189803921568628e-06, "loss": 0.3394, "step": 22010 }, { "epoch": 3.45, "learning_rate": 6.183529411764706e-06, "loss": 0.287, "step": 22020 }, { "epoch": 3.46, "learning_rate": 6.177254901960785e-06, "loss": 0.3015, "step": 22030 }, { "epoch": 3.46, "learning_rate": 6.170980392156864e-06, "loss": 0.3934, "step": 22040 }, { "epoch": 3.46, "learning_rate": 6.164705882352941e-06, "loss": 0.272, "step": 22050 }, { "epoch": 3.46, "learning_rate": 6.15843137254902e-06, "loss": 0.4084, "step": 22060 }, { "epoch": 3.46, "learning_rate": 6.152156862745099e-06, "loss": 0.2978, "step": 22070 }, { "epoch": 3.46, "learning_rate": 6.145882352941177e-06, "loss": 0.2924, "step": 22080 }, { "epoch": 3.47, "learning_rate": 6.139607843137255e-06, "loss": 0.3113, "step": 22090 }, { "epoch": 3.47, "learning_rate": 6.133333333333334e-06, "loss": 0.2647, "step": 22100 }, { "epoch": 3.47, "learning_rate": 6.127058823529413e-06, "loss": 0.442, "step": 22110 }, { "epoch": 3.47, "learning_rate": 6.1207843137254906e-06, "loss": 0.3169, "step": 22120 }, { "epoch": 3.47, "learning_rate": 6.114509803921569e-06, "loss": 0.2181, "step": 22130 }, { "epoch": 3.47, "learning_rate": 6.108235294117647e-06, "loss": 0.2393, "step": 22140 }, { "epoch": 3.47, "learning_rate": 6.101960784313726e-06, "loss": 0.1104, "step": 22150 }, { "epoch": 3.48, "learning_rate": 6.0956862745098045e-06, "loss": 0.3532, "step": 22160 }, { "epoch": 3.48, "learning_rate": 6.089411764705883e-06, "loss": 0.2957, "step": 22170 }, { "epoch": 3.48, "learning_rate": 6.083137254901961e-06, "loss": 0.4758, "step": 22180 }, { "epoch": 3.48, "learning_rate": 6.07686274509804e-06, "loss": 0.3497, "step": 22190 }, { "epoch": 3.48, "learning_rate": 6.070588235294118e-06, "loss": 0.297, "step": 22200 }, { "epoch": 3.48, "learning_rate": 6.064313725490196e-06, "loss": 0.3351, "step": 22210 }, { "epoch": 3.49, "learning_rate": 6.058039215686275e-06, "loss": 0.3607, "step": 22220 }, { "epoch": 3.49, "learning_rate": 6.051764705882354e-06, "loss": 0.3728, "step": 22230 }, { "epoch": 3.49, "learning_rate": 6.045490196078431e-06, "loss": 0.3178, "step": 22240 }, { "epoch": 3.49, "learning_rate": 6.03921568627451e-06, "loss": 0.3194, "step": 22250 }, { "epoch": 3.49, "learning_rate": 6.032941176470589e-06, "loss": 0.3292, "step": 22260 }, { "epoch": 3.49, "learning_rate": 6.026666666666668e-06, "loss": 0.2281, "step": 22270 }, { "epoch": 3.49, "learning_rate": 6.020392156862745e-06, "loss": 0.1824, "step": 22280 }, { "epoch": 3.5, "learning_rate": 6.014117647058824e-06, "loss": 0.0847, "step": 22290 }, { "epoch": 3.5, "learning_rate": 6.007843137254903e-06, "loss": 0.3106, "step": 22300 }, { "epoch": 3.5, "learning_rate": 6.001568627450981e-06, "loss": 0.2409, "step": 22310 }, { "epoch": 3.5, "learning_rate": 5.995294117647059e-06, "loss": 0.3339, "step": 22320 }, { "epoch": 3.5, "learning_rate": 5.989019607843138e-06, "loss": 0.462, "step": 22330 }, { "epoch": 3.5, "learning_rate": 5.982745098039216e-06, "loss": 0.3086, "step": 22340 }, { "epoch": 3.51, "learning_rate": 5.976470588235295e-06, "loss": 0.1435, "step": 22350 }, { "epoch": 3.51, "learning_rate": 5.970196078431373e-06, "loss": 0.3488, "step": 22360 }, { "epoch": 3.51, "learning_rate": 5.963921568627451e-06, "loss": 0.1163, "step": 22370 }, { "epoch": 3.51, "learning_rate": 5.95764705882353e-06, "loss": 0.2784, "step": 22380 }, { "epoch": 3.51, "learning_rate": 5.9513725490196085e-06, "loss": 0.4844, "step": 22390 }, { "epoch": 3.51, "learning_rate": 5.945098039215686e-06, "loss": 0.3983, "step": 22400 }, { "epoch": 3.52, "learning_rate": 5.938823529411765e-06, "loss": 0.3001, "step": 22410 }, { "epoch": 3.52, "learning_rate": 5.932549019607844e-06, "loss": 0.2631, "step": 22420 }, { "epoch": 3.52, "learning_rate": 5.9262745098039224e-06, "loss": 0.3404, "step": 22430 }, { "epoch": 3.52, "learning_rate": 5.92e-06, "loss": 0.2815, "step": 22440 }, { "epoch": 3.52, "learning_rate": 5.913725490196079e-06, "loss": 0.3247, "step": 22450 }, { "epoch": 3.52, "learning_rate": 5.907450980392158e-06, "loss": 0.4596, "step": 22460 }, { "epoch": 3.52, "learning_rate": 5.9011764705882355e-06, "loss": 0.3834, "step": 22470 }, { "epoch": 3.53, "learning_rate": 5.894901960784314e-06, "loss": 0.3683, "step": 22480 }, { "epoch": 3.53, "learning_rate": 5.888627450980393e-06, "loss": 0.3638, "step": 22490 }, { "epoch": 3.53, "learning_rate": 5.882352941176471e-06, "loss": 0.3319, "step": 22500 }, { "epoch": 3.53, "learning_rate": 5.876078431372549e-06, "loss": 0.2029, "step": 22510 }, { "epoch": 3.53, "learning_rate": 5.869803921568628e-06, "loss": 0.4986, "step": 22520 }, { "epoch": 3.53, "learning_rate": 5.863529411764707e-06, "loss": 0.1616, "step": 22530 }, { "epoch": 3.54, "learning_rate": 5.857254901960785e-06, "loss": 0.5609, "step": 22540 }, { "epoch": 3.54, "learning_rate": 5.850980392156863e-06, "loss": 0.2917, "step": 22550 }, { "epoch": 3.54, "learning_rate": 5.844705882352941e-06, "loss": 0.2516, "step": 22560 }, { "epoch": 3.54, "learning_rate": 5.83843137254902e-06, "loss": 0.4093, "step": 22570 }, { "epoch": 3.54, "learning_rate": 5.832156862745099e-06, "loss": 0.1713, "step": 22580 }, { "epoch": 3.54, "learning_rate": 5.825882352941176e-06, "loss": 0.3056, "step": 22590 }, { "epoch": 3.55, "learning_rate": 5.819607843137255e-06, "loss": 0.3793, "step": 22600 }, { "epoch": 3.55, "learning_rate": 5.813333333333334e-06, "loss": 0.4206, "step": 22610 }, { "epoch": 3.55, "learning_rate": 5.807058823529413e-06, "loss": 0.3064, "step": 22620 }, { "epoch": 3.55, "learning_rate": 5.80078431372549e-06, "loss": 0.3326, "step": 22630 }, { "epoch": 3.55, "learning_rate": 5.794509803921569e-06, "loss": 0.4033, "step": 22640 }, { "epoch": 3.55, "learning_rate": 5.788235294117648e-06, "loss": 0.2483, "step": 22650 }, { "epoch": 3.55, "learning_rate": 5.781960784313726e-06, "loss": 0.3099, "step": 22660 }, { "epoch": 3.56, "learning_rate": 5.775686274509804e-06, "loss": 0.3653, "step": 22670 }, { "epoch": 3.56, "learning_rate": 5.7694117647058835e-06, "loss": 0.4163, "step": 22680 }, { "epoch": 3.56, "learning_rate": 5.763137254901962e-06, "loss": 0.3538, "step": 22690 }, { "epoch": 3.56, "learning_rate": 5.7568627450980396e-06, "loss": 0.2336, "step": 22700 }, { "epoch": 3.56, "learning_rate": 5.750588235294118e-06, "loss": 0.3732, "step": 22710 }, { "epoch": 3.56, "learning_rate": 5.744313725490197e-06, "loss": 0.4794, "step": 22720 }, { "epoch": 3.57, "learning_rate": 5.738039215686275e-06, "loss": 0.3026, "step": 22730 }, { "epoch": 3.57, "learning_rate": 5.7317647058823535e-06, "loss": 0.6086, "step": 22740 }, { "epoch": 3.57, "learning_rate": 5.725490196078431e-06, "loss": 0.3586, "step": 22750 }, { "epoch": 3.57, "learning_rate": 5.71921568627451e-06, "loss": 0.2248, "step": 22760 }, { "epoch": 3.57, "learning_rate": 5.712941176470589e-06, "loss": 0.3199, "step": 22770 }, { "epoch": 3.57, "learning_rate": 5.706666666666667e-06, "loss": 0.2613, "step": 22780 }, { "epoch": 3.57, "learning_rate": 5.700392156862745e-06, "loss": 0.1409, "step": 22790 }, { "epoch": 3.58, "learning_rate": 5.694117647058824e-06, "loss": 0.2595, "step": 22800 }, { "epoch": 3.58, "learning_rate": 5.687843137254903e-06, "loss": 0.2032, "step": 22810 }, { "epoch": 3.58, "learning_rate": 5.68156862745098e-06, "loss": 0.28, "step": 22820 }, { "epoch": 3.58, "learning_rate": 5.675294117647059e-06, "loss": 0.255, "step": 22830 }, { "epoch": 3.58, "learning_rate": 5.669019607843138e-06, "loss": 0.3291, "step": 22840 }, { "epoch": 3.58, "learning_rate": 5.662745098039216e-06, "loss": 0.3033, "step": 22850 }, { "epoch": 3.59, "learning_rate": 5.656470588235294e-06, "loss": 0.5413, "step": 22860 }, { "epoch": 3.59, "learning_rate": 5.650196078431373e-06, "loss": 0.1827, "step": 22870 }, { "epoch": 3.59, "learning_rate": 5.643921568627452e-06, "loss": 0.2419, "step": 22880 }, { "epoch": 3.59, "learning_rate": 5.63764705882353e-06, "loss": 0.3047, "step": 22890 }, { "epoch": 3.59, "learning_rate": 5.631372549019608e-06, "loss": 0.1902, "step": 22900 }, { "epoch": 3.59, "learning_rate": 5.625098039215686e-06, "loss": 0.2311, "step": 22910 }, { "epoch": 3.6, "learning_rate": 5.618823529411765e-06, "loss": 0.3128, "step": 22920 }, { "epoch": 3.6, "learning_rate": 5.612549019607844e-06, "loss": 0.2971, "step": 22930 }, { "epoch": 3.6, "learning_rate": 5.606274509803922e-06, "loss": 0.3142, "step": 22940 }, { "epoch": 3.6, "learning_rate": 5.600000000000001e-06, "loss": 0.5576, "step": 22950 }, { "epoch": 3.6, "learning_rate": 5.593725490196079e-06, "loss": 0.5117, "step": 22960 }, { "epoch": 3.6, "learning_rate": 5.5874509803921575e-06, "loss": 0.4667, "step": 22970 }, { "epoch": 3.6, "learning_rate": 5.581176470588235e-06, "loss": 0.2978, "step": 22980 }, { "epoch": 3.61, "learning_rate": 5.5749019607843145e-06, "loss": 0.2945, "step": 22990 }, { "epoch": 3.61, "learning_rate": 5.568627450980393e-06, "loss": 0.3326, "step": 23000 }, { "epoch": 3.61, "learning_rate": 5.562352941176471e-06, "loss": 0.3256, "step": 23010 }, { "epoch": 3.61, "learning_rate": 5.556078431372549e-06, "loss": 0.3996, "step": 23020 }, { "epoch": 3.61, "learning_rate": 5.549803921568628e-06, "loss": 0.3687, "step": 23030 }, { "epoch": 3.61, "learning_rate": 5.543529411764707e-06, "loss": 0.3361, "step": 23040 }, { "epoch": 3.62, "learning_rate": 5.5372549019607845e-06, "loss": 0.3215, "step": 23050 }, { "epoch": 3.62, "learning_rate": 5.530980392156863e-06, "loss": 0.1895, "step": 23060 }, { "epoch": 3.62, "learning_rate": 5.524705882352942e-06, "loss": 0.2253, "step": 23070 }, { "epoch": 3.62, "learning_rate": 5.51843137254902e-06, "loss": 0.2576, "step": 23080 }, { "epoch": 3.62, "learning_rate": 5.512156862745098e-06, "loss": 0.2271, "step": 23090 }, { "epoch": 3.62, "learning_rate": 5.505882352941177e-06, "loss": 0.2629, "step": 23100 }, { "epoch": 3.63, "learning_rate": 5.499607843137255e-06, "loss": 0.2931, "step": 23110 }, { "epoch": 3.63, "learning_rate": 5.493333333333334e-06, "loss": 0.3659, "step": 23120 }, { "epoch": 3.63, "learning_rate": 5.487058823529412e-06, "loss": 0.4464, "step": 23130 }, { "epoch": 3.63, "learning_rate": 5.48078431372549e-06, "loss": 0.2321, "step": 23140 }, { "epoch": 3.63, "learning_rate": 5.474509803921569e-06, "loss": 0.3611, "step": 23150 }, { "epoch": 3.63, "learning_rate": 5.468235294117648e-06, "loss": 0.2949, "step": 23160 }, { "epoch": 3.63, "learning_rate": 5.461960784313725e-06, "loss": 0.3396, "step": 23170 }, { "epoch": 3.64, "learning_rate": 5.455686274509804e-06, "loss": 0.3646, "step": 23180 }, { "epoch": 3.64, "learning_rate": 5.449411764705883e-06, "loss": 0.2444, "step": 23190 }, { "epoch": 3.64, "learning_rate": 5.443137254901962e-06, "loss": 0.2918, "step": 23200 }, { "epoch": 3.64, "learning_rate": 5.436862745098039e-06, "loss": 0.4721, "step": 23210 }, { "epoch": 3.64, "learning_rate": 5.4305882352941186e-06, "loss": 0.4258, "step": 23220 }, { "epoch": 3.64, "learning_rate": 5.424313725490197e-06, "loss": 0.206, "step": 23230 }, { "epoch": 3.65, "learning_rate": 5.418039215686275e-06, "loss": 0.4885, "step": 23240 }, { "epoch": 3.65, "learning_rate": 5.411764705882353e-06, "loss": 0.391, "step": 23250 }, { "epoch": 3.65, "learning_rate": 5.4054901960784325e-06, "loss": 0.4066, "step": 23260 }, { "epoch": 3.65, "learning_rate": 5.39921568627451e-06, "loss": 0.4933, "step": 23270 }, { "epoch": 3.65, "learning_rate": 5.3929411764705886e-06, "loss": 0.204, "step": 23280 }, { "epoch": 3.65, "learning_rate": 5.386666666666667e-06, "loss": 0.349, "step": 23290 }, { "epoch": 3.65, "learning_rate": 5.380392156862746e-06, "loss": 0.2816, "step": 23300 }, { "epoch": 3.66, "learning_rate": 5.374117647058824e-06, "loss": 0.2211, "step": 23310 }, { "epoch": 3.66, "learning_rate": 5.3678431372549025e-06, "loss": 0.1313, "step": 23320 }, { "epoch": 3.66, "learning_rate": 5.36156862745098e-06, "loss": 0.2509, "step": 23330 }, { "epoch": 3.66, "learning_rate": 5.355294117647059e-06, "loss": 0.3012, "step": 23340 }, { "epoch": 3.66, "learning_rate": 5.349019607843138e-06, "loss": 0.4367, "step": 23350 }, { "epoch": 3.66, "learning_rate": 5.3427450980392155e-06, "loss": 0.3744, "step": 23360 }, { "epoch": 3.67, "learning_rate": 5.336470588235294e-06, "loss": 0.1066, "step": 23370 }, { "epoch": 3.67, "learning_rate": 5.330196078431373e-06, "loss": 0.4749, "step": 23380 }, { "epoch": 3.67, "learning_rate": 5.323921568627452e-06, "loss": 0.2062, "step": 23390 }, { "epoch": 3.67, "learning_rate": 5.317647058823529e-06, "loss": 0.4018, "step": 23400 }, { "epoch": 3.67, "learning_rate": 5.311372549019608e-06, "loss": 0.439, "step": 23410 }, { "epoch": 3.67, "learning_rate": 5.305098039215687e-06, "loss": 0.1355, "step": 23420 }, { "epoch": 3.68, "learning_rate": 5.298823529411765e-06, "loss": 0.2334, "step": 23430 }, { "epoch": 3.68, "learning_rate": 5.292549019607843e-06, "loss": 0.3339, "step": 23440 }, { "epoch": 3.68, "learning_rate": 5.286274509803923e-06, "loss": 0.2578, "step": 23450 }, { "epoch": 3.68, "learning_rate": 5.28e-06, "loss": 0.4641, "step": 23460 }, { "epoch": 3.68, "learning_rate": 5.273725490196079e-06, "loss": 0.4322, "step": 23470 }, { "epoch": 3.68, "learning_rate": 5.267450980392157e-06, "loss": 0.2521, "step": 23480 }, { "epoch": 3.68, "learning_rate": 5.2611764705882365e-06, "loss": 0.3453, "step": 23490 }, { "epoch": 3.69, "learning_rate": 5.254901960784314e-06, "loss": 0.2481, "step": 23500 }, { "epoch": 3.69, "learning_rate": 5.248627450980393e-06, "loss": 0.2984, "step": 23510 }, { "epoch": 3.69, "learning_rate": 5.24235294117647e-06, "loss": 0.3497, "step": 23520 }, { "epoch": 3.69, "learning_rate": 5.23607843137255e-06, "loss": 0.3718, "step": 23530 }, { "epoch": 3.69, "learning_rate": 5.229803921568628e-06, "loss": 0.1612, "step": 23540 }, { "epoch": 3.69, "learning_rate": 5.2235294117647065e-06, "loss": 0.4743, "step": 23550 }, { "epoch": 3.7, "learning_rate": 5.217254901960784e-06, "loss": 0.2661, "step": 23560 }, { "epoch": 3.7, "learning_rate": 5.2109803921568635e-06, "loss": 0.1501, "step": 23570 }, { "epoch": 3.7, "learning_rate": 5.204705882352942e-06, "loss": 0.4788, "step": 23580 }, { "epoch": 3.7, "learning_rate": 5.19843137254902e-06, "loss": 0.3264, "step": 23590 }, { "epoch": 3.7, "learning_rate": 5.192156862745098e-06, "loss": 0.385, "step": 23600 }, { "epoch": 3.7, "learning_rate": 5.185882352941177e-06, "loss": 0.4229, "step": 23610 }, { "epoch": 3.71, "learning_rate": 5.179607843137255e-06, "loss": 0.5238, "step": 23620 }, { "epoch": 3.71, "learning_rate": 5.1733333333333335e-06, "loss": 0.5224, "step": 23630 }, { "epoch": 3.71, "learning_rate": 5.167058823529412e-06, "loss": 0.3669, "step": 23640 }, { "epoch": 3.71, "learning_rate": 5.160784313725491e-06, "loss": 0.2825, "step": 23650 }, { "epoch": 3.71, "learning_rate": 5.154509803921569e-06, "loss": 0.3713, "step": 23660 }, { "epoch": 3.71, "learning_rate": 5.148235294117647e-06, "loss": 0.4378, "step": 23670 }, { "epoch": 3.71, "learning_rate": 5.141960784313725e-06, "loss": 0.4405, "step": 23680 }, { "epoch": 3.72, "learning_rate": 5.135686274509804e-06, "loss": 0.4131, "step": 23690 }, { "epoch": 3.72, "learning_rate": 5.129411764705883e-06, "loss": 0.2855, "step": 23700 }, { "epoch": 3.72, "learning_rate": 5.123137254901961e-06, "loss": 0.339, "step": 23710 }, { "epoch": 3.72, "learning_rate": 5.11686274509804e-06, "loss": 0.3575, "step": 23720 }, { "epoch": 3.72, "learning_rate": 5.110588235294118e-06, "loss": 0.3148, "step": 23730 }, { "epoch": 3.72, "learning_rate": 5.104313725490197e-06, "loss": 0.3298, "step": 23740 }, { "epoch": 3.73, "learning_rate": 5.098039215686274e-06, "loss": 0.3884, "step": 23750 }, { "epoch": 3.73, "learning_rate": 5.091764705882354e-06, "loss": 0.4189, "step": 23760 }, { "epoch": 3.73, "learning_rate": 5.085490196078432e-06, "loss": 0.1289, "step": 23770 }, { "epoch": 3.73, "learning_rate": 5.07921568627451e-06, "loss": 0.2621, "step": 23780 }, { "epoch": 3.73, "learning_rate": 5.072941176470588e-06, "loss": 0.1583, "step": 23790 }, { "epoch": 3.73, "learning_rate": 5.0666666666666676e-06, "loss": 0.4014, "step": 23800 }, { "epoch": 3.73, "learning_rate": 5.060392156862746e-06, "loss": 0.358, "step": 23810 }, { "epoch": 3.74, "learning_rate": 5.054117647058824e-06, "loss": 0.2629, "step": 23820 }, { "epoch": 3.74, "learning_rate": 5.047843137254902e-06, "loss": 0.2564, "step": 23830 }, { "epoch": 3.74, "learning_rate": 5.0415686274509815e-06, "loss": 0.3719, "step": 23840 }, { "epoch": 3.74, "learning_rate": 5.035294117647059e-06, "loss": 0.3233, "step": 23850 }, { "epoch": 3.74, "learning_rate": 5.0290196078431376e-06, "loss": 0.4364, "step": 23860 }, { "epoch": 3.74, "learning_rate": 5.022745098039215e-06, "loss": 0.4308, "step": 23870 }, { "epoch": 3.75, "learning_rate": 5.0164705882352945e-06, "loss": 0.3311, "step": 23880 }, { "epoch": 3.75, "learning_rate": 5.010196078431373e-06, "loss": 0.307, "step": 23890 }, { "epoch": 3.75, "learning_rate": 5.0039215686274515e-06, "loss": 0.1966, "step": 23900 }, { "epoch": 3.75, "learning_rate": 4.99764705882353e-06, "loss": 0.5216, "step": 23910 }, { "epoch": 3.75, "learning_rate": 4.9913725490196076e-06, "loss": 0.3298, "step": 23920 }, { "epoch": 3.75, "learning_rate": 4.985098039215687e-06, "loss": 0.2518, "step": 23930 }, { "epoch": 3.76, "learning_rate": 4.978823529411765e-06, "loss": 0.3248, "step": 23940 }, { "epoch": 3.76, "learning_rate": 4.972549019607844e-06, "loss": 0.2267, "step": 23950 }, { "epoch": 3.76, "learning_rate": 4.966274509803922e-06, "loss": 0.273, "step": 23960 }, { "epoch": 3.76, "learning_rate": 4.960000000000001e-06, "loss": 0.2015, "step": 23970 }, { "epoch": 3.76, "learning_rate": 4.953725490196079e-06, "loss": 0.1944, "step": 23980 }, { "epoch": 3.76, "learning_rate": 4.947450980392157e-06, "loss": 0.4129, "step": 23990 }, { "epoch": 3.76, "learning_rate": 4.941176470588236e-06, "loss": 0.2336, "step": 24000 }, { "epoch": 3.77, "learning_rate": 4.934901960784314e-06, "loss": 0.2867, "step": 24010 }, { "epoch": 3.77, "learning_rate": 4.928627450980393e-06, "loss": 0.5853, "step": 24020 }, { "epoch": 3.77, "learning_rate": 4.922352941176471e-06, "loss": 0.1643, "step": 24030 }, { "epoch": 3.77, "learning_rate": 4.916078431372549e-06, "loss": 0.2319, "step": 24040 }, { "epoch": 3.77, "learning_rate": 4.909803921568628e-06, "loss": 0.2861, "step": 24050 }, { "epoch": 3.77, "learning_rate": 4.903529411764706e-06, "loss": 0.4778, "step": 24060 }, { "epoch": 3.78, "learning_rate": 4.897254901960785e-06, "loss": 0.3719, "step": 24070 }, { "epoch": 3.78, "learning_rate": 4.890980392156863e-06, "loss": 0.3104, "step": 24080 }, { "epoch": 3.78, "learning_rate": 4.884705882352942e-06, "loss": 0.287, "step": 24090 }, { "epoch": 3.78, "learning_rate": 4.87843137254902e-06, "loss": 0.4196, "step": 24100 }, { "epoch": 3.78, "learning_rate": 4.872156862745099e-06, "loss": 0.1613, "step": 24110 }, { "epoch": 3.78, "learning_rate": 4.865882352941177e-06, "loss": 0.2659, "step": 24120 }, { "epoch": 3.79, "learning_rate": 4.859607843137255e-06, "loss": 0.3756, "step": 24130 }, { "epoch": 3.79, "learning_rate": 4.853333333333334e-06, "loss": 0.389, "step": 24140 }, { "epoch": 3.79, "learning_rate": 4.847058823529412e-06, "loss": 0.313, "step": 24150 }, { "epoch": 3.79, "learning_rate": 4.840784313725491e-06, "loss": 0.2495, "step": 24160 }, { "epoch": 3.79, "learning_rate": 4.834509803921569e-06, "loss": 0.349, "step": 24170 }, { "epoch": 3.79, "learning_rate": 4.828235294117647e-06, "loss": 0.2783, "step": 24180 }, { "epoch": 3.79, "learning_rate": 4.821960784313726e-06, "loss": 0.3273, "step": 24190 }, { "epoch": 3.8, "learning_rate": 4.815686274509804e-06, "loss": 0.309, "step": 24200 }, { "epoch": 3.8, "learning_rate": 4.809411764705883e-06, "loss": 0.2513, "step": 24210 }, { "epoch": 3.8, "learning_rate": 4.803137254901961e-06, "loss": 0.621, "step": 24220 }, { "epoch": 3.8, "learning_rate": 4.7968627450980394e-06, "loss": 0.3172, "step": 24230 }, { "epoch": 3.8, "learning_rate": 4.790588235294118e-06, "loss": 0.2701, "step": 24240 }, { "epoch": 3.8, "learning_rate": 4.784313725490196e-06, "loss": 0.3008, "step": 24250 }, { "epoch": 3.81, "learning_rate": 4.778039215686275e-06, "loss": 0.195, "step": 24260 }, { "epoch": 3.81, "learning_rate": 4.771764705882353e-06, "loss": 0.2647, "step": 24270 }, { "epoch": 3.81, "learning_rate": 4.765490196078432e-06, "loss": 0.304, "step": 24280 }, { "epoch": 3.81, "learning_rate": 4.75921568627451e-06, "loss": 0.2196, "step": 24290 }, { "epoch": 3.81, "learning_rate": 4.752941176470589e-06, "loss": 0.2826, "step": 24300 }, { "epoch": 3.81, "learning_rate": 4.746666666666667e-06, "loss": 0.4066, "step": 24310 }, { "epoch": 3.81, "learning_rate": 4.740392156862746e-06, "loss": 0.2571, "step": 24320 }, { "epoch": 3.82, "learning_rate": 4.734117647058824e-06, "loss": 0.3834, "step": 24330 }, { "epoch": 3.82, "learning_rate": 4.727843137254902e-06, "loss": 0.1781, "step": 24340 }, { "epoch": 3.82, "learning_rate": 4.721568627450981e-06, "loss": 0.3864, "step": 24350 }, { "epoch": 3.82, "learning_rate": 4.715294117647059e-06, "loss": 0.2317, "step": 24360 }, { "epoch": 3.82, "learning_rate": 4.709019607843138e-06, "loss": 0.2239, "step": 24370 }, { "epoch": 3.82, "learning_rate": 4.702745098039216e-06, "loss": 0.3988, "step": 24380 }, { "epoch": 3.83, "learning_rate": 4.696470588235294e-06, "loss": 0.2347, "step": 24390 }, { "epoch": 3.83, "learning_rate": 4.690196078431373e-06, "loss": 0.1655, "step": 24400 }, { "epoch": 3.83, "learning_rate": 4.683921568627451e-06, "loss": 0.2624, "step": 24410 }, { "epoch": 3.83, "learning_rate": 4.67764705882353e-06, "loss": 0.2457, "step": 24420 }, { "epoch": 3.83, "learning_rate": 4.671372549019608e-06, "loss": 0.3744, "step": 24430 }, { "epoch": 3.83, "learning_rate": 4.6650980392156866e-06, "loss": 0.4075, "step": 24440 }, { "epoch": 3.84, "learning_rate": 4.658823529411765e-06, "loss": 0.3702, "step": 24450 }, { "epoch": 3.84, "learning_rate": 4.6525490196078435e-06, "loss": 0.3554, "step": 24460 }, { "epoch": 3.84, "learning_rate": 4.646274509803922e-06, "loss": 0.2586, "step": 24470 }, { "epoch": 3.84, "learning_rate": 4.6400000000000005e-06, "loss": 0.3246, "step": 24480 }, { "epoch": 3.84, "learning_rate": 4.633725490196079e-06, "loss": 0.1987, "step": 24490 }, { "epoch": 3.84, "learning_rate": 4.627450980392157e-06, "loss": 0.4617, "step": 24500 }, { "epoch": 3.84, "learning_rate": 4.621176470588236e-06, "loss": 0.4459, "step": 24510 }, { "epoch": 3.85, "learning_rate": 4.614901960784314e-06, "loss": 0.4515, "step": 24520 }, { "epoch": 3.85, "learning_rate": 4.608627450980393e-06, "loss": 0.1898, "step": 24530 }, { "epoch": 3.85, "learning_rate": 4.602352941176471e-06, "loss": 0.3266, "step": 24540 }, { "epoch": 3.85, "learning_rate": 4.596078431372549e-06, "loss": 0.3003, "step": 24550 }, { "epoch": 3.85, "learning_rate": 4.589803921568628e-06, "loss": 0.3409, "step": 24560 }, { "epoch": 3.85, "learning_rate": 4.583529411764706e-06, "loss": 0.2909, "step": 24570 }, { "epoch": 3.86, "learning_rate": 4.577254901960785e-06, "loss": 0.4174, "step": 24580 }, { "epoch": 3.86, "learning_rate": 4.570980392156863e-06, "loss": 0.268, "step": 24590 }, { "epoch": 3.86, "learning_rate": 4.564705882352941e-06, "loss": 0.2072, "step": 24600 }, { "epoch": 3.86, "learning_rate": 4.55843137254902e-06, "loss": 0.3457, "step": 24610 }, { "epoch": 3.86, "learning_rate": 4.552156862745098e-06, "loss": 0.4213, "step": 24620 }, { "epoch": 3.86, "learning_rate": 4.545882352941177e-06, "loss": 0.398, "step": 24630 }, { "epoch": 3.87, "learning_rate": 4.539607843137255e-06, "loss": 0.401, "step": 24640 }, { "epoch": 3.87, "learning_rate": 4.533333333333334e-06, "loss": 0.2373, "step": 24650 }, { "epoch": 3.87, "learning_rate": 4.527058823529412e-06, "loss": 0.2786, "step": 24660 }, { "epoch": 3.87, "learning_rate": 4.520784313725491e-06, "loss": 0.3434, "step": 24670 }, { "epoch": 3.87, "learning_rate": 4.514509803921569e-06, "loss": 0.1008, "step": 24680 }, { "epoch": 3.87, "learning_rate": 4.508235294117647e-06, "loss": 0.3966, "step": 24690 }, { "epoch": 3.87, "learning_rate": 4.501960784313726e-06, "loss": 0.3781, "step": 24700 }, { "epoch": 3.88, "learning_rate": 4.4956862745098045e-06, "loss": 0.2536, "step": 24710 }, { "epoch": 3.88, "learning_rate": 4.489411764705883e-06, "loss": 0.2623, "step": 24720 }, { "epoch": 3.88, "learning_rate": 4.4831372549019615e-06, "loss": 0.3711, "step": 24730 }, { "epoch": 3.88, "learning_rate": 4.476862745098039e-06, "loss": 0.5189, "step": 24740 }, { "epoch": 3.88, "learning_rate": 4.4705882352941184e-06, "loss": 0.2207, "step": 24750 }, { "epoch": 3.88, "learning_rate": 4.464313725490196e-06, "loss": 0.2042, "step": 24760 }, { "epoch": 3.89, "learning_rate": 4.458039215686275e-06, "loss": 0.2143, "step": 24770 }, { "epoch": 3.89, "learning_rate": 4.451764705882353e-06, "loss": 0.2593, "step": 24780 }, { "epoch": 3.89, "learning_rate": 4.4454901960784315e-06, "loss": 0.1756, "step": 24790 }, { "epoch": 3.89, "learning_rate": 4.43921568627451e-06, "loss": 0.2708, "step": 24800 }, { "epoch": 3.89, "learning_rate": 4.4329411764705884e-06, "loss": 0.3586, "step": 24810 }, { "epoch": 3.89, "learning_rate": 4.426666666666667e-06, "loss": 0.3573, "step": 24820 }, { "epoch": 3.89, "learning_rate": 4.420392156862745e-06, "loss": 0.3139, "step": 24830 }, { "epoch": 3.9, "learning_rate": 4.414117647058824e-06, "loss": 0.3016, "step": 24840 }, { "epoch": 3.9, "learning_rate": 4.407843137254902e-06, "loss": 0.3712, "step": 24850 }, { "epoch": 3.9, "learning_rate": 4.401568627450981e-06, "loss": 0.3798, "step": 24860 }, { "epoch": 3.9, "learning_rate": 4.395294117647059e-06, "loss": 0.4728, "step": 24870 }, { "epoch": 3.9, "learning_rate": 4.389019607843138e-06, "loss": 0.5253, "step": 24880 }, { "epoch": 3.9, "learning_rate": 4.382745098039216e-06, "loss": 0.219, "step": 24890 }, { "epoch": 3.91, "learning_rate": 4.376470588235294e-06, "loss": 0.4219, "step": 24900 }, { "epoch": 3.91, "learning_rate": 4.370196078431373e-06, "loss": 0.2848, "step": 24910 }, { "epoch": 3.91, "learning_rate": 4.363921568627451e-06, "loss": 0.4223, "step": 24920 }, { "epoch": 3.91, "learning_rate": 4.35764705882353e-06, "loss": 0.3831, "step": 24930 }, { "epoch": 3.91, "learning_rate": 4.351372549019608e-06, "loss": 0.29, "step": 24940 }, { "epoch": 3.91, "learning_rate": 4.345098039215686e-06, "loss": 0.2312, "step": 24950 }, { "epoch": 3.92, "learning_rate": 4.3388235294117656e-06, "loss": 0.2451, "step": 24960 }, { "epoch": 3.92, "learning_rate": 4.332549019607843e-06, "loss": 0.4812, "step": 24970 }, { "epoch": 3.92, "learning_rate": 4.3262745098039225e-06, "loss": 0.4033, "step": 24980 }, { "epoch": 3.92, "learning_rate": 4.32e-06, "loss": 0.1514, "step": 24990 }, { "epoch": 3.92, "learning_rate": 4.313725490196079e-06, "loss": 0.2522, "step": 25000 }, { "epoch": 3.92, "learning_rate": 4.307450980392157e-06, "loss": 0.4003, "step": 25010 }, { "epoch": 3.92, "learning_rate": 4.3011764705882356e-06, "loss": 0.2281, "step": 25020 }, { "epoch": 3.93, "learning_rate": 4.294901960784314e-06, "loss": 0.2464, "step": 25030 }, { "epoch": 3.93, "learning_rate": 4.2886274509803925e-06, "loss": 0.5021, "step": 25040 }, { "epoch": 3.93, "learning_rate": 4.282352941176471e-06, "loss": 0.3742, "step": 25050 }, { "epoch": 3.93, "learning_rate": 4.2760784313725495e-06, "loss": 0.3564, "step": 25060 }, { "epoch": 3.93, "learning_rate": 4.269803921568628e-06, "loss": 0.2964, "step": 25070 }, { "epoch": 3.93, "learning_rate": 4.263529411764706e-06, "loss": 0.1332, "step": 25080 }, { "epoch": 3.94, "learning_rate": 4.257254901960785e-06, "loss": 0.237, "step": 25090 }, { "epoch": 3.94, "learning_rate": 4.250980392156863e-06, "loss": 0.4658, "step": 25100 }, { "epoch": 3.94, "learning_rate": 4.244705882352941e-06, "loss": 0.4726, "step": 25110 }, { "epoch": 3.94, "learning_rate": 4.23843137254902e-06, "loss": 0.4457, "step": 25120 }, { "epoch": 3.94, "learning_rate": 4.232156862745098e-06, "loss": 0.311, "step": 25130 }, { "epoch": 3.94, "learning_rate": 4.225882352941177e-06, "loss": 0.4378, "step": 25140 }, { "epoch": 3.95, "learning_rate": 4.219607843137255e-06, "loss": 0.1323, "step": 25150 }, { "epoch": 3.95, "learning_rate": 4.213333333333333e-06, "loss": 0.2633, "step": 25160 }, { "epoch": 3.95, "learning_rate": 4.207058823529412e-06, "loss": 0.1922, "step": 25170 }, { "epoch": 3.95, "learning_rate": 4.20078431372549e-06, "loss": 0.2337, "step": 25180 }, { "epoch": 3.95, "learning_rate": 4.194509803921569e-06, "loss": 0.2976, "step": 25190 }, { "epoch": 3.95, "learning_rate": 4.188235294117647e-06, "loss": 0.422, "step": 25200 }, { "epoch": 3.95, "learning_rate": 4.181960784313726e-06, "loss": 0.2545, "step": 25210 }, { "epoch": 3.96, "learning_rate": 4.175686274509804e-06, "loss": 0.2287, "step": 25220 }, { "epoch": 3.96, "learning_rate": 4.169411764705883e-06, "loss": 0.3911, "step": 25230 }, { "epoch": 3.96, "learning_rate": 4.163137254901961e-06, "loss": 0.1544, "step": 25240 }, { "epoch": 3.96, "learning_rate": 4.15686274509804e-06, "loss": 0.1616, "step": 25250 }, { "epoch": 3.96, "learning_rate": 4.150588235294118e-06, "loss": 0.5267, "step": 25260 }, { "epoch": 3.96, "learning_rate": 4.144313725490197e-06, "loss": 0.3285, "step": 25270 }, { "epoch": 3.97, "learning_rate": 4.138039215686275e-06, "loss": 0.3379, "step": 25280 }, { "epoch": 3.97, "learning_rate": 4.1317647058823535e-06, "loss": 0.3125, "step": 25290 }, { "epoch": 3.97, "learning_rate": 4.125490196078432e-06, "loss": 0.3495, "step": 25300 }, { "epoch": 3.97, "learning_rate": 4.1192156862745105e-06, "loss": 0.3336, "step": 25310 }, { "epoch": 3.97, "learning_rate": 4.112941176470588e-06, "loss": 0.304, "step": 25320 }, { "epoch": 3.97, "learning_rate": 4.1066666666666674e-06, "loss": 0.2138, "step": 25330 }, { "epoch": 3.97, "learning_rate": 4.100392156862745e-06, "loss": 0.2003, "step": 25340 }, { "epoch": 3.98, "learning_rate": 4.094117647058824e-06, "loss": 0.2517, "step": 25350 }, { "epoch": 3.98, "learning_rate": 4.087843137254902e-06, "loss": 0.3191, "step": 25360 }, { "epoch": 3.98, "learning_rate": 4.0815686274509805e-06, "loss": 0.398, "step": 25370 }, { "epoch": 3.98, "learning_rate": 4.075294117647059e-06, "loss": 0.4274, "step": 25380 }, { "epoch": 3.98, "learning_rate": 4.0690196078431374e-06, "loss": 0.2428, "step": 25390 }, { "epoch": 3.98, "learning_rate": 4.062745098039216e-06, "loss": 0.3154, "step": 25400 }, { "epoch": 3.99, "learning_rate": 4.056470588235294e-06, "loss": 0.2873, "step": 25410 }, { "epoch": 3.99, "learning_rate": 4.050196078431373e-06, "loss": 0.2401, "step": 25420 }, { "epoch": 3.99, "learning_rate": 4.043921568627451e-06, "loss": 0.28, "step": 25430 }, { "epoch": 3.99, "learning_rate": 4.03764705882353e-06, "loss": 0.4423, "step": 25440 }, { "epoch": 3.99, "learning_rate": 4.031372549019608e-06, "loss": 0.2021, "step": 25450 }, { "epoch": 3.99, "learning_rate": 4.025098039215687e-06, "loss": 0.3742, "step": 25460 }, { "epoch": 4.0, "learning_rate": 4.018823529411765e-06, "loss": 0.4138, "step": 25470 }, { "epoch": 4.0, "learning_rate": 4.012549019607844e-06, "loss": 0.3185, "step": 25480 }, { "epoch": 4.0, "learning_rate": 4.006274509803922e-06, "loss": 0.3359, "step": 25490 }, { "epoch": 4.0, "learning_rate": 4.000000000000001e-06, "loss": 0.3094, "step": 25500 }, { "epoch": 4.0, "eval_accuracy": 0.9935555555555555, "eval_loss": 0.026872247457504272, "eval_runtime": 182.6962, "eval_samples_per_second": 49.262, "eval_steps_per_second": 6.158, "step": 25500 }, { "epoch": 4.0, "learning_rate": 3.993725490196078e-06, "loss": 0.5308, "step": 25510 }, { "epoch": 4.0, "learning_rate": 3.987450980392158e-06, "loss": 0.3065, "step": 25520 }, { "epoch": 4.0, "learning_rate": 3.981176470588235e-06, "loss": 0.2491, "step": 25530 }, { "epoch": 4.01, "learning_rate": 3.9749019607843146e-06, "loss": 0.2459, "step": 25540 }, { "epoch": 4.01, "learning_rate": 3.968627450980392e-06, "loss": 0.2854, "step": 25550 }, { "epoch": 4.01, "learning_rate": 3.962352941176471e-06, "loss": 0.2536, "step": 25560 }, { "epoch": 4.01, "learning_rate": 3.956078431372549e-06, "loss": 0.2934, "step": 25570 }, { "epoch": 4.01, "learning_rate": 3.949803921568628e-06, "loss": 0.3308, "step": 25580 }, { "epoch": 4.01, "learning_rate": 3.943529411764706e-06, "loss": 0.3403, "step": 25590 }, { "epoch": 4.02, "learning_rate": 3.9372549019607846e-06, "loss": 0.3369, "step": 25600 }, { "epoch": 4.02, "learning_rate": 3.930980392156863e-06, "loss": 0.206, "step": 25610 }, { "epoch": 4.02, "learning_rate": 3.9247058823529415e-06, "loss": 0.3595, "step": 25620 }, { "epoch": 4.02, "learning_rate": 3.91843137254902e-06, "loss": 0.2242, "step": 25630 }, { "epoch": 4.02, "learning_rate": 3.9121568627450985e-06, "loss": 0.4616, "step": 25640 }, { "epoch": 4.02, "learning_rate": 3.905882352941177e-06, "loss": 0.3446, "step": 25650 }, { "epoch": 4.03, "learning_rate": 3.899607843137255e-06, "loss": 0.3024, "step": 25660 }, { "epoch": 4.03, "learning_rate": 3.893333333333333e-06, "loss": 0.3117, "step": 25670 }, { "epoch": 4.03, "learning_rate": 3.887058823529412e-06, "loss": 0.3924, "step": 25680 }, { "epoch": 4.03, "learning_rate": 3.88078431372549e-06, "loss": 0.4802, "step": 25690 }, { "epoch": 4.03, "learning_rate": 3.874509803921569e-06, "loss": 0.1926, "step": 25700 }, { "epoch": 4.03, "learning_rate": 3.868235294117648e-06, "loss": 0.1405, "step": 25710 }, { "epoch": 4.03, "learning_rate": 3.861960784313725e-06, "loss": 0.209, "step": 25720 }, { "epoch": 4.04, "learning_rate": 3.855686274509805e-06, "loss": 0.2438, "step": 25730 }, { "epoch": 4.04, "learning_rate": 3.849411764705882e-06, "loss": 0.2731, "step": 25740 }, { "epoch": 4.04, "learning_rate": 3.843137254901962e-06, "loss": 0.4251, "step": 25750 }, { "epoch": 4.04, "learning_rate": 3.836862745098039e-06, "loss": 0.5005, "step": 25760 }, { "epoch": 4.04, "learning_rate": 3.830588235294118e-06, "loss": 0.1843, "step": 25770 }, { "epoch": 4.04, "learning_rate": 3.824313725490196e-06, "loss": 0.131, "step": 25780 }, { "epoch": 4.05, "learning_rate": 3.818039215686275e-06, "loss": 0.3682, "step": 25790 }, { "epoch": 4.05, "learning_rate": 3.8117647058823532e-06, "loss": 0.3869, "step": 25800 }, { "epoch": 4.05, "learning_rate": 3.8054901960784317e-06, "loss": 0.3601, "step": 25810 }, { "epoch": 4.05, "learning_rate": 3.79921568627451e-06, "loss": 0.358, "step": 25820 }, { "epoch": 4.05, "learning_rate": 3.7929411764705886e-06, "loss": 0.1746, "step": 25830 }, { "epoch": 4.05, "learning_rate": 3.7866666666666667e-06, "loss": 0.1658, "step": 25840 }, { "epoch": 4.05, "learning_rate": 3.7803921568627456e-06, "loss": 0.3768, "step": 25850 }, { "epoch": 4.06, "learning_rate": 3.7741176470588236e-06, "loss": 0.1445, "step": 25860 }, { "epoch": 4.06, "learning_rate": 3.7678431372549025e-06, "loss": 0.3535, "step": 25870 }, { "epoch": 4.06, "learning_rate": 3.7615686274509806e-06, "loss": 0.4963, "step": 25880 }, { "epoch": 4.06, "learning_rate": 3.755294117647059e-06, "loss": 0.2007, "step": 25890 }, { "epoch": 4.06, "learning_rate": 3.7490196078431375e-06, "loss": 0.555, "step": 25900 }, { "epoch": 4.06, "learning_rate": 3.742745098039216e-06, "loss": 0.1997, "step": 25910 }, { "epoch": 4.07, "learning_rate": 3.736470588235294e-06, "loss": 0.3774, "step": 25920 }, { "epoch": 4.07, "learning_rate": 3.730196078431373e-06, "loss": 0.2067, "step": 25930 }, { "epoch": 4.07, "learning_rate": 3.723921568627451e-06, "loss": 0.1082, "step": 25940 }, { "epoch": 4.07, "learning_rate": 3.71764705882353e-06, "loss": 0.4187, "step": 25950 }, { "epoch": 4.07, "learning_rate": 3.7113725490196084e-06, "loss": 0.4062, "step": 25960 }, { "epoch": 4.07, "learning_rate": 3.7050980392156864e-06, "loss": 0.1896, "step": 25970 }, { "epoch": 4.08, "learning_rate": 3.6988235294117653e-06, "loss": 0.4701, "step": 25980 }, { "epoch": 4.08, "learning_rate": 3.6925490196078434e-06, "loss": 0.3285, "step": 25990 }, { "epoch": 4.08, "learning_rate": 3.6862745098039223e-06, "loss": 0.1288, "step": 26000 }, { "epoch": 4.08, "learning_rate": 3.6800000000000003e-06, "loss": 0.3045, "step": 26010 }, { "epoch": 4.08, "learning_rate": 3.673725490196079e-06, "loss": 0.2839, "step": 26020 }, { "epoch": 4.08, "learning_rate": 3.667450980392157e-06, "loss": 0.2376, "step": 26030 }, { "epoch": 4.08, "learning_rate": 3.6611764705882358e-06, "loss": 0.3831, "step": 26040 }, { "epoch": 4.09, "learning_rate": 3.654901960784314e-06, "loss": 0.5642, "step": 26050 }, { "epoch": 4.09, "learning_rate": 3.6486274509803927e-06, "loss": 0.2989, "step": 26060 }, { "epoch": 4.09, "learning_rate": 3.6423529411764708e-06, "loss": 0.4428, "step": 26070 }, { "epoch": 4.09, "learning_rate": 3.6360784313725497e-06, "loss": 0.4086, "step": 26080 }, { "epoch": 4.09, "learning_rate": 3.6298039215686277e-06, "loss": 0.5487, "step": 26090 }, { "epoch": 4.09, "learning_rate": 3.623529411764706e-06, "loss": 0.3256, "step": 26100 }, { "epoch": 4.1, "learning_rate": 3.6172549019607842e-06, "loss": 0.2581, "step": 26110 }, { "epoch": 4.1, "learning_rate": 3.610980392156863e-06, "loss": 0.2972, "step": 26120 }, { "epoch": 4.1, "learning_rate": 3.604705882352941e-06, "loss": 0.2889, "step": 26130 }, { "epoch": 4.1, "learning_rate": 3.59843137254902e-06, "loss": 0.3312, "step": 26140 }, { "epoch": 4.1, "learning_rate": 3.592156862745098e-06, "loss": 0.2974, "step": 26150 }, { "epoch": 4.1, "learning_rate": 3.5858823529411766e-06, "loss": 0.2402, "step": 26160 }, { "epoch": 4.11, "learning_rate": 3.579607843137255e-06, "loss": 0.3488, "step": 26170 }, { "epoch": 4.11, "learning_rate": 3.5733333333333336e-06, "loss": 0.2705, "step": 26180 }, { "epoch": 4.11, "learning_rate": 3.5670588235294116e-06, "loss": 0.3502, "step": 26190 }, { "epoch": 4.11, "learning_rate": 3.5607843137254905e-06, "loss": 0.2862, "step": 26200 }, { "epoch": 4.11, "learning_rate": 3.554509803921569e-06, "loss": 0.3453, "step": 26210 }, { "epoch": 4.11, "learning_rate": 3.5482352941176475e-06, "loss": 0.2388, "step": 26220 }, { "epoch": 4.11, "learning_rate": 3.541960784313726e-06, "loss": 0.4257, "step": 26230 }, { "epoch": 4.12, "learning_rate": 3.535686274509804e-06, "loss": 0.2949, "step": 26240 }, { "epoch": 4.12, "learning_rate": 3.529411764705883e-06, "loss": 0.3896, "step": 26250 }, { "epoch": 4.12, "learning_rate": 3.523137254901961e-06, "loss": 0.35, "step": 26260 }, { "epoch": 4.12, "learning_rate": 3.51686274509804e-06, "loss": 0.305, "step": 26270 }, { "epoch": 4.12, "learning_rate": 3.510588235294118e-06, "loss": 0.1966, "step": 26280 }, { "epoch": 4.12, "learning_rate": 3.5043137254901964e-06, "loss": 0.1797, "step": 26290 }, { "epoch": 4.13, "learning_rate": 3.498039215686275e-06, "loss": 0.2974, "step": 26300 }, { "epoch": 4.13, "learning_rate": 3.4917647058823533e-06, "loss": 0.264, "step": 26310 }, { "epoch": 4.13, "learning_rate": 3.4854901960784314e-06, "loss": 0.3848, "step": 26320 }, { "epoch": 4.13, "learning_rate": 3.4792156862745103e-06, "loss": 0.1224, "step": 26330 }, { "epoch": 4.13, "learning_rate": 3.4729411764705883e-06, "loss": 0.4456, "step": 26340 }, { "epoch": 4.13, "learning_rate": 3.4666666666666672e-06, "loss": 0.1914, "step": 26350 }, { "epoch": 4.13, "learning_rate": 3.4603921568627453e-06, "loss": 0.587, "step": 26360 }, { "epoch": 4.14, "learning_rate": 3.4541176470588237e-06, "loss": 0.3596, "step": 26370 }, { "epoch": 4.14, "learning_rate": 3.4478431372549022e-06, "loss": 0.1735, "step": 26380 }, { "epoch": 4.14, "learning_rate": 3.4415686274509807e-06, "loss": 0.3133, "step": 26390 }, { "epoch": 4.14, "learning_rate": 3.4352941176470587e-06, "loss": 0.2478, "step": 26400 }, { "epoch": 4.14, "learning_rate": 3.4290196078431376e-06, "loss": 0.334, "step": 26410 }, { "epoch": 4.14, "learning_rate": 3.4227450980392157e-06, "loss": 0.113, "step": 26420 }, { "epoch": 4.15, "learning_rate": 3.4164705882352946e-06, "loss": 0.3647, "step": 26430 }, { "epoch": 4.15, "learning_rate": 3.4101960784313726e-06, "loss": 0.1323, "step": 26440 }, { "epoch": 4.15, "learning_rate": 3.403921568627451e-06, "loss": 0.2675, "step": 26450 }, { "epoch": 4.15, "learning_rate": 3.39764705882353e-06, "loss": 0.3742, "step": 26460 }, { "epoch": 4.15, "learning_rate": 3.391372549019608e-06, "loss": 0.3095, "step": 26470 }, { "epoch": 4.15, "learning_rate": 3.385098039215687e-06, "loss": 0.2549, "step": 26480 }, { "epoch": 4.16, "learning_rate": 3.378823529411765e-06, "loss": 0.2703, "step": 26490 }, { "epoch": 4.16, "learning_rate": 3.3725490196078435e-06, "loss": 0.2437, "step": 26500 }, { "epoch": 4.16, "learning_rate": 3.366274509803922e-06, "loss": 0.3054, "step": 26510 }, { "epoch": 4.16, "learning_rate": 3.3600000000000004e-06, "loss": 0.4325, "step": 26520 }, { "epoch": 4.16, "learning_rate": 3.3537254901960785e-06, "loss": 0.3045, "step": 26530 }, { "epoch": 4.16, "learning_rate": 3.3474509803921574e-06, "loss": 0.2479, "step": 26540 }, { "epoch": 4.16, "learning_rate": 3.3411764705882354e-06, "loss": 0.1483, "step": 26550 }, { "epoch": 4.17, "learning_rate": 3.3349019607843143e-06, "loss": 0.3133, "step": 26560 }, { "epoch": 4.17, "learning_rate": 3.3286274509803924e-06, "loss": 0.2548, "step": 26570 }, { "epoch": 4.17, "learning_rate": 3.322352941176471e-06, "loss": 0.2884, "step": 26580 }, { "epoch": 4.17, "learning_rate": 3.3160784313725493e-06, "loss": 0.4415, "step": 26590 }, { "epoch": 4.17, "learning_rate": 3.309803921568628e-06, "loss": 0.5359, "step": 26600 }, { "epoch": 4.17, "learning_rate": 3.303529411764706e-06, "loss": 0.3087, "step": 26610 }, { "epoch": 4.18, "learning_rate": 3.2972549019607848e-06, "loss": 0.1654, "step": 26620 }, { "epoch": 4.18, "learning_rate": 3.290980392156863e-06, "loss": 0.2994, "step": 26630 }, { "epoch": 4.18, "learning_rate": 3.2847058823529417e-06, "loss": 0.3124, "step": 26640 }, { "epoch": 4.18, "learning_rate": 3.2784313725490198e-06, "loss": 0.3385, "step": 26650 }, { "epoch": 4.18, "learning_rate": 3.2721568627450982e-06, "loss": 0.3674, "step": 26660 }, { "epoch": 4.18, "learning_rate": 3.2658823529411763e-06, "loss": 0.1857, "step": 26670 }, { "epoch": 4.19, "learning_rate": 3.259607843137255e-06, "loss": 0.3431, "step": 26680 }, { "epoch": 4.19, "learning_rate": 3.2533333333333332e-06, "loss": 0.1259, "step": 26690 }, { "epoch": 4.19, "learning_rate": 3.247058823529412e-06, "loss": 0.3395, "step": 26700 }, { "epoch": 4.19, "learning_rate": 3.2407843137254906e-06, "loss": 0.4773, "step": 26710 }, { "epoch": 4.19, "learning_rate": 3.234509803921569e-06, "loss": 0.4332, "step": 26720 }, { "epoch": 4.19, "learning_rate": 3.2282352941176476e-06, "loss": 0.1739, "step": 26730 }, { "epoch": 4.19, "learning_rate": 3.2219607843137256e-06, "loss": 0.4886, "step": 26740 }, { "epoch": 4.2, "learning_rate": 3.2156862745098045e-06, "loss": 0.211, "step": 26750 }, { "epoch": 4.2, "learning_rate": 3.2094117647058826e-06, "loss": 0.3034, "step": 26760 }, { "epoch": 4.2, "learning_rate": 3.2031372549019615e-06, "loss": 0.186, "step": 26770 }, { "epoch": 4.2, "learning_rate": 3.1968627450980395e-06, "loss": 0.3721, "step": 26780 }, { "epoch": 4.2, "learning_rate": 3.190588235294118e-06, "loss": 0.3557, "step": 26790 }, { "epoch": 4.2, "learning_rate": 3.184313725490196e-06, "loss": 0.2961, "step": 26800 }, { "epoch": 4.21, "learning_rate": 3.178039215686275e-06, "loss": 0.3183, "step": 26810 }, { "epoch": 4.21, "learning_rate": 3.171764705882353e-06, "loss": 0.1062, "step": 26820 }, { "epoch": 4.21, "learning_rate": 3.165490196078432e-06, "loss": 0.4412, "step": 26830 }, { "epoch": 4.21, "learning_rate": 3.15921568627451e-06, "loss": 0.3607, "step": 26840 }, { "epoch": 4.21, "learning_rate": 3.1529411764705884e-06, "loss": 0.2671, "step": 26850 }, { "epoch": 4.21, "learning_rate": 3.146666666666667e-06, "loss": 0.1514, "step": 26860 }, { "epoch": 4.21, "learning_rate": 3.1403921568627454e-06, "loss": 0.2637, "step": 26870 }, { "epoch": 4.22, "learning_rate": 3.1341176470588234e-06, "loss": 0.3278, "step": 26880 }, { "epoch": 4.22, "learning_rate": 3.1278431372549023e-06, "loss": 0.2629, "step": 26890 }, { "epoch": 4.22, "learning_rate": 3.1215686274509804e-06, "loss": 0.2841, "step": 26900 }, { "epoch": 4.22, "learning_rate": 3.1152941176470593e-06, "loss": 0.1682, "step": 26910 }, { "epoch": 4.22, "learning_rate": 3.1090196078431373e-06, "loss": 0.1811, "step": 26920 }, { "epoch": 4.22, "learning_rate": 3.102745098039216e-06, "loss": 0.2413, "step": 26930 }, { "epoch": 4.23, "learning_rate": 3.0964705882352943e-06, "loss": 0.1469, "step": 26940 }, { "epoch": 4.23, "learning_rate": 3.0901960784313727e-06, "loss": 0.343, "step": 26950 }, { "epoch": 4.23, "learning_rate": 3.083921568627451e-06, "loss": 0.4836, "step": 26960 }, { "epoch": 4.23, "learning_rate": 3.0776470588235297e-06, "loss": 0.3213, "step": 26970 }, { "epoch": 4.23, "learning_rate": 3.071372549019608e-06, "loss": 0.2812, "step": 26980 }, { "epoch": 4.23, "learning_rate": 3.0650980392156866e-06, "loss": 0.4616, "step": 26990 }, { "epoch": 4.24, "learning_rate": 3.058823529411765e-06, "loss": 0.2088, "step": 27000 }, { "epoch": 4.24, "learning_rate": 3.052549019607843e-06, "loss": 0.2382, "step": 27010 }, { "epoch": 4.24, "learning_rate": 3.046274509803922e-06, "loss": 0.2086, "step": 27020 }, { "epoch": 4.24, "learning_rate": 3.04e-06, "loss": 0.3068, "step": 27030 }, { "epoch": 4.24, "learning_rate": 3.033725490196079e-06, "loss": 0.3676, "step": 27040 }, { "epoch": 4.24, "learning_rate": 3.027450980392157e-06, "loss": 0.2736, "step": 27050 }, { "epoch": 4.24, "learning_rate": 3.0211764705882355e-06, "loss": 0.4387, "step": 27060 }, { "epoch": 4.25, "learning_rate": 3.014901960784314e-06, "loss": 0.2978, "step": 27070 }, { "epoch": 4.25, "learning_rate": 3.0086274509803925e-06, "loss": 0.3083, "step": 27080 }, { "epoch": 4.25, "learning_rate": 3.0023529411764705e-06, "loss": 0.2016, "step": 27090 }, { "epoch": 4.25, "learning_rate": 2.9960784313725494e-06, "loss": 0.2815, "step": 27100 }, { "epoch": 4.25, "learning_rate": 2.9898039215686275e-06, "loss": 0.4741, "step": 27110 }, { "epoch": 4.25, "learning_rate": 2.9835294117647064e-06, "loss": 0.1623, "step": 27120 }, { "epoch": 4.26, "learning_rate": 2.9772549019607844e-06, "loss": 0.6056, "step": 27130 }, { "epoch": 4.26, "learning_rate": 2.970980392156863e-06, "loss": 0.389, "step": 27140 }, { "epoch": 4.26, "learning_rate": 2.9647058823529414e-06, "loss": 0.3245, "step": 27150 }, { "epoch": 4.26, "learning_rate": 2.95843137254902e-06, "loss": 0.2663, "step": 27160 }, { "epoch": 4.26, "learning_rate": 2.952156862745098e-06, "loss": 0.4298, "step": 27170 }, { "epoch": 4.26, "learning_rate": 2.945882352941177e-06, "loss": 0.3548, "step": 27180 }, { "epoch": 4.27, "learning_rate": 2.939607843137255e-06, "loss": 0.4985, "step": 27190 }, { "epoch": 4.27, "learning_rate": 2.9333333333333338e-06, "loss": 0.1604, "step": 27200 }, { "epoch": 4.27, "learning_rate": 2.927058823529412e-06, "loss": 0.3099, "step": 27210 }, { "epoch": 4.27, "learning_rate": 2.9207843137254903e-06, "loss": 0.3697, "step": 27220 }, { "epoch": 4.27, "learning_rate": 2.914509803921569e-06, "loss": 0.369, "step": 27230 }, { "epoch": 4.27, "learning_rate": 2.9082352941176472e-06, "loss": 0.2633, "step": 27240 }, { "epoch": 4.27, "learning_rate": 2.901960784313726e-06, "loss": 0.4083, "step": 27250 }, { "epoch": 4.28, "learning_rate": 2.895686274509804e-06, "loss": 0.3777, "step": 27260 }, { "epoch": 4.28, "learning_rate": 2.8894117647058827e-06, "loss": 0.2136, "step": 27270 }, { "epoch": 4.28, "learning_rate": 2.883137254901961e-06, "loss": 0.2263, "step": 27280 }, { "epoch": 4.28, "learning_rate": 2.8768627450980396e-06, "loss": 0.327, "step": 27290 }, { "epoch": 4.28, "learning_rate": 2.8705882352941177e-06, "loss": 0.2035, "step": 27300 }, { "epoch": 4.28, "learning_rate": 2.8643137254901966e-06, "loss": 0.3547, "step": 27310 }, { "epoch": 4.29, "learning_rate": 2.8580392156862746e-06, "loss": 0.298, "step": 27320 }, { "epoch": 4.29, "learning_rate": 2.8517647058823535e-06, "loss": 0.3237, "step": 27330 }, { "epoch": 4.29, "learning_rate": 2.8454901960784316e-06, "loss": 0.2628, "step": 27340 }, { "epoch": 4.29, "learning_rate": 2.83921568627451e-06, "loss": 0.2851, "step": 27350 }, { "epoch": 4.29, "learning_rate": 2.8329411764705885e-06, "loss": 0.1729, "step": 27360 }, { "epoch": 4.29, "learning_rate": 2.826666666666667e-06, "loss": 0.4959, "step": 27370 }, { "epoch": 4.29, "learning_rate": 2.820392156862745e-06, "loss": 0.1145, "step": 27380 }, { "epoch": 4.3, "learning_rate": 2.814117647058824e-06, "loss": 0.3599, "step": 27390 }, { "epoch": 4.3, "learning_rate": 2.807843137254902e-06, "loss": 0.4116, "step": 27400 }, { "epoch": 4.3, "learning_rate": 2.801568627450981e-06, "loss": 0.2756, "step": 27410 }, { "epoch": 4.3, "learning_rate": 2.795294117647059e-06, "loss": 0.2749, "step": 27420 }, { "epoch": 4.3, "learning_rate": 2.7890196078431374e-06, "loss": 0.5884, "step": 27430 }, { "epoch": 4.3, "learning_rate": 2.7827450980392155e-06, "loss": 0.2567, "step": 27440 }, { "epoch": 4.31, "learning_rate": 2.7764705882352944e-06, "loss": 0.3105, "step": 27450 }, { "epoch": 4.31, "learning_rate": 2.7701960784313724e-06, "loss": 0.2971, "step": 27460 }, { "epoch": 4.31, "learning_rate": 2.7639215686274513e-06, "loss": 0.205, "step": 27470 }, { "epoch": 4.31, "learning_rate": 2.75764705882353e-06, "loss": 0.1881, "step": 27480 }, { "epoch": 4.31, "learning_rate": 2.751372549019608e-06, "loss": 0.3286, "step": 27490 }, { "epoch": 4.31, "learning_rate": 2.7450980392156867e-06, "loss": 0.3679, "step": 27500 }, { "epoch": 4.32, "learning_rate": 2.738823529411765e-06, "loss": 0.2481, "step": 27510 }, { "epoch": 4.32, "learning_rate": 2.7325490196078437e-06, "loss": 0.2636, "step": 27520 }, { "epoch": 4.32, "learning_rate": 2.7262745098039217e-06, "loss": 0.1417, "step": 27530 }, { "epoch": 4.32, "learning_rate": 2.7200000000000002e-06, "loss": 0.2432, "step": 27540 }, { "epoch": 4.32, "learning_rate": 2.7137254901960787e-06, "loss": 0.2336, "step": 27550 }, { "epoch": 4.32, "learning_rate": 2.707450980392157e-06, "loss": 0.2053, "step": 27560 }, { "epoch": 4.32, "learning_rate": 2.7011764705882352e-06, "loss": 0.3232, "step": 27570 }, { "epoch": 4.33, "learning_rate": 2.694901960784314e-06, "loss": 0.4829, "step": 27580 }, { "epoch": 4.33, "learning_rate": 2.688627450980392e-06, "loss": 0.1949, "step": 27590 }, { "epoch": 4.33, "learning_rate": 2.682352941176471e-06, "loss": 0.2907, "step": 27600 }, { "epoch": 4.33, "learning_rate": 2.676078431372549e-06, "loss": 0.1293, "step": 27610 }, { "epoch": 4.33, "learning_rate": 2.6698039215686276e-06, "loss": 0.2764, "step": 27620 }, { "epoch": 4.33, "learning_rate": 2.663529411764706e-06, "loss": 0.2722, "step": 27630 }, { "epoch": 4.34, "learning_rate": 2.6572549019607845e-06, "loss": 0.1254, "step": 27640 }, { "epoch": 4.34, "learning_rate": 2.6509803921568626e-06, "loss": 0.2593, "step": 27650 }, { "epoch": 4.34, "learning_rate": 2.6447058823529415e-06, "loss": 0.5245, "step": 27660 }, { "epoch": 4.34, "learning_rate": 2.6384313725490195e-06, "loss": 0.2127, "step": 27670 }, { "epoch": 4.34, "learning_rate": 2.6321568627450984e-06, "loss": 0.2245, "step": 27680 }, { "epoch": 4.34, "learning_rate": 2.6258823529411765e-06, "loss": 0.2426, "step": 27690 }, { "epoch": 4.35, "learning_rate": 2.619607843137255e-06, "loss": 0.4143, "step": 27700 }, { "epoch": 4.35, "learning_rate": 2.6133333333333334e-06, "loss": 0.3419, "step": 27710 }, { "epoch": 4.35, "learning_rate": 2.607058823529412e-06, "loss": 0.1768, "step": 27720 }, { "epoch": 4.35, "learning_rate": 2.600784313725491e-06, "loss": 0.2391, "step": 27730 }, { "epoch": 4.35, "learning_rate": 2.594509803921569e-06, "loss": 0.3221, "step": 27740 }, { "epoch": 4.35, "learning_rate": 2.5882352941176473e-06, "loss": 0.6179, "step": 27750 }, { "epoch": 4.35, "learning_rate": 2.581960784313726e-06, "loss": 0.4007, "step": 27760 }, { "epoch": 4.36, "learning_rate": 2.5756862745098043e-06, "loss": 0.3001, "step": 27770 }, { "epoch": 4.36, "learning_rate": 2.5694117647058823e-06, "loss": 0.3147, "step": 27780 }, { "epoch": 4.36, "learning_rate": 2.5631372549019612e-06, "loss": 0.2706, "step": 27790 }, { "epoch": 4.36, "learning_rate": 2.5568627450980393e-06, "loss": 0.4256, "step": 27800 }, { "epoch": 4.36, "learning_rate": 2.550588235294118e-06, "loss": 0.305, "step": 27810 }, { "epoch": 4.36, "learning_rate": 2.5443137254901962e-06, "loss": 0.2934, "step": 27820 }, { "epoch": 4.37, "learning_rate": 2.5380392156862747e-06, "loss": 0.3118, "step": 27830 }, { "epoch": 4.37, "learning_rate": 2.531764705882353e-06, "loss": 0.4471, "step": 27840 }, { "epoch": 4.37, "learning_rate": 2.5254901960784317e-06, "loss": 0.3244, "step": 27850 }, { "epoch": 4.37, "learning_rate": 2.5192156862745097e-06, "loss": 0.2607, "step": 27860 }, { "epoch": 4.37, "learning_rate": 2.5129411764705886e-06, "loss": 0.237, "step": 27870 }, { "epoch": 4.37, "learning_rate": 2.5066666666666667e-06, "loss": 0.3304, "step": 27880 }, { "epoch": 4.37, "learning_rate": 2.5003921568627456e-06, "loss": 0.2849, "step": 27890 }, { "epoch": 4.38, "learning_rate": 2.4941176470588236e-06, "loss": 0.3624, "step": 27900 }, { "epoch": 4.38, "learning_rate": 2.487843137254902e-06, "loss": 0.3358, "step": 27910 }, { "epoch": 4.38, "learning_rate": 2.4815686274509806e-06, "loss": 0.3436, "step": 27920 }, { "epoch": 4.38, "learning_rate": 2.475294117647059e-06, "loss": 0.2848, "step": 27930 }, { "epoch": 4.38, "learning_rate": 2.4690196078431375e-06, "loss": 0.3017, "step": 27940 }, { "epoch": 4.38, "learning_rate": 2.462745098039216e-06, "loss": 0.4136, "step": 27950 }, { "epoch": 4.39, "learning_rate": 2.4564705882352945e-06, "loss": 0.2735, "step": 27960 }, { "epoch": 4.39, "learning_rate": 2.450196078431373e-06, "loss": 0.3427, "step": 27970 }, { "epoch": 4.39, "learning_rate": 2.443921568627451e-06, "loss": 0.3158, "step": 27980 }, { "epoch": 4.39, "learning_rate": 2.4376470588235295e-06, "loss": 0.2665, "step": 27990 }, { "epoch": 4.39, "learning_rate": 2.431372549019608e-06, "loss": 0.1491, "step": 28000 }, { "epoch": 4.39, "learning_rate": 2.4250980392156864e-06, "loss": 0.2459, "step": 28010 }, { "epoch": 4.4, "learning_rate": 2.418823529411765e-06, "loss": 0.2218, "step": 28020 }, { "epoch": 4.4, "learning_rate": 2.4125490196078434e-06, "loss": 0.3521, "step": 28030 }, { "epoch": 4.4, "learning_rate": 2.406274509803922e-06, "loss": 0.3537, "step": 28040 }, { "epoch": 4.4, "learning_rate": 2.4000000000000003e-06, "loss": 0.2675, "step": 28050 }, { "epoch": 4.4, "learning_rate": 2.3937254901960784e-06, "loss": 0.3277, "step": 28060 }, { "epoch": 4.4, "learning_rate": 2.387450980392157e-06, "loss": 0.3629, "step": 28070 }, { "epoch": 4.4, "learning_rate": 2.3811764705882353e-06, "loss": 0.0832, "step": 28080 }, { "epoch": 4.41, "learning_rate": 2.374901960784314e-06, "loss": 0.3309, "step": 28090 }, { "epoch": 4.41, "learning_rate": 2.3686274509803927e-06, "loss": 0.3345, "step": 28100 }, { "epoch": 4.41, "learning_rate": 2.3623529411764707e-06, "loss": 0.2495, "step": 28110 }, { "epoch": 4.41, "learning_rate": 2.3560784313725492e-06, "loss": 0.3147, "step": 28120 }, { "epoch": 4.41, "learning_rate": 2.3498039215686277e-06, "loss": 0.5246, "step": 28130 }, { "epoch": 4.41, "learning_rate": 2.343529411764706e-06, "loss": 0.2784, "step": 28140 }, { "epoch": 4.42, "learning_rate": 2.3372549019607846e-06, "loss": 0.494, "step": 28150 }, { "epoch": 4.42, "learning_rate": 2.330980392156863e-06, "loss": 0.4655, "step": 28160 }, { "epoch": 4.42, "learning_rate": 2.3247058823529416e-06, "loss": 0.219, "step": 28170 }, { "epoch": 4.42, "learning_rate": 2.3184313725490196e-06, "loss": 0.2938, "step": 28180 }, { "epoch": 4.42, "learning_rate": 2.312156862745098e-06, "loss": 0.4779, "step": 28190 }, { "epoch": 4.42, "learning_rate": 2.3058823529411766e-06, "loss": 0.3364, "step": 28200 }, { "epoch": 4.43, "learning_rate": 2.299607843137255e-06, "loss": 0.3105, "step": 28210 }, { "epoch": 4.43, "learning_rate": 2.2933333333333335e-06, "loss": 0.3383, "step": 28220 }, { "epoch": 4.43, "learning_rate": 2.287058823529412e-06, "loss": 0.2243, "step": 28230 }, { "epoch": 4.43, "learning_rate": 2.2807843137254905e-06, "loss": 0.2249, "step": 28240 }, { "epoch": 4.43, "learning_rate": 2.274509803921569e-06, "loss": 0.2912, "step": 28250 }, { "epoch": 4.43, "learning_rate": 2.268235294117647e-06, "loss": 0.1974, "step": 28260 }, { "epoch": 4.43, "learning_rate": 2.2619607843137255e-06, "loss": 0.1958, "step": 28270 }, { "epoch": 4.44, "learning_rate": 2.255686274509804e-06, "loss": 0.2831, "step": 28280 }, { "epoch": 4.44, "learning_rate": 2.2494117647058824e-06, "loss": 0.3601, "step": 28290 }, { "epoch": 4.44, "learning_rate": 2.243137254901961e-06, "loss": 0.3637, "step": 28300 }, { "epoch": 4.44, "learning_rate": 2.2368627450980394e-06, "loss": 0.2367, "step": 28310 }, { "epoch": 4.44, "learning_rate": 2.230588235294118e-06, "loss": 0.3955, "step": 28320 }, { "epoch": 4.44, "learning_rate": 2.2243137254901963e-06, "loss": 0.4779, "step": 28330 }, { "epoch": 4.45, "learning_rate": 2.2180392156862744e-06, "loss": 0.2849, "step": 28340 }, { "epoch": 4.45, "learning_rate": 2.2117647058823533e-06, "loss": 0.3683, "step": 28350 }, { "epoch": 4.45, "learning_rate": 2.2054901960784318e-06, "loss": 0.235, "step": 28360 }, { "epoch": 4.45, "learning_rate": 2.1992156862745102e-06, "loss": 0.2864, "step": 28370 }, { "epoch": 4.45, "learning_rate": 2.1929411764705887e-06, "loss": 0.2268, "step": 28380 }, { "epoch": 4.45, "learning_rate": 2.1866666666666668e-06, "loss": 0.3957, "step": 28390 }, { "epoch": 4.45, "learning_rate": 2.1803921568627452e-06, "loss": 0.4301, "step": 28400 }, { "epoch": 4.46, "learning_rate": 2.1741176470588237e-06, "loss": 0.1981, "step": 28410 }, { "epoch": 4.46, "learning_rate": 2.167843137254902e-06, "loss": 0.364, "step": 28420 }, { "epoch": 4.46, "learning_rate": 2.1615686274509807e-06, "loss": 0.3526, "step": 28430 }, { "epoch": 4.46, "learning_rate": 2.155294117647059e-06, "loss": 0.1453, "step": 28440 }, { "epoch": 4.46, "learning_rate": 2.1490196078431376e-06, "loss": 0.4385, "step": 28450 }, { "epoch": 4.46, "learning_rate": 2.1427450980392157e-06, "loss": 0.2104, "step": 28460 }, { "epoch": 4.47, "learning_rate": 2.136470588235294e-06, "loss": 0.1577, "step": 28470 }, { "epoch": 4.47, "learning_rate": 2.1301960784313726e-06, "loss": 0.326, "step": 28480 }, { "epoch": 4.47, "learning_rate": 2.123921568627451e-06, "loss": 0.2163, "step": 28490 }, { "epoch": 4.47, "learning_rate": 2.1176470588235296e-06, "loss": 0.2472, "step": 28500 }, { "epoch": 4.47, "learning_rate": 2.111372549019608e-06, "loss": 0.4743, "step": 28510 }, { "epoch": 4.47, "learning_rate": 2.1050980392156865e-06, "loss": 0.1112, "step": 28520 }, { "epoch": 4.48, "learning_rate": 2.098823529411765e-06, "loss": 0.3006, "step": 28530 }, { "epoch": 4.48, "learning_rate": 2.092549019607843e-06, "loss": 0.1186, "step": 28540 }, { "epoch": 4.48, "learning_rate": 2.0862745098039215e-06, "loss": 0.1177, "step": 28550 }, { "epoch": 4.48, "learning_rate": 2.08e-06, "loss": 0.2044, "step": 28560 }, { "epoch": 4.48, "learning_rate": 2.0737254901960785e-06, "loss": 0.3716, "step": 28570 }, { "epoch": 4.48, "learning_rate": 2.067450980392157e-06, "loss": 0.3377, "step": 28580 }, { "epoch": 4.48, "learning_rate": 2.0611764705882354e-06, "loss": 0.3696, "step": 28590 }, { "epoch": 4.49, "learning_rate": 2.054901960784314e-06, "loss": 0.2307, "step": 28600 }, { "epoch": 4.49, "learning_rate": 2.0486274509803924e-06, "loss": 0.2457, "step": 28610 }, { "epoch": 4.49, "learning_rate": 2.042352941176471e-06, "loss": 0.1568, "step": 28620 }, { "epoch": 4.49, "learning_rate": 2.0360784313725493e-06, "loss": 0.172, "step": 28630 }, { "epoch": 4.49, "learning_rate": 2.029803921568628e-06, "loss": 0.3621, "step": 28640 }, { "epoch": 4.49, "learning_rate": 2.0235294117647063e-06, "loss": 0.2508, "step": 28650 }, { "epoch": 4.5, "learning_rate": 2.0172549019607847e-06, "loss": 0.3894, "step": 28660 }, { "epoch": 4.5, "learning_rate": 2.010980392156863e-06, "loss": 0.5151, "step": 28670 }, { "epoch": 4.5, "learning_rate": 2.0047058823529413e-06, "loss": 0.2422, "step": 28680 }, { "epoch": 4.5, "learning_rate": 1.9984313725490197e-06, "loss": 0.3541, "step": 28690 }, { "epoch": 4.5, "learning_rate": 1.9921568627450982e-06, "loss": 0.1317, "step": 28700 }, { "epoch": 4.5, "learning_rate": 1.9858823529411767e-06, "loss": 0.4524, "step": 28710 }, { "epoch": 4.51, "learning_rate": 1.979607843137255e-06, "loss": 0.2924, "step": 28720 }, { "epoch": 4.51, "learning_rate": 1.9733333333333336e-06, "loss": 0.2284, "step": 28730 }, { "epoch": 4.51, "learning_rate": 1.967058823529412e-06, "loss": 0.4344, "step": 28740 }, { "epoch": 4.51, "learning_rate": 1.96078431372549e-06, "loss": 0.3931, "step": 28750 }, { "epoch": 4.51, "learning_rate": 1.9545098039215686e-06, "loss": 0.3492, "step": 28760 }, { "epoch": 4.51, "learning_rate": 1.948235294117647e-06, "loss": 0.254, "step": 28770 }, { "epoch": 4.51, "learning_rate": 1.9419607843137256e-06, "loss": 0.2125, "step": 28780 }, { "epoch": 4.52, "learning_rate": 1.935686274509804e-06, "loss": 0.2925, "step": 28790 }, { "epoch": 4.52, "learning_rate": 1.9294117647058825e-06, "loss": 0.4822, "step": 28800 }, { "epoch": 4.52, "learning_rate": 1.923137254901961e-06, "loss": 0.1988, "step": 28810 }, { "epoch": 4.52, "learning_rate": 1.916862745098039e-06, "loss": 0.2173, "step": 28820 }, { "epoch": 4.52, "learning_rate": 1.9105882352941175e-06, "loss": 0.235, "step": 28830 }, { "epoch": 4.52, "learning_rate": 1.904313725490196e-06, "loss": 0.3781, "step": 28840 }, { "epoch": 4.53, "learning_rate": 1.8980392156862745e-06, "loss": 0.2724, "step": 28850 }, { "epoch": 4.53, "learning_rate": 1.8917647058823532e-06, "loss": 0.2694, "step": 28860 }, { "epoch": 4.53, "learning_rate": 1.8854901960784317e-06, "loss": 0.2738, "step": 28870 }, { "epoch": 4.53, "learning_rate": 1.8792156862745101e-06, "loss": 0.3664, "step": 28880 }, { "epoch": 4.53, "learning_rate": 1.8729411764705884e-06, "loss": 0.3726, "step": 28890 }, { "epoch": 4.53, "learning_rate": 1.8666666666666669e-06, "loss": 0.3786, "step": 28900 }, { "epoch": 4.53, "learning_rate": 1.8603921568627453e-06, "loss": 0.3109, "step": 28910 }, { "epoch": 4.54, "learning_rate": 1.8541176470588238e-06, "loss": 0.3287, "step": 28920 }, { "epoch": 4.54, "learning_rate": 1.847843137254902e-06, "loss": 0.3783, "step": 28930 }, { "epoch": 4.54, "learning_rate": 1.8415686274509806e-06, "loss": 0.1754, "step": 28940 }, { "epoch": 4.54, "learning_rate": 1.835294117647059e-06, "loss": 0.0731, "step": 28950 }, { "epoch": 4.54, "learning_rate": 1.8290196078431375e-06, "loss": 0.3421, "step": 28960 }, { "epoch": 4.54, "learning_rate": 1.8227450980392158e-06, "loss": 0.219, "step": 28970 }, { "epoch": 4.55, "learning_rate": 1.8164705882352942e-06, "loss": 0.3296, "step": 28980 }, { "epoch": 4.55, "learning_rate": 1.8101960784313727e-06, "loss": 0.4036, "step": 28990 }, { "epoch": 4.55, "learning_rate": 1.8039215686274512e-06, "loss": 0.4787, "step": 29000 }, { "epoch": 4.55, "learning_rate": 1.7976470588235295e-06, "loss": 0.1904, "step": 29010 }, { "epoch": 4.55, "learning_rate": 1.791372549019608e-06, "loss": 0.3395, "step": 29020 }, { "epoch": 4.55, "learning_rate": 1.7850980392156864e-06, "loss": 0.4661, "step": 29030 }, { "epoch": 4.56, "learning_rate": 1.7788235294117649e-06, "loss": 0.3325, "step": 29040 }, { "epoch": 4.56, "learning_rate": 1.7725490196078431e-06, "loss": 0.2867, "step": 29050 }, { "epoch": 4.56, "learning_rate": 1.7662745098039216e-06, "loss": 0.2827, "step": 29060 }, { "epoch": 4.56, "learning_rate": 1.76e-06, "loss": 0.3487, "step": 29070 }, { "epoch": 4.56, "learning_rate": 1.7537254901960786e-06, "loss": 0.208, "step": 29080 }, { "epoch": 4.56, "learning_rate": 1.7474509803921568e-06, "loss": 0.3774, "step": 29090 }, { "epoch": 4.56, "learning_rate": 1.7411764705882353e-06, "loss": 0.2773, "step": 29100 }, { "epoch": 4.57, "learning_rate": 1.734901960784314e-06, "loss": 0.3384, "step": 29110 }, { "epoch": 4.57, "learning_rate": 1.7286274509803925e-06, "loss": 0.1541, "step": 29120 }, { "epoch": 4.57, "learning_rate": 1.722352941176471e-06, "loss": 0.2588, "step": 29130 }, { "epoch": 4.57, "learning_rate": 1.7160784313725492e-06, "loss": 0.2718, "step": 29140 }, { "epoch": 4.57, "learning_rate": 1.7098039215686277e-06, "loss": 0.1671, "step": 29150 }, { "epoch": 4.57, "learning_rate": 1.7035294117647062e-06, "loss": 0.201, "step": 29160 }, { "epoch": 4.58, "learning_rate": 1.6972549019607844e-06, "loss": 0.2531, "step": 29170 }, { "epoch": 4.58, "learning_rate": 1.6909803921568629e-06, "loss": 0.3498, "step": 29180 }, { "epoch": 4.58, "learning_rate": 1.6847058823529414e-06, "loss": 0.3486, "step": 29190 }, { "epoch": 4.58, "learning_rate": 1.6784313725490198e-06, "loss": 0.467, "step": 29200 }, { "epoch": 4.58, "learning_rate": 1.672156862745098e-06, "loss": 0.1887, "step": 29210 }, { "epoch": 4.58, "learning_rate": 1.6658823529411766e-06, "loss": 0.2832, "step": 29220 }, { "epoch": 4.59, "learning_rate": 1.659607843137255e-06, "loss": 0.2831, "step": 29230 }, { "epoch": 4.59, "learning_rate": 1.6533333333333335e-06, "loss": 0.301, "step": 29240 }, { "epoch": 4.59, "learning_rate": 1.6470588235294118e-06, "loss": 0.3519, "step": 29250 }, { "epoch": 4.59, "learning_rate": 1.6407843137254903e-06, "loss": 0.3303, "step": 29260 }, { "epoch": 4.59, "learning_rate": 1.6345098039215687e-06, "loss": 0.2338, "step": 29270 }, { "epoch": 4.59, "learning_rate": 1.6282352941176472e-06, "loss": 0.2349, "step": 29280 }, { "epoch": 4.59, "learning_rate": 1.6219607843137255e-06, "loss": 0.2735, "step": 29290 }, { "epoch": 4.6, "learning_rate": 1.615686274509804e-06, "loss": 0.3238, "step": 29300 }, { "epoch": 4.6, "learning_rate": 1.6094117647058824e-06, "loss": 0.3793, "step": 29310 }, { "epoch": 4.6, "learning_rate": 1.603137254901961e-06, "loss": 0.3902, "step": 29320 }, { "epoch": 4.6, "learning_rate": 1.5968627450980392e-06, "loss": 0.3559, "step": 29330 }, { "epoch": 4.6, "learning_rate": 1.5905882352941176e-06, "loss": 0.412, "step": 29340 }, { "epoch": 4.6, "learning_rate": 1.5843137254901961e-06, "loss": 0.3254, "step": 29350 }, { "epoch": 4.61, "learning_rate": 1.5780392156862748e-06, "loss": 0.3644, "step": 29360 }, { "epoch": 4.61, "learning_rate": 1.5717647058823533e-06, "loss": 0.3777, "step": 29370 }, { "epoch": 4.61, "learning_rate": 1.5654901960784315e-06, "loss": 0.2346, "step": 29380 }, { "epoch": 4.61, "learning_rate": 1.55921568627451e-06, "loss": 0.2756, "step": 29390 }, { "epoch": 4.61, "learning_rate": 1.5529411764705885e-06, "loss": 0.3786, "step": 29400 }, { "epoch": 4.61, "learning_rate": 1.546666666666667e-06, "loss": 0.4337, "step": 29410 }, { "epoch": 4.61, "learning_rate": 1.5403921568627452e-06, "loss": 0.1156, "step": 29420 }, { "epoch": 4.62, "learning_rate": 1.5341176470588237e-06, "loss": 0.3356, "step": 29430 }, { "epoch": 4.62, "learning_rate": 1.5278431372549022e-06, "loss": 0.28, "step": 29440 }, { "epoch": 4.62, "learning_rate": 1.5215686274509807e-06, "loss": 0.3751, "step": 29450 }, { "epoch": 4.62, "learning_rate": 1.515294117647059e-06, "loss": 0.1531, "step": 29460 }, { "epoch": 4.62, "learning_rate": 1.5090196078431374e-06, "loss": 0.2649, "step": 29470 }, { "epoch": 4.62, "learning_rate": 1.5027450980392159e-06, "loss": 0.1809, "step": 29480 }, { "epoch": 4.63, "learning_rate": 1.4964705882352941e-06, "loss": 0.3974, "step": 29490 }, { "epoch": 4.63, "learning_rate": 1.4901960784313726e-06, "loss": 0.2815, "step": 29500 }, { "epoch": 4.63, "learning_rate": 1.483921568627451e-06, "loss": 0.3315, "step": 29510 }, { "epoch": 4.63, "learning_rate": 1.4776470588235296e-06, "loss": 0.169, "step": 29520 }, { "epoch": 4.63, "learning_rate": 1.4713725490196078e-06, "loss": 0.3053, "step": 29530 }, { "epoch": 4.63, "learning_rate": 1.4650980392156863e-06, "loss": 0.1855, "step": 29540 }, { "epoch": 4.64, "learning_rate": 1.4588235294117648e-06, "loss": 0.2555, "step": 29550 }, { "epoch": 4.64, "learning_rate": 1.4525490196078432e-06, "loss": 0.2113, "step": 29560 }, { "epoch": 4.64, "learning_rate": 1.4462745098039215e-06, "loss": 0.2061, "step": 29570 }, { "epoch": 4.64, "learning_rate": 1.44e-06, "loss": 0.2099, "step": 29580 }, { "epoch": 4.64, "learning_rate": 1.4337254901960785e-06, "loss": 0.1734, "step": 29590 }, { "epoch": 4.64, "learning_rate": 1.427450980392157e-06, "loss": 0.2943, "step": 29600 }, { "epoch": 4.64, "learning_rate": 1.4211764705882356e-06, "loss": 0.3199, "step": 29610 }, { "epoch": 4.65, "learning_rate": 1.4149019607843139e-06, "loss": 0.3265, "step": 29620 }, { "epoch": 4.65, "learning_rate": 1.4086274509803924e-06, "loss": 0.3834, "step": 29630 }, { "epoch": 4.65, "learning_rate": 1.4023529411764708e-06, "loss": 0.2507, "step": 29640 }, { "epoch": 4.65, "learning_rate": 1.3960784313725493e-06, "loss": 0.362, "step": 29650 }, { "epoch": 4.65, "learning_rate": 1.3898039215686276e-06, "loss": 0.2565, "step": 29660 }, { "epoch": 4.65, "learning_rate": 1.383529411764706e-06, "loss": 0.455, "step": 29670 }, { "epoch": 4.66, "learning_rate": 1.3772549019607845e-06, "loss": 0.2247, "step": 29680 }, { "epoch": 4.66, "learning_rate": 1.370980392156863e-06, "loss": 0.426, "step": 29690 }, { "epoch": 4.66, "learning_rate": 1.3647058823529413e-06, "loss": 0.2118, "step": 29700 }, { "epoch": 4.66, "learning_rate": 1.3584313725490197e-06, "loss": 0.2269, "step": 29710 }, { "epoch": 4.66, "learning_rate": 1.3521568627450982e-06, "loss": 0.4097, "step": 29720 }, { "epoch": 4.66, "learning_rate": 1.3458823529411767e-06, "loss": 0.2901, "step": 29730 }, { "epoch": 4.67, "learning_rate": 1.339607843137255e-06, "loss": 0.274, "step": 29740 }, { "epoch": 4.67, "learning_rate": 1.3333333333333334e-06, "loss": 0.3403, "step": 29750 }, { "epoch": 4.67, "learning_rate": 1.3270588235294119e-06, "loss": 0.3182, "step": 29760 }, { "epoch": 4.67, "learning_rate": 1.3207843137254904e-06, "loss": 0.1559, "step": 29770 }, { "epoch": 4.67, "learning_rate": 1.3145098039215686e-06, "loss": 0.4059, "step": 29780 }, { "epoch": 4.67, "learning_rate": 1.308235294117647e-06, "loss": 0.2608, "step": 29790 }, { "epoch": 4.67, "learning_rate": 1.3019607843137256e-06, "loss": 0.3684, "step": 29800 }, { "epoch": 4.68, "learning_rate": 1.2956862745098038e-06, "loss": 0.2666, "step": 29810 }, { "epoch": 4.68, "learning_rate": 1.2894117647058823e-06, "loss": 0.2868, "step": 29820 }, { "epoch": 4.68, "learning_rate": 1.2831372549019608e-06, "loss": 0.2607, "step": 29830 }, { "epoch": 4.68, "learning_rate": 1.2768627450980393e-06, "loss": 0.2529, "step": 29840 }, { "epoch": 4.68, "learning_rate": 1.2705882352941175e-06, "loss": 0.3858, "step": 29850 }, { "epoch": 4.68, "learning_rate": 1.2643137254901962e-06, "loss": 0.2435, "step": 29860 }, { "epoch": 4.69, "learning_rate": 1.2580392156862747e-06, "loss": 0.2008, "step": 29870 }, { "epoch": 4.69, "learning_rate": 1.2517647058823532e-06, "loss": 0.3689, "step": 29880 }, { "epoch": 4.69, "learning_rate": 1.2454901960784314e-06, "loss": 0.3345, "step": 29890 }, { "epoch": 4.69, "learning_rate": 1.23921568627451e-06, "loss": 0.3677, "step": 29900 }, { "epoch": 4.69, "learning_rate": 1.2329411764705884e-06, "loss": 0.4659, "step": 29910 }, { "epoch": 4.69, "learning_rate": 1.2266666666666666e-06, "loss": 0.2071, "step": 29920 }, { "epoch": 4.69, "learning_rate": 1.2203921568627453e-06, "loss": 0.3619, "step": 29930 }, { "epoch": 4.7, "learning_rate": 1.2141176470588236e-06, "loss": 0.3971, "step": 29940 }, { "epoch": 4.7, "learning_rate": 1.207843137254902e-06, "loss": 0.3113, "step": 29950 }, { "epoch": 4.7, "learning_rate": 1.2015686274509805e-06, "loss": 0.3234, "step": 29960 }, { "epoch": 4.7, "learning_rate": 1.195294117647059e-06, "loss": 0.359, "step": 29970 }, { "epoch": 4.7, "learning_rate": 1.1890196078431373e-06, "loss": 0.3307, "step": 29980 }, { "epoch": 4.7, "learning_rate": 1.1827450980392158e-06, "loss": 0.2722, "step": 29990 }, { "epoch": 4.71, "learning_rate": 1.1764705882352942e-06, "loss": 0.3245, "step": 30000 }, { "epoch": 4.71, "learning_rate": 1.1701960784313727e-06, "loss": 0.2177, "step": 30010 }, { "epoch": 4.71, "learning_rate": 1.163921568627451e-06, "loss": 0.4123, "step": 30020 }, { "epoch": 4.71, "learning_rate": 1.1576470588235294e-06, "loss": 0.1948, "step": 30030 }, { "epoch": 4.71, "learning_rate": 1.151372549019608e-06, "loss": 0.1997, "step": 30040 }, { "epoch": 4.71, "learning_rate": 1.1450980392156864e-06, "loss": 0.1779, "step": 30050 }, { "epoch": 4.72, "learning_rate": 1.1388235294117649e-06, "loss": 0.2205, "step": 30060 }, { "epoch": 4.72, "learning_rate": 1.1325490196078433e-06, "loss": 0.3205, "step": 30070 }, { "epoch": 4.72, "learning_rate": 1.1262745098039216e-06, "loss": 0.2881, "step": 30080 }, { "epoch": 4.72, "learning_rate": 1.12e-06, "loss": 0.4199, "step": 30090 }, { "epoch": 4.72, "learning_rate": 1.1137254901960786e-06, "loss": 0.3522, "step": 30100 }, { "epoch": 4.72, "learning_rate": 1.107450980392157e-06, "loss": 0.3433, "step": 30110 }, { "epoch": 4.72, "learning_rate": 1.1011764705882353e-06, "loss": 0.1951, "step": 30120 }, { "epoch": 4.73, "learning_rate": 1.0949019607843138e-06, "loss": 0.3994, "step": 30130 }, { "epoch": 4.73, "learning_rate": 1.0886274509803922e-06, "loss": 0.4659, "step": 30140 }, { "epoch": 4.73, "learning_rate": 1.0823529411764707e-06, "loss": 0.1751, "step": 30150 }, { "epoch": 4.73, "learning_rate": 1.076078431372549e-06, "loss": 0.4366, "step": 30160 }, { "epoch": 4.73, "learning_rate": 1.0698039215686275e-06, "loss": 0.3946, "step": 30170 }, { "epoch": 4.73, "learning_rate": 1.0635294117647061e-06, "loss": 0.3067, "step": 30180 }, { "epoch": 4.74, "learning_rate": 1.0572549019607844e-06, "loss": 0.211, "step": 30190 }, { "epoch": 4.74, "learning_rate": 1.0509803921568629e-06, "loss": 0.3062, "step": 30200 }, { "epoch": 4.74, "learning_rate": 1.0447058823529414e-06, "loss": 0.161, "step": 30210 }, { "epoch": 4.74, "learning_rate": 1.0384313725490196e-06, "loss": 0.1295, "step": 30220 }, { "epoch": 4.74, "learning_rate": 1.032156862745098e-06, "loss": 0.3393, "step": 30230 }, { "epoch": 4.74, "learning_rate": 1.0258823529411766e-06, "loss": 0.3295, "step": 30240 }, { "epoch": 4.75, "learning_rate": 1.019607843137255e-06, "loss": 0.2343, "step": 30250 }, { "epoch": 4.75, "learning_rate": 1.0133333333333333e-06, "loss": 0.2837, "step": 30260 }, { "epoch": 4.75, "learning_rate": 1.0070588235294118e-06, "loss": 0.2491, "step": 30270 }, { "epoch": 4.75, "learning_rate": 1.0007843137254903e-06, "loss": 0.2958, "step": 30280 }, { "epoch": 4.75, "learning_rate": 9.945098039215687e-07, "loss": 0.3158, "step": 30290 }, { "epoch": 4.75, "learning_rate": 9.88235294117647e-07, "loss": 0.317, "step": 30300 }, { "epoch": 4.75, "learning_rate": 9.819607843137257e-07, "loss": 0.4471, "step": 30310 }, { "epoch": 4.76, "learning_rate": 9.756862745098042e-07, "loss": 0.225, "step": 30320 }, { "epoch": 4.76, "learning_rate": 9.694117647058824e-07, "loss": 0.2301, "step": 30330 }, { "epoch": 4.76, "learning_rate": 9.631372549019609e-07, "loss": 0.3364, "step": 30340 }, { "epoch": 4.76, "learning_rate": 9.568627450980394e-07, "loss": 0.3344, "step": 30350 }, { "epoch": 4.76, "learning_rate": 9.505882352941177e-07, "loss": 0.1931, "step": 30360 }, { "epoch": 4.76, "learning_rate": 9.443137254901961e-07, "loss": 0.4155, "step": 30370 }, { "epoch": 4.77, "learning_rate": 9.380392156862746e-07, "loss": 0.1514, "step": 30380 }, { "epoch": 4.77, "learning_rate": 9.317647058823529e-07, "loss": 0.1925, "step": 30390 }, { "epoch": 4.77, "learning_rate": 9.254901960784314e-07, "loss": 0.2315, "step": 30400 }, { "epoch": 4.77, "learning_rate": 9.192156862745098e-07, "loss": 0.5299, "step": 30410 }, { "epoch": 4.77, "learning_rate": 9.129411764705883e-07, "loss": 0.4844, "step": 30420 }, { "epoch": 4.77, "learning_rate": 9.066666666666668e-07, "loss": 0.1673, "step": 30430 }, { "epoch": 4.77, "learning_rate": 9.003921568627452e-07, "loss": 0.3329, "step": 30440 }, { "epoch": 4.78, "learning_rate": 8.941176470588237e-07, "loss": 0.2168, "step": 30450 }, { "epoch": 4.78, "learning_rate": 8.878431372549021e-07, "loss": 0.3626, "step": 30460 }, { "epoch": 4.78, "learning_rate": 8.815686274509805e-07, "loss": 0.2559, "step": 30470 }, { "epoch": 4.78, "learning_rate": 8.752941176470589e-07, "loss": 0.2243, "step": 30480 }, { "epoch": 4.78, "learning_rate": 8.690196078431374e-07, "loss": 0.2072, "step": 30490 }, { "epoch": 4.78, "learning_rate": 8.627450980392157e-07, "loss": 0.3356, "step": 30500 }, { "epoch": 4.79, "learning_rate": 8.564705882352941e-07, "loss": 0.2771, "step": 30510 }, { "epoch": 4.79, "learning_rate": 8.501960784313726e-07, "loss": 0.3827, "step": 30520 }, { "epoch": 4.79, "learning_rate": 8.43921568627451e-07, "loss": 0.4209, "step": 30530 }, { "epoch": 4.79, "learning_rate": 8.376470588235294e-07, "loss": 0.2533, "step": 30540 }, { "epoch": 4.79, "learning_rate": 8.313725490196078e-07, "loss": 0.1973, "step": 30550 }, { "epoch": 4.79, "learning_rate": 8.250980392156864e-07, "loss": 0.4049, "step": 30560 }, { "epoch": 4.8, "learning_rate": 8.188235294117649e-07, "loss": 0.2178, "step": 30570 }, { "epoch": 4.8, "learning_rate": 8.125490196078432e-07, "loss": 0.3775, "step": 30580 }, { "epoch": 4.8, "learning_rate": 8.062745098039217e-07, "loss": 0.3849, "step": 30590 }, { "epoch": 4.8, "learning_rate": 8.000000000000001e-07, "loss": 0.21, "step": 30600 }, { "epoch": 4.8, "learning_rate": 7.937254901960785e-07, "loss": 0.4502, "step": 30610 }, { "epoch": 4.8, "learning_rate": 7.874509803921569e-07, "loss": 0.2912, "step": 30620 }, { "epoch": 4.8, "learning_rate": 7.811764705882354e-07, "loss": 0.3612, "step": 30630 }, { "epoch": 4.81, "learning_rate": 7.749019607843138e-07, "loss": 0.2751, "step": 30640 }, { "epoch": 4.81, "learning_rate": 7.686274509803922e-07, "loss": 0.2618, "step": 30650 }, { "epoch": 4.81, "learning_rate": 7.623529411764706e-07, "loss": 0.368, "step": 30660 }, { "epoch": 4.81, "learning_rate": 7.56078431372549e-07, "loss": 0.2872, "step": 30670 }, { "epoch": 4.81, "learning_rate": 7.498039215686276e-07, "loss": 0.3772, "step": 30680 }, { "epoch": 4.81, "learning_rate": 7.43529411764706e-07, "loss": 0.1926, "step": 30690 }, { "epoch": 4.82, "learning_rate": 7.372549019607844e-07, "loss": 0.1626, "step": 30700 }, { "epoch": 4.82, "learning_rate": 7.309803921568629e-07, "loss": 0.5383, "step": 30710 }, { "epoch": 4.82, "learning_rate": 7.247058823529412e-07, "loss": 0.1338, "step": 30720 }, { "epoch": 4.82, "learning_rate": 7.184313725490197e-07, "loss": 0.389, "step": 30730 }, { "epoch": 4.82, "learning_rate": 7.121568627450981e-07, "loss": 0.1685, "step": 30740 }, { "epoch": 4.82, "learning_rate": 7.058823529411766e-07, "loss": 0.2696, "step": 30750 }, { "epoch": 4.83, "learning_rate": 6.996078431372549e-07, "loss": 0.3507, "step": 30760 }, { "epoch": 4.83, "learning_rate": 6.933333333333334e-07, "loss": 0.5086, "step": 30770 }, { "epoch": 4.83, "learning_rate": 6.870588235294118e-07, "loss": 0.43, "step": 30780 }, { "epoch": 4.83, "learning_rate": 6.807843137254902e-07, "loss": 0.2851, "step": 30790 }, { "epoch": 4.83, "learning_rate": 6.745098039215686e-07, "loss": 0.3227, "step": 30800 }, { "epoch": 4.83, "learning_rate": 6.682352941176472e-07, "loss": 0.1461, "step": 30810 }, { "epoch": 4.83, "learning_rate": 6.619607843137256e-07, "loss": 0.1121, "step": 30820 }, { "epoch": 4.84, "learning_rate": 6.55686274509804e-07, "loss": 0.2951, "step": 30830 }, { "epoch": 4.84, "learning_rate": 6.494117647058824e-07, "loss": 0.372, "step": 30840 }, { "epoch": 4.84, "learning_rate": 6.431372549019609e-07, "loss": 0.3411, "step": 30850 }, { "epoch": 4.84, "learning_rate": 6.368627450980393e-07, "loss": 0.2917, "step": 30860 }, { "epoch": 4.84, "learning_rate": 6.305882352941177e-07, "loss": 0.2638, "step": 30870 }, { "epoch": 4.84, "learning_rate": 6.243137254901961e-07, "loss": 0.1118, "step": 30880 }, { "epoch": 4.85, "learning_rate": 6.180392156862746e-07, "loss": 0.5434, "step": 30890 }, { "epoch": 4.85, "learning_rate": 6.11764705882353e-07, "loss": 0.3246, "step": 30900 }, { "epoch": 4.85, "learning_rate": 6.054901960784314e-07, "loss": 0.2603, "step": 30910 }, { "epoch": 4.85, "learning_rate": 5.992156862745099e-07, "loss": 0.1642, "step": 30920 }, { "epoch": 4.85, "learning_rate": 5.929411764705883e-07, "loss": 0.2747, "step": 30930 }, { "epoch": 4.85, "learning_rate": 5.866666666666667e-07, "loss": 0.4072, "step": 30940 }, { "epoch": 4.85, "learning_rate": 5.803921568627451e-07, "loss": 0.3326, "step": 30950 }, { "epoch": 4.86, "learning_rate": 5.741176470588236e-07, "loss": 0.3616, "step": 30960 }, { "epoch": 4.86, "learning_rate": 5.67843137254902e-07, "loss": 0.4393, "step": 30970 }, { "epoch": 4.86, "learning_rate": 5.615686274509804e-07, "loss": 0.3758, "step": 30980 }, { "epoch": 4.86, "learning_rate": 5.552941176470589e-07, "loss": 0.2784, "step": 30990 }, { "epoch": 4.86, "learning_rate": 5.490196078431373e-07, "loss": 0.2006, "step": 31000 }, { "epoch": 4.86, "learning_rate": 5.427450980392157e-07, "loss": 0.2581, "step": 31010 }, { "epoch": 4.87, "learning_rate": 5.364705882352941e-07, "loss": 0.2585, "step": 31020 }, { "epoch": 4.87, "learning_rate": 5.301960784313726e-07, "loss": 0.2546, "step": 31030 }, { "epoch": 4.87, "learning_rate": 5.239215686274511e-07, "loss": 0.2636, "step": 31040 }, { "epoch": 4.87, "learning_rate": 5.176470588235294e-07, "loss": 0.2011, "step": 31050 }, { "epoch": 4.87, "learning_rate": 5.113725490196079e-07, "loss": 0.5135, "step": 31060 }, { "epoch": 4.87, "learning_rate": 5.050980392156863e-07, "loss": 0.2147, "step": 31070 }, { "epoch": 4.88, "learning_rate": 4.988235294117647e-07, "loss": 0.1797, "step": 31080 }, { "epoch": 4.88, "learning_rate": 4.925490196078432e-07, "loss": 0.2201, "step": 31090 }, { "epoch": 4.88, "learning_rate": 4.862745098039216e-07, "loss": 0.3215, "step": 31100 }, { "epoch": 4.88, "learning_rate": 4.800000000000001e-07, "loss": 0.2302, "step": 31110 }, { "epoch": 4.88, "learning_rate": 4.737254901960785e-07, "loss": 0.3469, "step": 31120 }, { "epoch": 4.88, "learning_rate": 4.674509803921569e-07, "loss": 0.24, "step": 31130 }, { "epoch": 4.88, "learning_rate": 4.6117647058823533e-07, "loss": 0.2391, "step": 31140 }, { "epoch": 4.89, "learning_rate": 4.5490196078431375e-07, "loss": 0.188, "step": 31150 }, { "epoch": 4.89, "learning_rate": 4.486274509803922e-07, "loss": 0.3695, "step": 31160 }, { "epoch": 4.89, "learning_rate": 4.4235294117647065e-07, "loss": 0.2804, "step": 31170 }, { "epoch": 4.89, "learning_rate": 4.3607843137254907e-07, "loss": 0.3835, "step": 31180 }, { "epoch": 4.89, "learning_rate": 4.298039215686275e-07, "loss": 0.2645, "step": 31190 }, { "epoch": 4.89, "learning_rate": 4.235294117647059e-07, "loss": 0.1862, "step": 31200 }, { "epoch": 4.9, "learning_rate": 4.1725490196078433e-07, "loss": 0.2864, "step": 31210 }, { "epoch": 4.9, "learning_rate": 4.109803921568628e-07, "loss": 0.3496, "step": 31220 }, { "epoch": 4.9, "learning_rate": 4.0470588235294123e-07, "loss": 0.2768, "step": 31230 }, { "epoch": 4.9, "learning_rate": 3.9843137254901965e-07, "loss": 0.2957, "step": 31240 }, { "epoch": 4.9, "learning_rate": 3.921568627450981e-07, "loss": 0.3308, "step": 31250 }, { "epoch": 4.9, "learning_rate": 3.858823529411765e-07, "loss": 0.4046, "step": 31260 }, { "epoch": 4.91, "learning_rate": 3.796078431372549e-07, "loss": 0.3904, "step": 31270 }, { "epoch": 4.91, "learning_rate": 3.733333333333334e-07, "loss": 0.4637, "step": 31280 }, { "epoch": 4.91, "learning_rate": 3.670588235294118e-07, "loss": 0.4154, "step": 31290 }, { "epoch": 4.91, "learning_rate": 3.6078431372549024e-07, "loss": 0.1937, "step": 31300 }, { "epoch": 4.91, "learning_rate": 3.5450980392156866e-07, "loss": 0.476, "step": 31310 }, { "epoch": 4.91, "learning_rate": 3.482352941176471e-07, "loss": 0.2692, "step": 31320 }, { "epoch": 4.91, "learning_rate": 3.419607843137255e-07, "loss": 0.2386, "step": 31330 }, { "epoch": 4.92, "learning_rate": 3.35686274509804e-07, "loss": 0.2526, "step": 31340 }, { "epoch": 4.92, "learning_rate": 3.294117647058824e-07, "loss": 0.2648, "step": 31350 }, { "epoch": 4.92, "learning_rate": 3.231372549019608e-07, "loss": 0.2061, "step": 31360 }, { "epoch": 4.92, "learning_rate": 3.1686274509803924e-07, "loss": 0.2833, "step": 31370 }, { "epoch": 4.92, "learning_rate": 3.1058823529411767e-07, "loss": 0.2244, "step": 31380 }, { "epoch": 4.92, "learning_rate": 3.0431372549019614e-07, "loss": 0.244, "step": 31390 }, { "epoch": 4.93, "learning_rate": 2.9803921568627456e-07, "loss": 0.2236, "step": 31400 }, { "epoch": 4.93, "learning_rate": 2.9176470588235293e-07, "loss": 0.2667, "step": 31410 }, { "epoch": 4.93, "learning_rate": 2.854901960784314e-07, "loss": 0.3927, "step": 31420 }, { "epoch": 4.93, "learning_rate": 2.7921568627450983e-07, "loss": 0.1682, "step": 31430 }, { "epoch": 4.93, "learning_rate": 2.7294117647058825e-07, "loss": 0.5228, "step": 31440 }, { "epoch": 4.93, "learning_rate": 2.666666666666667e-07, "loss": 0.4514, "step": 31450 }, { "epoch": 4.93, "learning_rate": 2.6039215686274515e-07, "loss": 0.305, "step": 31460 }, { "epoch": 4.94, "learning_rate": 2.5411764705882357e-07, "loss": 0.1607, "step": 31470 }, { "epoch": 4.94, "learning_rate": 2.47843137254902e-07, "loss": 0.348, "step": 31480 }, { "epoch": 4.94, "learning_rate": 2.415686274509804e-07, "loss": 0.3534, "step": 31490 }, { "epoch": 4.94, "learning_rate": 2.3529411764705883e-07, "loss": 0.4916, "step": 31500 }, { "epoch": 4.94, "learning_rate": 2.2901960784313726e-07, "loss": 0.4352, "step": 31510 }, { "epoch": 4.94, "learning_rate": 2.227450980392157e-07, "loss": 0.187, "step": 31520 }, { "epoch": 4.95, "learning_rate": 2.1647058823529413e-07, "loss": 0.4045, "step": 31530 }, { "epoch": 4.95, "learning_rate": 2.1019607843137255e-07, "loss": 0.4347, "step": 31540 }, { "epoch": 4.95, "learning_rate": 2.03921568627451e-07, "loss": 0.185, "step": 31550 }, { "epoch": 4.95, "learning_rate": 1.9764705882352942e-07, "loss": 0.2212, "step": 31560 }, { "epoch": 4.95, "learning_rate": 1.9137254901960784e-07, "loss": 0.2741, "step": 31570 }, { "epoch": 4.95, "learning_rate": 1.850980392156863e-07, "loss": 0.3345, "step": 31580 }, { "epoch": 4.96, "learning_rate": 1.788235294117647e-07, "loss": 0.4033, "step": 31590 }, { "epoch": 4.96, "learning_rate": 1.7254901960784313e-07, "loss": 0.2918, "step": 31600 }, { "epoch": 4.96, "learning_rate": 1.662745098039216e-07, "loss": 0.2242, "step": 31610 }, { "epoch": 4.96, "learning_rate": 1.6e-07, "loss": 0.4202, "step": 31620 }, { "epoch": 4.96, "learning_rate": 1.5372549019607843e-07, "loss": 0.3196, "step": 31630 }, { "epoch": 4.96, "learning_rate": 1.4745098039215687e-07, "loss": 0.3565, "step": 31640 }, { "epoch": 4.96, "learning_rate": 1.4117647058823532e-07, "loss": 0.2188, "step": 31650 }, { "epoch": 4.97, "learning_rate": 1.3490196078431372e-07, "loss": 0.1255, "step": 31660 }, { "epoch": 4.97, "learning_rate": 1.2862745098039217e-07, "loss": 0.1861, "step": 31670 }, { "epoch": 4.97, "learning_rate": 1.2235294117647061e-07, "loss": 0.2371, "step": 31680 }, { "epoch": 4.97, "learning_rate": 1.1607843137254902e-07, "loss": 0.1907, "step": 31690 }, { "epoch": 4.97, "learning_rate": 1.0980392156862746e-07, "loss": 0.2327, "step": 31700 }, { "epoch": 4.97, "learning_rate": 1.0352941176470589e-07, "loss": 0.2482, "step": 31710 }, { "epoch": 4.98, "learning_rate": 9.725490196078431e-08, "loss": 0.2663, "step": 31720 }, { "epoch": 4.98, "learning_rate": 9.098039215686275e-08, "loss": 0.1831, "step": 31730 }, { "epoch": 4.98, "learning_rate": 8.470588235294117e-08, "loss": 0.2204, "step": 31740 }, { "epoch": 4.98, "learning_rate": 7.843137254901961e-08, "loss": 0.2438, "step": 31750 }, { "epoch": 4.98, "learning_rate": 7.215686274509804e-08, "loss": 0.3439, "step": 31760 }, { "epoch": 4.98, "learning_rate": 6.588235294117648e-08, "loss": 0.3954, "step": 31770 }, { "epoch": 4.99, "learning_rate": 5.960784313725491e-08, "loss": 0.2694, "step": 31780 }, { "epoch": 4.99, "learning_rate": 5.3333333333333334e-08, "loss": 0.2756, "step": 31790 }, { "epoch": 4.99, "learning_rate": 4.705882352941176e-08, "loss": 0.1508, "step": 31800 }, { "epoch": 4.99, "learning_rate": 4.0784313725490205e-08, "loss": 0.3217, "step": 31810 }, { "epoch": 4.99, "learning_rate": 3.450980392156863e-08, "loss": 0.2986, "step": 31820 }, { "epoch": 4.99, "learning_rate": 2.8235294117647062e-08, "loss": 0.3344, "step": 31830 }, { "epoch": 4.99, "learning_rate": 2.196078431372549e-08, "loss": 0.2008, "step": 31840 }, { "epoch": 5.0, "learning_rate": 1.5686274509803922e-08, "loss": 0.1681, "step": 31850 }, { "epoch": 5.0, "learning_rate": 9.411764705882354e-09, "loss": 0.2271, "step": 31860 }, { "epoch": 5.0, "learning_rate": 3.1372549019607845e-09, "loss": 0.2981, "step": 31870 }, { "epoch": 5.0, "eval_accuracy": 0.9948888888888889, "eval_loss": 0.024681543931365013, "eval_runtime": 182.6599, "eval_samples_per_second": 49.272, "eval_steps_per_second": 6.159, "step": 31875 }, { "epoch": 5.0, "step": 31875, "total_flos": 1.976187420555264e+19, "train_loss": 0.38248717208376115, "train_runtime": 17385.9511, "train_samples_per_second": 14.667, "train_steps_per_second": 1.833 } ], "logging_steps": 10, "max_steps": 31875, "num_input_tokens_seen": 0, "num_train_epochs": 5, "save_steps": 500, "total_flos": 1.976187420555264e+19, "trial_name": null, "trial_params": null }