/home/yuqian_fu here1 here2 {'gpu': '0', 'data': 'art_painting', 'ntr': None, 'translate': None, 'autoaug': 'CA_multiple', 'n': 3, 'stride': 5, 'factor_num': 16, 'epochs': 70, 'nbatch': 100, 'batchsize': 6, 'lr': 0.01, 'lr_scheduler': 'cosine', 'svroot': '/data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/saved-PACS//art_painting/CA_multiple_16fa_v2_ep70_lr0.01_cosine_base0.01_bs6_lamCa_1_lamRe1_adt4_cls1_EW2_70_rmTrue_rnTrue_str5_WithStyleAttackExp1', 'clsadapt': True, 'lambda_causal': 1.0, 'lambda_re': 1.0, 'randm': True, 'randn': True, 'network': 'resnet18'} stride: 5 /data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/data/PACS/art_painting_train.hdf5 torch.Size([1840, 3, 227, 227]) torch.Size([1840]) --------------------------CA_multiple-------------------------- ---------------------------16 factors----------------- randm: True randn: True n: 3 randm: False /data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/data/PACS/art_painting_val.hdf5 torch.Size([208, 3, 227, 227]) torch.Size([208]) -------------------------------------loading pretrain weights---------------------------------- Epoch 1, weight, value: tensor([[ 0.0106, -0.0051, 0.0193, ..., 0.0075, 0.0158, -0.0062], [-0.0130, -0.0094, -0.0199, ..., 0.0148, 0.0091, 0.0089], [ 0.0217, 0.0123, -0.0198, ..., -0.0208, 0.0086, -0.0179], ..., [-0.0085, -0.0153, 0.0125, ..., 0.0016, 0.0065, 0.0184], [ 0.0111, -0.0125, 0.0214, ..., -0.0080, 0.0129, -0.0186], [-0.0095, 0.0164, 0.0024, ..., -0.0037, 0.0123, 0.0207]], device='cuda:0'), grad: None Epoch 1, bias, value: tensor([ 0.0156, -0.0099, -0.0060, -0.0064, -0.0070, 0.0143, 0.0059], device='cuda:0'), grad: None 306 0.01 changing lr ---------------------saving model at epoch 0---------------------------------------------------- epoch 0, time 421.76, cls_loss 11.6044 cls_loss_mapping 1.8986 cls_loss_causal 1.9230 re_mapping 0.4058 re_causal 0.4051 /// teacc 43.27 lr 0.00999497 Epoch 2, weight, value: tensor([[ 0.1878, 0.1646, 0.2100, ..., -0.0196, 0.0020, -0.0184], [-0.0649, -0.0708, -0.0772, ..., 0.0933, 0.0843, 0.0784], [-0.0386, -0.0153, -0.0629, ..., 0.0813, 0.0574, 0.0432], ..., [-0.0308, -0.0403, -0.0028, ..., 0.0012, 0.0620, 0.0384], [ 0.0231, 0.0282, 0.0095, ..., -0.1716, -0.1842, -0.2046], [-0.0140, 0.0138, -0.0027, ..., -0.0242, -0.0221, 0.0152]], device='cuda:0'), grad: tensor([[-0.2456, -0.2186, -0.2211, ..., -0.0522, -0.0388, -0.0158], [-0.0894, -0.0504, -0.0532, ..., -0.0684, -0.0511, -0.0178], [ 0.0267, 0.0195, 0.0190, ..., 0.0144, 0.0101, 0.0034], ..., [-0.0347, -0.0211, -0.0188, ..., -0.0300, -0.0179, -0.0050], [ 0.0645, 0.0405, 0.0429, ..., 0.0440, 0.0329, 0.0115], [ 0.2252, 0.1888, 0.1958, ..., 0.0678, 0.0501, 0.0192]], device='cuda:0') Epoch 2, bias, value: tensor([-0.0260, 0.0482, -0.0075, -0.0372, 0.0386, -0.0156, 0.0055], device='cuda:0'), grad: tensor([-0.2336, -0.2844, 0.0637, 0.0666, -0.1208, 0.2062, 0.3025], device='cuda:0') 306 0.009994965332706574 changing lr epoch 1, time 424.69, cls_loss 2.0222 cls_loss_mapping 1.6048 cls_loss_causal 1.8042 re_mapping 0.1042 re_causal 0.1037 /// teacc 42.31 lr 0.00997987 Epoch 3, weight, value: tensor([[ 0.1727, 0.1568, 0.1972, ..., -0.0276, -0.0070, -0.0236], [-0.1169, -0.1227, -0.1246, ..., 0.0940, 0.0914, 0.0822], [-0.0560, -0.0360, -0.0763, ..., 0.0755, 0.0478, 0.0340], ..., [-0.0177, -0.0277, 0.0238, ..., -0.0073, 0.0621, 0.0368], [ 0.0245, 0.0343, 0.0103, ..., -0.1734, -0.1889, -0.2067], [ 0.0536, 0.0736, 0.0462, ..., -0.0229, -0.0228, 0.0129]], device='cuda:0'), grad: tensor([[ 4.2572e-02, 3.2379e-02, 3.3173e-02, ..., 1.3580e-02, 9.7656e-03, 4.5853e-03], [ 2.6875e-03, 1.4639e-03, 1.4391e-03, ..., 1.2617e-03, 9.5606e-04, 5.2929e-04], [ 1.7932e-01, 1.0278e-01, 1.1157e-01, ..., 6.3354e-02, 4.5898e-02, 2.2125e-02], ..., [-2.3206e-01, -1.4355e-01, -1.5369e-01, ..., -7.8003e-02, -5.5969e-02, -2.5665e-02], [ 7.9679e-04, 4.0364e-04, 3.8266e-04, ..., 3.4571e-04, 2.4199e-04, 1.1581e-04], [ 4.0680e-02, 2.0706e-02, 2.0401e-02, ..., 1.9073e-02, 1.4320e-02, 7.4272e-03]], device='cuda:0') Epoch 3, bias, value: tensor([-0.0599, 0.0352, -0.0025, -0.0539, 0.0827, -0.0219, 0.0262], device='cuda:0'), grad: tensor([ 0.0748, 0.0074, 0.2949, -0.1169, -0.3730, 0.0019, 0.1111], device='cuda:0') 306 0.009979871469976196 changing lr epoch 2, time 425.75, cls_loss 1.2357 cls_loss_mapping 1.2540 cls_loss_causal 1.6276 re_mapping 0.0906 re_causal 0.0899 /// teacc 37.50 lr 0.00995475 Epoch 4, weight, value: tensor([[ 0.1852, 0.1727, 0.2084, ..., -0.0183, -0.0016, -0.0191], [-0.1073, -0.1069, -0.1134, ..., 0.0933, 0.0911, 0.0836], [-0.0701, -0.0514, -0.0840, ..., 0.0735, 0.0460, 0.0309], ..., [-0.0694, -0.0779, -0.0236, ..., -0.0143, 0.0575, 0.0329], [ 0.0138, 0.0213, -0.0054, ..., -0.1757, -0.1922, -0.2086], [ 0.1033, 0.1200, 0.0955, ..., -0.0222, -0.0174, 0.0177]], device='cuda:0'), grad: tensor([[-2.4323e-02, -1.7715e-02, -1.8951e-02, ..., -3.5667e-03, -2.6474e-03, -1.3180e-03], [ 2.2745e-04, 1.7011e-04, 1.8191e-04, ..., 2.9743e-05, 2.2605e-05, 1.0028e-05], [ 2.1515e-03, 1.2884e-03, 1.3847e-03, ..., 5.4646e-04, 3.7622e-04, 2.5415e-04], ..., [ 1.3123e-02, 9.5825e-03, 1.0254e-02, ..., 1.9026e-03, 1.4153e-03, 6.9475e-04], [ 3.9124e-04, 3.5381e-04, 3.7313e-04, ..., 8.3372e-06, 1.3404e-05, 1.1101e-06], [ 6.6795e-03, 5.0278e-03, 5.3711e-03, ..., 8.4639e-04, 6.4564e-04, 2.7847e-04]], device='cuda:0') Epoch 4, bias, value: tensor([-0.0368, 0.0249, -0.0431, -0.0514, 0.0937, -0.0193, 0.0377], device='cuda:0'), grad: tensor([-0.0337, 0.0003, 0.0039, 0.0023, 0.0181, 0.0003, 0.0088], device='cuda:0') 306 0.009954748808839675 changing lr ---------------------saving model at epoch 3---------------------------------------------------- epoch 3, time 431.13, cls_loss 0.8660 cls_loss_mapping 0.9373 cls_loss_causal 1.4483 re_mapping 0.0863 re_causal 0.0854 /// teacc 48.56 lr 0.00991965 Epoch 5, weight, value: tensor([[ 0.1969, 0.1932, 0.2240, ..., -0.0106, 0.0083, -0.0084], [-0.1106, -0.1103, -0.1220, ..., 0.0942, 0.0883, 0.0799], [-0.0832, -0.0670, -0.0937, ..., 0.0688, 0.0420, 0.0268], ..., [-0.0814, -0.0917, -0.0321, ..., -0.0199, 0.0522, 0.0261], [ 0.0065, 0.0142, -0.0115, ..., -0.1784, -0.1959, -0.2123], [ 0.1163, 0.1315, 0.1102, ..., -0.0206, -0.0112, 0.0235]], device='cuda:0'), grad: tensor([[ 5.9433e-03, 2.8019e-03, 2.9926e-03, ..., 1.8187e-03, 1.3828e-03, 1.1148e-03], [-5.0812e-02, -2.6550e-02, -2.8870e-02, ..., -1.4725e-02, -1.0307e-02, -8.5678e-03], [ 4.2009e-04, 1.5187e-04, 1.5056e-04, ..., 1.4293e-04, 1.2743e-04, 9.7096e-05], ..., [ 5.1514e-02, 2.4612e-02, 2.6321e-02, ..., 1.5656e-02, 1.1826e-02, 9.5596e-03], [-1.1978e-02, -2.7523e-03, -2.3022e-03, ..., -4.5547e-03, -4.4975e-03, -3.3283e-03], [ 3.6449e-03, 1.3599e-03, 1.3704e-03, ..., 1.2245e-03, 1.0624e-03, 8.1873e-04]], device='cuda:0') Epoch 5, bias, value: tensor([-0.0341, 0.0254, -0.0705, -0.0229, 0.1050, -0.0339, 0.0366], device='cuda:0'), grad: tensor([ 0.0132, -0.0865, 0.0014, 0.0052, 0.1113, -0.0564, 0.0118], device='cuda:0') 306 0.009919647942993149 changing lr ---------------------saving model at epoch 4---------------------------------------------------- epoch 4, time 432.30, cls_loss 0.6247 cls_loss_mapping 0.7047 cls_loss_causal 1.2669 re_mapping 0.0851 re_causal 0.0840 /// teacc 72.12 lr 0.00987464 Epoch 6, weight, value: tensor([[ 0.1957, 0.1933, 0.2247, ..., -0.0115, 0.0068, -0.0090], [-0.1062, -0.1071, -0.1227, ..., 0.0857, 0.0783, 0.0707], [-0.0818, -0.0668, -0.0912, ..., 0.0748, 0.0483, 0.0343], ..., [-0.0890, -0.1005, -0.0373, ..., -0.0144, 0.0582, 0.0325], [-0.0086, 0.0016, -0.0259, ..., -0.1826, -0.1993, -0.2163], [ 0.1395, 0.1505, 0.1312, ..., -0.0195, -0.0080, 0.0255]], device='cuda:0'), grad: tensor([[ 0.0750, 0.0472, 0.0440, ..., 0.0164, 0.0148, 0.0158], [-0.0565, -0.0350, -0.0352, ..., -0.0148, -0.0146, -0.0162], [ 0.0156, 0.0057, 0.0061, ..., 0.0040, 0.0032, 0.0034], ..., [ 0.0461, 0.0297, 0.0238, ..., 0.0068, 0.0044, 0.0037], [-0.0116, -0.0024, -0.0030, ..., -0.0031, -0.0021, -0.0022], [-0.0692, -0.0453, -0.0359, ..., -0.0096, -0.0058, -0.0047]], device='cuda:0') Epoch 6, bias, value: tensor([-0.0279, 0.0292, -0.0618, -0.0331, 0.0900, -0.0454, 0.0549], device='cuda:0'), grad: tensor([ 0.1162, -0.0898, 0.0411, 0.0019, 0.0696, -0.0381, -0.1008], device='cuda:0') 306 0.009874639560909117 changing lr epoch 5, time 426.44, cls_loss 0.4380 cls_loss_mapping 0.5281 cls_loss_causal 1.2226 re_mapping 0.0861 re_causal 0.0850 /// teacc 42.31 lr 0.00981981 Epoch 7, weight, value: tensor([[ 0.2098, 0.2093, 0.2438, ..., -0.0112, 0.0055, -0.0106], [-0.0974, -0.0969, -0.1146, ..., 0.0915, 0.0849, 0.0769], [-0.0816, -0.0701, -0.0943, ..., 0.0685, 0.0410, 0.0278], ..., [-0.1026, -0.1140, -0.0502, ..., -0.0106, 0.0625, 0.0370], [-0.0079, 0.0012, -0.0266, ..., -0.1819, -0.1995, -0.2156], [ 0.1374, 0.1491, 0.1289, ..., -0.0233, -0.0101, 0.0223]], device='cuda:0'), grad: tensor([[-0.0735, -0.0327, -0.0339, ..., -0.0254, -0.0235, -0.0245], [ 0.0145, 0.0052, 0.0056, ..., 0.0056, 0.0052, 0.0054], [-0.0057, -0.0010, -0.0008, ..., -0.0023, -0.0022, -0.0022], ..., [ 0.0115, 0.0060, 0.0061, ..., 0.0034, 0.0032, 0.0033], [ 0.0025, 0.0004, 0.0004, ..., 0.0010, 0.0009, 0.0009], [ 0.0484, 0.0214, 0.0222, ..., 0.0167, 0.0155, 0.0161]], device='cuda:0') Epoch 7, bias, value: tensor([-0.0215, 0.0466, -0.0830, -0.0234, 0.0787, -0.0530, 0.0614], device='cuda:0'), grad: tensor([-0.1722, 0.0393, -0.0172, 0.0070, 0.0224, 0.0075, 0.1133], device='cuda:0') 306 0.009819814303479266 changing lr ---------------------saving model at epoch 6---------------------------------------------------- epoch 6, time 427.73, cls_loss 0.3658 cls_loss_mapping 0.4601 cls_loss_causal 1.1575 re_mapping 0.0819 re_causal 0.0808 /// teacc 78.85 lr 0.00975528 Epoch 8, weight, value: tensor([[ 0.2240, 0.2268, 0.2634, ..., -0.0166, 0.0008, -0.0141], [-0.0867, -0.0825, -0.1027, ..., 0.0913, 0.0859, 0.0786], [-0.0880, -0.0760, -0.0994, ..., 0.0664, 0.0399, 0.0277], ..., [-0.1128, -0.1233, -0.0609, ..., -0.0075, 0.0650, 0.0393], [-0.0067, 0.0008, -0.0251, ..., -0.1829, -0.1997, -0.2169], [ 0.1252, 0.1315, 0.1129, ..., -0.0217, -0.0112, 0.0210]], device='cuda:0'), grad: tensor([[-1.6870e-03, -1.3332e-03, -9.8896e-04, ..., -2.5082e-04, -2.2459e-04, -2.6464e-04], [ 2.4211e-04, 6.1810e-05, 4.8459e-05, ..., 1.4257e-04, 1.3053e-04, 1.4758e-04], [ 2.3880e-03, 1.4086e-03, 1.0519e-03, ..., 7.5197e-04, 6.8378e-04, 7.8249e-04], ..., [ 8.5688e-04, 1.7285e-04, 1.3876e-04, ..., 5.4407e-04, 4.9782e-04, 5.6219e-04], [-1.4257e-04, -3.1739e-05, -2.5257e-06, ..., -5.1796e-05, -5.4002e-05, -5.2422e-05], [ 4.9734e-04, 1.1754e-04, 9.1851e-05, ..., 3.0231e-04, 2.7633e-04, 3.1233e-04]], device='cuda:0') Epoch 8, bias, value: tensor([-0.0263, 0.0549, -0.0846, -0.0161, 0.0702, -0.0534, 0.0610], device='cuda:0'), grad: tensor([-0.0018, 0.0007, 0.0042, -0.0068, 0.0026, -0.0004, 0.0015], device='cuda:0') 306 0.009755282581475767 changing lr epoch 7, time 428.13, cls_loss 0.3124 cls_loss_mapping 0.3850 cls_loss_causal 1.0907 re_mapping 0.0829 re_causal 0.0819 /// teacc 77.88 lr 0.00968117 Epoch 9, weight, value: tensor([[ 0.2525, 0.2555, 0.2922, ..., -0.0049, 0.0127, -0.0029], [-0.0902, -0.0873, -0.1086, ..., 0.0879, 0.0829, 0.0759], [-0.0888, -0.0797, -0.1016, ..., 0.0671, 0.0397, 0.0283], ..., [-0.1185, -0.1264, -0.0642, ..., -0.0080, 0.0637, 0.0385], [-0.0113, -0.0028, -0.0307, ..., -0.1846, -0.2024, -0.2191], [ 0.1178, 0.1233, 0.1060, ..., -0.0286, -0.0175, 0.0142]], device='cuda:0'), grad: tensor([[ 2.7314e-05, -1.1764e-05, -1.4231e-05, ..., 2.3901e-05, 2.5332e-05, 2.7269e-05], [ 3.2067e-04, 1.0395e-04, 6.3479e-05, ..., 1.3578e-04, 1.5271e-04, 1.6749e-04], [-8.8024e-04, -2.0969e-04, -1.2898e-04, ..., -3.6907e-04, -3.9673e-04, -4.4155e-04], ..., [ 2.5272e-05, 4.2580e-06, 2.9281e-06, ..., 1.1168e-05, 1.1407e-05, 1.2547e-05], [ 5.6177e-05, 1.7658e-05, 1.0453e-05, ..., 2.3872e-05, 2.6867e-05, 2.9534e-05], [ 8.3685e-05, 2.6420e-05, 2.0012e-05, ..., 3.2485e-05, 3.4392e-05, 3.7998e-05]], device='cuda:0') Epoch 9, bias, value: tensor([-0.0317, 0.0606, -0.0707, -0.0167, 0.0594, -0.0630, 0.0678], device='cuda:0'), grad: tensor([ 1.6165e-04, 1.2770e-03, -3.2120e-03, 1.1988e-03, 8.2850e-05, 2.2674e-04, 2.6727e-04], device='cuda:0') 306 0.009681174353198686 changing lr epoch 8, time 424.95, cls_loss 0.2002 cls_loss_mapping 0.3516 cls_loss_causal 1.0047 re_mapping 0.0820 re_causal 0.0811 /// teacc 33.65 lr 0.00959764 Epoch 10, weight, value: tensor([[ 0.2614, 0.2600, 0.2987, ..., -0.0098, 0.0075, -0.0080], [-0.0923, -0.0862, -0.1080, ..., 0.0822, 0.0782, 0.0712], [-0.0955, -0.0838, -0.1060, ..., 0.0645, 0.0378, 0.0267], ..., [-0.1192, -0.1275, -0.0653, ..., -0.0048, 0.0652, 0.0409], [-0.0104, -0.0028, -0.0313, ..., -0.1847, -0.2024, -0.2193], [ 0.1168, 0.1226, 0.1041, ..., -0.0272, -0.0155, 0.0157]], device='cuda:0'), grad: tensor([[ 2.3193e-03, 1.1292e-03, 1.1568e-03, ..., 1.0481e-03, 1.2121e-03, 1.2636e-03], [ 2.1305e-03, 1.9920e-04, 2.6274e-04, ..., 1.1711e-03, 1.3361e-03, 1.3800e-03], [-5.6152e-03, -1.1911e-03, -1.2016e-03, ..., -2.9316e-03, -3.5038e-03, -3.5591e-03], ..., [-1.7195e-03, -1.0481e-03, -1.0672e-03, ..., -7.2527e-04, -8.3351e-04, -8.7643e-04], [-7.3552e-05, -1.5363e-05, -6.2399e-06, ..., -2.7969e-05, -3.7313e-05, -3.1382e-05], [ 4.4250e-04, 7.7963e-05, 8.1897e-05, ..., 2.3389e-04, 2.7561e-04, 2.8086e-04]], device='cuda:0') Epoch 10, bias, value: tensor([-0.0325, 0.0560, -0.0762, -0.0128, 0.0681, -0.0589, 0.0619], device='cuda:0'), grad: tensor([ 0.0045, 0.0079, -0.0192, 0.0078, -0.0023, -0.0003, 0.0016], device='cuda:0') 306 0.009597638862757255 changing lr ---------------------saving model at epoch 9---------------------------------------------------- epoch 9, time 428.66, cls_loss 0.1527 cls_loss_mapping 0.2936 cls_loss_causal 0.9805 re_mapping 0.0773 re_causal 0.0766 /// teacc 84.13 lr 0.00950484 Epoch 11, weight, value: tensor([[ 0.2543, 0.2539, 0.2933, ..., -0.0140, 0.0034, -0.0122], [-0.0843, -0.0777, -0.1006, ..., 0.0825, 0.0791, 0.0716], [-0.0936, -0.0842, -0.1065, ..., 0.0610, 0.0347, 0.0243], ..., [-0.1203, -0.1270, -0.0658, ..., -0.0010, 0.0682, 0.0443], [-0.0038, 0.0005, -0.0281, ..., -0.1811, -0.1990, -0.2157], [ 0.1156, 0.1199, 0.1026, ..., -0.0263, -0.0146, 0.0164]], device='cuda:0'), grad: tensor([[ 2.4765e-02, 6.7558e-03, 6.9962e-03, ..., 1.8234e-02, 1.8661e-02, 1.9272e-02], [ 4.1313e-03, 1.1272e-03, 1.1683e-03, ..., 3.0441e-03, 3.1166e-03, 3.2177e-03], [ 7.8964e-03, 2.1553e-03, 2.2316e-03, ..., 5.8098e-03, 5.9509e-03, 6.1417e-03], ..., [-3.7140e-02, -1.0139e-02, -1.0498e-02, ..., -2.7344e-02, -2.7985e-02, -2.8900e-02], [ 1.7428e-04, 4.7594e-05, 4.9263e-05, ..., 1.2815e-04, 1.3125e-04, 1.3554e-04], [ 8.5831e-05, 2.3320e-05, 2.4110e-05, ..., 6.3360e-05, 6.4850e-05, 6.6936e-05]], device='cuda:0') Epoch 11, bias, value: tensor([-0.0377, 0.0606, -0.0670, -0.0292, 0.0683, -0.0484, 0.0588], device='cuda:0'), grad: tensor([ 0.0754, 0.0126, 0.0241, 0.0003, -0.1132, 0.0005, 0.0003], device='cuda:0') 306 0.009504844339512096 changing lr epoch 10, time 432.92, cls_loss 0.1024 cls_loss_mapping 0.2664 cls_loss_causal 0.9671 re_mapping 0.0781 re_causal 0.0776 /// teacc 78.85 lr 0.00940298 Epoch 12, weight, value: tensor([[ 0.2556, 0.2549, 0.2939, ..., -0.0154, 0.0019, -0.0134], [-0.0869, -0.0772, -0.1012, ..., 0.0757, 0.0722, 0.0644], [-0.0960, -0.0876, -0.1101, ..., 0.0613, 0.0361, 0.0255], ..., [-0.1176, -0.1258, -0.0646, ..., 0.0041, 0.0723, 0.0490], [-0.0108, -0.0027, -0.0305, ..., -0.1790, -0.1969, -0.2136], [ 0.1190, 0.1213, 0.1047, ..., -0.0270, -0.0154, 0.0159]], device='cuda:0'), grad: tensor([[ 4.6939e-06, -6.5342e-06, -7.6517e-06, ..., 2.5146e-06, 3.9227e-06, 4.4741e-06], [ 2.5302e-05, 8.0988e-06, 5.8897e-06, ..., 4.1500e-06, 5.7928e-06, 6.3777e-06], [ 6.5279e-04, 9.4473e-05, 5.7399e-05, ..., 3.0136e-04, 3.3832e-04, 3.9124e-04], ..., [-3.4866e-03, -3.1996e-04, -1.4079e-04, ..., -1.8482e-03, -2.0466e-03, -2.3766e-03], [-6.5207e-05, -2.3693e-05, -1.9684e-05, ..., -6.7875e-06, -3.9041e-06, -1.0878e-06], [-1.3351e-04, -4.1932e-05, -2.8417e-05, ..., -1.0617e-05, -1.9982e-05, -2.1055e-05]], device='cuda:0') Epoch 12, bias, value: tensor([-0.0359, 0.0466, -0.0613, -0.0207, 0.0722, -0.0636, 0.0679], device='cuda:0'), grad: tensor([ 3.7521e-05, 6.2823e-05, 1.5831e-03, 7.4615e-03, -8.5754e-03, -2.2757e-04, -3.4595e-04], device='cuda:0') 306 0.009402977659283692 changing lr epoch 11, time 428.82, cls_loss 0.1045 cls_loss_mapping 0.2426 cls_loss_causal 0.9590 re_mapping 0.0746 re_causal 0.0742 /// teacc 73.08 lr 0.00929224 Epoch 13, weight, value: tensor([[ 0.2544, 0.2549, 0.2941, ..., -0.0189, -0.0027, -0.0172], [-0.0718, -0.0682, -0.0925, ..., 0.0745, 0.0723, 0.0643], [-0.0980, -0.0885, -0.1105, ..., 0.0600, 0.0352, 0.0247], ..., [-0.1210, -0.1289, -0.0685, ..., 0.0058, 0.0737, 0.0503], [-0.0109, -0.0042, -0.0327, ..., -0.1759, -0.1932, -0.2096], [ 0.1122, 0.1174, 0.1006, ..., -0.0270, -0.0160, 0.0152]], device='cuda:0'), grad: tensor([[-3.5858e-03, -1.9474e-03, -1.8644e-03, ..., -1.1988e-03, -1.4086e-03, -1.4277e-03], [ 3.4409e-03, 1.7090e-03, 1.6346e-03, ..., 1.1806e-03, 1.3924e-03, 1.4219e-03], [-1.8692e-03, -2.7037e-04, -2.5535e-04, ..., -6.8378e-04, -8.3160e-04, -9.0551e-04], ..., [ 3.8457e-04, 1.0127e-04, 9.6619e-05, ..., 1.2803e-04, 1.5509e-04, 1.6749e-04], [ 8.3780e-04, 1.4365e-04, 1.3626e-04, ..., 3.0255e-04, 3.6740e-04, 3.9887e-04], [ 1.8275e-04, 4.3809e-05, 4.1515e-05, ..., 6.1333e-05, 7.4029e-05, 8.0884e-05]], device='cuda:0') Epoch 13, bias, value: tensor([-0.0398, 0.0663, -0.0684, -0.0271, 0.0654, -0.0569, 0.0658], device='cuda:0'), grad: tensor([-0.0050, 0.0056, -0.0061, 0.0013, 0.0010, 0.0026, 0.0005], device='cuda:0') 306 0.009292243968009333 changing lr ---------------------saving model at epoch 12---------------------------------------------------- epoch 12, time 433.28, cls_loss 0.1435 cls_loss_mapping 0.2366 cls_loss_causal 0.9508 re_mapping 0.0761 re_causal 0.0760 /// teacc 84.62 lr 0.00917287 Epoch 14, weight, value: tensor([[ 0.2648, 0.2631, 0.3023, ..., -0.0203, -0.0039, -0.0183], [-0.0708, -0.0647, -0.0908, ..., 0.0738, 0.0713, 0.0633], [-0.1016, -0.0902, -0.1106, ..., 0.0580, 0.0337, 0.0228], ..., [-0.1377, -0.1389, -0.0794, ..., 0.0060, 0.0728, 0.0497], [-0.0093, -0.0050, -0.0338, ..., -0.1737, -0.1908, -0.2077], [ 0.1134, 0.1155, 0.1006, ..., -0.0270, -0.0161, 0.0155]], device='cuda:0'), grad: tensor([[-1.0452e-02, -4.9820e-03, -4.7569e-03, ..., -1.1320e-03, -1.4286e-03, -1.5659e-03], [ 9.8109e-05, 4.7505e-05, 4.5270e-05, ..., 1.0222e-05, 1.2979e-05, 1.4357e-05], [ 4.1217e-05, 1.9357e-05, 1.8522e-05, ..., 4.6566e-06, 5.8375e-06, 6.3404e-06], ..., [ 1.2827e-03, 6.1560e-04, 5.8699e-04, ..., 1.3649e-04, 1.7273e-04, 1.9014e-04], [ 6.7101e-03, 3.0994e-03, 2.9678e-03, ..., 7.8487e-04, 9.7942e-04, 1.0557e-03], [ 1.6441e-03, 8.8120e-04, 8.3113e-04, ..., 1.2231e-04, 1.6546e-04, 1.9848e-04]], device='cuda:0') Epoch 14, bias, value: tensor([-0.0348, 0.0573, -0.0797, -0.0097, 0.0495, -0.0497, 0.0722], device='cuda:0'), grad: tensor([-1.6800e-02, 1.5473e-04, 6.7532e-05, 1.0681e-03, 2.0447e-03, 1.1192e-02, 2.2488e-03], device='cuda:0') 306 0.009172866268606516 changing lr epoch 13, time 428.41, cls_loss 0.0724 cls_loss_mapping 0.2087 cls_loss_causal 0.9581 re_mapping 0.0757 re_causal 0.0758 /// teacc 82.21 lr 0.00904508 Epoch 15, weight, value: tensor([[ 0.2678, 0.2653, 0.3044, ..., -0.0203, -0.0042, -0.0183], [-0.0688, -0.0640, -0.0901, ..., 0.0728, 0.0709, 0.0626], [-0.1001, -0.0903, -0.1106, ..., 0.0604, 0.0366, 0.0265], ..., [-0.1351, -0.1369, -0.0781, ..., 0.0080, 0.0743, 0.0514], [-0.0108, -0.0065, -0.0349, ..., -0.1723, -0.1900, -0.2064], [ 0.1121, 0.1141, 0.0995, ..., -0.0277, -0.0169, 0.0143]], device='cuda:0'), grad: tensor([[ 1.1940e-02, 4.6654e-03, 3.6144e-03, ..., 3.5648e-03, 3.6526e-03, 4.1809e-03], [ 7.1526e-04, 2.5606e-04, 1.9670e-04, ..., 2.4796e-04, 2.5439e-04, 2.9278e-04], [ 3.0756e-04, 9.9123e-05, 7.3135e-05, ..., 1.2970e-04, 1.3614e-04, 1.5116e-04], ..., [-1.6800e-02, -6.4583e-03, -4.9934e-03, ..., -5.1765e-03, -5.3024e-03, -6.0768e-03], [ 3.9649e-04, 1.4973e-04, 1.1551e-04, ..., 1.2648e-04, 1.2982e-04, 1.4853e-04], [ 3.3360e-03, 1.2503e-03, 9.6607e-04, ..., 1.0710e-03, 1.0958e-03, 1.2627e-03]], device='cuda:0') Epoch 15, bias, value: tensor([-0.0334, 0.0568, -0.0657, -0.0270, 0.0542, -0.0515, 0.0717], device='cuda:0'), grad: tensor([ 0.0220, 0.0015, 0.0007, 0.0002, -0.0316, 0.0008, 0.0065], device='cuda:0') 306 0.00904508497187474 changing lr epoch 14, time 427.18, cls_loss 0.0752 cls_loss_mapping 0.2063 cls_loss_causal 0.9385 re_mapping 0.0745 re_causal 0.0746 /// teacc 77.88 lr 0.00890916 Epoch 16, weight, value: tensor([[ 0.2597, 0.2606, 0.2997, ..., -0.0224, -0.0069, -0.0210], [-0.0762, -0.0676, -0.0934, ..., 0.0694, 0.0677, 0.0592], [-0.0925, -0.0868, -0.1069, ..., 0.0608, 0.0379, 0.0282], ..., [-0.1296, -0.1333, -0.0759, ..., 0.0107, 0.0755, 0.0534], [-0.0116, -0.0077, -0.0362, ..., -0.1713, -0.1888, -0.2052], [ 0.1154, 0.1160, 0.1024, ..., -0.0267, -0.0156, 0.0152]], device='cuda:0'), grad: tensor([[-1.1276e-02, -7.5569e-03, -7.1335e-03, ..., -9.6512e-04, -1.2064e-03, -1.4009e-03], [ 9.4399e-06, 6.8583e-06, 6.3330e-06, ..., 1.8068e-06, 1.8124e-06, 1.7043e-06], [ 3.1776e-03, 1.9855e-03, 1.8578e-03, ..., 3.5858e-04, 4.3797e-04, 4.9114e-04], ..., [-5.6791e-04, -1.5104e-04, -1.1533e-04, ..., -1.8704e-04, -2.1851e-04, -2.2566e-04], [ 1.1760e-04, 4.9055e-05, 4.1932e-05, ..., 2.5496e-05, 3.0220e-05, 3.1352e-05], [ 8.4915e-03, 5.6458e-03, 5.3215e-03, ..., 7.5531e-04, 9.4271e-04, 1.0891e-03]], device='cuda:0') Epoch 16, bias, value: tensor([-0.0451, 0.0468, -0.0566, -0.0274, 0.0629, -0.0538, 0.0782], device='cuda:0'), grad: tensor([-1.2550e-02, 7.4729e-06, 4.0016e-03, 7.3373e-05, -1.3828e-03, 2.2531e-04, 9.6130e-03], device='cuda:0') 306 0.008909157412340152 changing lr epoch 15, time 429.27, cls_loss 0.0493 cls_loss_mapping 0.1644 cls_loss_causal 0.8087 re_mapping 0.0737 re_causal 0.0738 /// teacc 84.13 lr 0.00876536 Epoch 17, weight, value: tensor([[ 0.2622, 0.2626, 0.3013, ..., -0.0197, -0.0045, -0.0183], [-0.0737, -0.0665, -0.0920, ..., 0.0687, 0.0676, 0.0594], [-0.1002, -0.0888, -0.1083, ..., 0.0566, 0.0336, 0.0238], ..., [-0.1308, -0.1338, -0.0773, ..., 0.0105, 0.0745, 0.0524], [-0.0091, -0.0070, -0.0355, ..., -0.1685, -0.1859, -0.2023], [ 0.1094, 0.1124, 0.0990, ..., -0.0280, -0.0173, 0.0129]], device='cuda:0'), grad: tensor([[ 3.2349e-03, 9.7322e-04, 1.0643e-03, ..., 9.9182e-04, 1.1444e-03, 1.1063e-03], [ 1.0931e-04, 3.7700e-05, 2.9683e-05, ..., 3.7044e-05, 4.3064e-05, 4.2886e-05], [ 3.2806e-04, 1.1384e-04, 1.1349e-04, ..., 1.0669e-04, 1.2398e-04, 1.2791e-04], ..., [-9.7275e-03, -3.2368e-03, -3.2558e-03, ..., -3.1281e-03, -3.6716e-03, -3.7193e-03], [ 2.3975e-03, 7.9107e-04, 8.0633e-04, ..., 7.6532e-04, 8.9788e-04, 9.0694e-04], [ 4.2582e-04, 1.5390e-04, 1.4496e-04, ..., 1.4293e-04, 1.7035e-04, 1.7917e-04]], device='cuda:0') Epoch 17, bias, value: tensor([-0.0407, 0.0508, -0.0761, -0.0104, 0.0615, -0.0460, 0.0660], device='cuda:0'), grad: tensor([ 0.0076, 0.0002, 0.0007, 0.0070, -0.0219, 0.0054, 0.0009], device='cuda:0') 306 0.00876535733001806 changing lr epoch 16, time 430.23, cls_loss 0.0479 cls_loss_mapping 0.1632 cls_loss_causal 0.8588 re_mapping 0.0722 re_causal 0.0726 /// teacc 83.65 lr 0.00861397 Epoch 18, weight, value: tensor([[ 0.2592, 0.2595, 0.2979, ..., -0.0207, -0.0055, -0.0192], [-0.0728, -0.0663, -0.0919, ..., 0.0676, 0.0669, 0.0586], [-0.0975, -0.0859, -0.1046, ..., 0.0558, 0.0330, 0.0236], ..., [-0.1267, -0.1304, -0.0751, ..., 0.0126, 0.0759, 0.0540], [-0.0114, -0.0086, -0.0370, ..., -0.1673, -0.1846, -0.2010], [ 0.1103, 0.1118, 0.0986, ..., -0.0272, -0.0167, 0.0132]], device='cuda:0'), grad: tensor([[ 1.6737e-04, 5.5701e-05, 5.5939e-05, ..., 5.0575e-05, 5.4538e-05, 6.0648e-05], [ 4.0913e-04, 1.3793e-04, 1.3995e-04, ..., 1.1837e-04, 1.2434e-04, 1.3864e-04], [-6.3133e-03, -1.0328e-03, -1.0738e-03, ..., -2.9068e-03, -2.9488e-03, -3.3131e-03], ..., [ 5.3711e-03, 8.8596e-04, 9.2173e-04, ..., 2.4643e-03, 2.5005e-03, 2.8095e-03], [ 1.8072e-04, 4.8667e-05, 4.9621e-05, ..., 6.4075e-05, 6.6817e-05, 7.4565e-05], [-3.5214e-04, -2.0921e-04, -2.0885e-04, ..., -1.7747e-05, -3.2753e-05, -3.3498e-05]], device='cuda:0') Epoch 18, bias, value: tensor([-0.0402, 0.0505, -0.0748, -0.0158, 0.0623, -0.0484, 0.0714], device='cuda:0'), grad: tensor([ 0.0004, 0.0009, -0.0178, 0.0014, 0.0151, 0.0004, -0.0005], device='cuda:0') 306 0.008613974319136962 changing lr epoch 17, time 429.92, cls_loss 0.0243 cls_loss_mapping 0.1391 cls_loss_causal 0.8789 re_mapping 0.0726 re_causal 0.0731 /// teacc 83.17 lr 0.00845531 Epoch 19, weight, value: tensor([[ 0.2583, 0.2586, 0.2964, ..., -0.0196, -0.0048, -0.0185], [-0.0752, -0.0660, -0.0914, ..., 0.0652, 0.0644, 0.0560], [-0.0976, -0.0868, -0.1054, ..., 0.0534, 0.0304, 0.0213], ..., [-0.1255, -0.1298, -0.0747, ..., 0.0132, 0.0766, 0.0549], [-0.0113, -0.0088, -0.0373, ..., -0.1651, -0.1825, -0.1985], [ 0.1100, 0.1122, 0.0997, ..., -0.0276, -0.0173, 0.0123]], device='cuda:0'), grad: tensor([[ 1.8403e-06, -8.6240e-07, -1.1362e-06, ..., 1.7639e-06, 1.7975e-06, 2.0228e-06], [ 8.0653e-07, 2.6636e-07, 1.9558e-07, ..., 2.4587e-07, 2.5518e-07, 2.9616e-07], [ 1.5311e-06, 3.5390e-07, 1.6950e-07, ..., 6.5006e-07, 5.9977e-07, 7.3202e-07], ..., [-3.5584e-05, -6.4559e-06, -2.1216e-06, ..., -1.6674e-05, -1.6719e-05, -1.9029e-05], [-7.9721e-07, -1.6205e-07, -3.1292e-07, ..., 3.2410e-07, 2.8685e-07, 2.4214e-07], [ 9.9838e-07, 6.4634e-07, 5.4948e-07, ..., 1.3597e-07, 1.3970e-07, 1.5274e-07]], device='cuda:0') Epoch 19, bias, value: tensor([-0.0405, 0.0421, -0.0691, -0.0113, 0.0613, -0.0455, 0.0680], device='cuda:0'), grad: tensor([ 8.4639e-06, 1.9073e-06, 3.9414e-06, 8.4400e-05, -9.7275e-05, -2.9840e-06, 1.5069e-06], device='cuda:0') 306 0.008455313244934327 changing lr epoch 18, time 426.88, cls_loss 0.0267 cls_loss_mapping 0.1306 cls_loss_causal 0.8283 re_mapping 0.0700 re_causal 0.0706 /// teacc 84.62 lr 0.00828969 Epoch 20, weight, value: tensor([[ 0.2555, 0.2564, 0.2938, ..., -0.0193, -0.0046, -0.0181], [-0.0724, -0.0644, -0.0896, ..., 0.0648, 0.0644, 0.0561], [-0.0933, -0.0848, -0.1034, ..., 0.0538, 0.0313, 0.0223], ..., [-0.1238, -0.1286, -0.0743, ..., 0.0136, 0.0762, 0.0548], [-0.0132, -0.0105, -0.0388, ..., -0.1628, -0.1805, -0.1963], [ 0.1110, 0.1128, 0.1005, ..., -0.0283, -0.0180, 0.0112]], device='cuda:0'), grad: tensor([[ 2.0943e-03, 6.0558e-04, 6.5231e-04, ..., 1.0023e-03, 9.3126e-04, 1.0624e-03], [ 3.7360e-04, 1.0639e-04, 1.0651e-04, ..., 1.6189e-04, 1.5295e-04, 1.7846e-04], [ 3.8662e-03, 9.4938e-04, 7.3290e-04, ..., 1.3113e-03, 1.5354e-03, 1.6270e-03], ..., [-8.3771e-03, -2.2507e-03, -2.1477e-03, ..., -3.5439e-03, -3.6964e-03, -4.0207e-03], [ 4.0007e-04, 1.0943e-04, 1.0467e-04, ..., 1.6475e-04, 1.6594e-04, 1.8454e-04], [ 8.8573e-05, 3.6508e-05, 7.8022e-05, ..., 1.6296e-04, 1.9848e-04, 1.7130e-04]], device='cuda:0') Epoch 20, bias, value: tensor([-0.0415, 0.0456, -0.0612, -0.0222, 0.0607, -0.0479, 0.0713], device='cuda:0'), grad: tensor([ 0.0055, 0.0010, 0.0094, 0.0041, -0.0212, 0.0010, 0.0003], device='cuda:0') 306 0.008289693629698565 changing lr ---------------------saving model at epoch 19---------------------------------------------------- epoch 19, time 432.12, cls_loss 0.0267 cls_loss_mapping 0.1288 cls_loss_causal 0.8148 re_mapping 0.0707 re_causal 0.0716 /// teacc 86.06 lr 0.00811745 Epoch 21, weight, value: tensor([[ 0.2600, 0.2587, 0.2960, ..., -0.0177, -0.0030, -0.0163], [-0.0687, -0.0632, -0.0889, ..., 0.0651, 0.0654, 0.0569], [-0.0958, -0.0865, -0.1049, ..., 0.0533, 0.0306, 0.0220], ..., [-0.1248, -0.1282, -0.0745, ..., 0.0130, 0.0747, 0.0536], [-0.0146, -0.0111, -0.0389, ..., -0.1617, -0.1793, -0.1949], [ 0.1092, 0.1118, 0.0995, ..., -0.0295, -0.0196, 0.0092]], device='cuda:0'), grad: tensor([[ 1.8910e-05, 6.5379e-06, 4.4592e-06, ..., 6.7987e-06, 7.1824e-06, 7.8008e-06], [-1.4128e-06, -1.0086e-06, -1.0710e-06, ..., -6.9104e-07, -1.0608e-06, -9.7603e-07], [ 4.8101e-05, 1.2361e-05, 5.9381e-06, ..., 8.7246e-06, 5.6922e-06, 8.7470e-06], ..., [-6.1803e-06, -3.6955e-06, -3.3882e-06, ..., -5.7295e-06, -7.3798e-06, -7.0296e-06], [-1.5056e-04, -3.7521e-05, -1.7092e-05, ..., -2.5272e-05, -1.4730e-05, -2.4661e-05], [ 3.8624e-05, 1.0028e-05, 4.9323e-06, ..., 6.9328e-06, 4.5486e-06, 6.9775e-06]], device='cuda:0') Epoch 21, bias, value: tensor([-0.0320, 0.0540, -0.0647, -0.0240, 0.0552, -0.0512, 0.0674], device='cuda:0'), grad: tensor([ 4.3154e-05, -3.6228e-07, 1.2267e-04, 1.3447e-04, -1.0043e-05, -3.8791e-04, 9.7752e-05], device='cuda:0') 306 0.00811744900929367 changing lr epoch 20, time 429.61, cls_loss 0.0334 cls_loss_mapping 0.1398 cls_loss_causal 0.8032 re_mapping 0.0706 re_causal 0.0716 /// teacc 84.62 lr 0.00793893 Epoch 22, weight, value: tensor([[ 0.2576, 0.2569, 0.2942, ..., -0.0177, -0.0028, -0.0160], [-0.0645, -0.0590, -0.0853, ..., 0.0635, 0.0634, 0.0551], [-0.0925, -0.0848, -0.1032, ..., 0.0532, 0.0312, 0.0227], ..., [-0.1247, -0.1288, -0.0756, ..., 0.0144, 0.0759, 0.0548], [-0.0171, -0.0135, -0.0409, ..., -0.1600, -0.1781, -0.1935], [ 0.1069, 0.1103, 0.0983, ..., -0.0300, -0.0205, 0.0080]], device='cuda:0'), grad: tensor([[-8.5533e-05, -6.3598e-05, -6.1393e-05, ..., -1.3635e-05, -1.5028e-05, -1.5661e-05], [-9.6858e-05, -2.0891e-05, -1.3158e-05, ..., -1.9193e-05, -2.6330e-05, -3.3528e-05], [ 1.8907e-04, 4.3601e-05, 3.0041e-05, ..., 4.0948e-05, 4.7296e-05, 5.7191e-05], ..., [ 8.9645e-05, 4.6551e-05, 4.2140e-05, ..., 1.4164e-05, 1.6123e-05, 1.9029e-05], [ 3.4887e-06, 1.4324e-06, 1.4063e-06, ..., -2.2911e-07, 7.2271e-07, 9.7789e-07], [-1.1152e-04, -9.6262e-06, -6.3889e-07, ..., -2.4766e-05, -2.5764e-05, -3.1590e-05]], device='cuda:0') Epoch 22, bias, value: tensor([-0.0325, 0.0521, -0.0610, -0.0267, 0.0597, -0.0502, 0.0632], device='cuda:0'), grad: tensor([-9.4354e-05, -2.4581e-04, 4.6301e-04, 2.9758e-05, 1.5485e-04, 6.5416e-06, -3.1376e-04], device='cuda:0') 306 0.007938926261462368 changing lr epoch 21, time 426.58, cls_loss 0.0379 cls_loss_mapping 0.1226 cls_loss_causal 0.8209 re_mapping 0.0672 re_causal 0.0683 /// teacc 80.77 lr 0.00775448 Epoch 23, weight, value: tensor([[ 0.2543, 0.2548, 0.2918, ..., -0.0181, -0.0035, -0.0166], [-0.0610, -0.0557, -0.0815, ..., 0.0627, 0.0633, 0.0549], [-0.0944, -0.0856, -0.1037, ..., 0.0523, 0.0305, 0.0222], ..., [-0.1257, -0.1288, -0.0764, ..., 0.0143, 0.0749, 0.0540], [-0.0155, -0.0136, -0.0409, ..., -0.1578, -0.1759, -0.1909], [ 0.1065, 0.1089, 0.0971, ..., -0.0297, -0.0205, 0.0078]], device='cuda:0'), grad: tensor([[-2.0206e-05, -1.4797e-05, -1.2912e-05, ..., -2.2482e-06, -2.2426e-06, -2.8517e-06], [ 3.0920e-07, 3.2224e-07, 3.0361e-07, ..., -7.0781e-08, -1.7323e-07, -1.7136e-07], [ 8.7991e-06, 3.8929e-06, 3.2075e-06, ..., 2.0005e-06, 2.0061e-06, 2.2966e-06], ..., [-2.0675e-07, 1.2089e-06, 1.0785e-06, ..., -2.2911e-06, -2.1961e-06, -2.2668e-06], [-6.8881e-06, -1.1437e-06, -6.6683e-07, ..., -1.4920e-06, -1.4491e-06, -1.6559e-06], [ 1.3739e-05, 9.4026e-06, 8.1286e-06, ..., 1.7323e-06, 1.7434e-06, 2.1514e-06]], device='cuda:0') Epoch 23, bias, value: tensor([-0.0375, 0.0531, -0.0654, -0.0235, 0.0556, -0.0433, 0.0657], device='cuda:0'), grad: tensor([-1.7449e-05, -2.8312e-07, 1.7524e-05, 1.3448e-05, -7.6182e-06, -2.0295e-05, 1.4588e-05], device='cuda:0') 306 0.007754484907260515 changing lr epoch 22, time 426.67, cls_loss 0.0169 cls_loss_mapping 0.1011 cls_loss_causal 0.7419 re_mapping 0.0658 re_causal 0.0668 /// teacc 85.58 lr 0.00756450 Epoch 24, weight, value: tensor([[ 0.2518, 0.2533, 0.2901, ..., -0.0181, -0.0037, -0.0170], [-0.0619, -0.0557, -0.0812, ..., 0.0615, 0.0621, 0.0538], [-0.0946, -0.0862, -0.1040, ..., 0.0516, 0.0301, 0.0220], ..., [-0.1201, -0.1249, -0.0734, ..., 0.0155, 0.0755, 0.0550], [-0.0152, -0.0134, -0.0403, ..., -0.1562, -0.1744, -0.1893], [ 0.1044, 0.1070, 0.0952, ..., -0.0296, -0.0204, 0.0076]], device='cuda:0'), grad: tensor([[ 3.6269e-05, 1.0885e-05, 5.3830e-06, ..., 2.0787e-05, 2.2650e-05, 2.5585e-05], [ 7.8753e-06, 2.4289e-06, 1.2554e-06, ..., 4.4480e-06, 4.8541e-06, 5.4799e-06], [ 9.0837e-05, 2.9549e-05, 1.5378e-05, ..., 5.3287e-05, 5.8591e-05, 6.5625e-05], ..., [-1.1673e-03, -3.6597e-04, -1.8895e-04, ..., -6.7091e-04, -7.3385e-04, -8.2588e-04], [ 8.7166e-04, 2.7275e-04, 1.4091e-04, ..., 5.0020e-04, 5.4693e-04, 6.1560e-04], [ 4.8757e-05, 1.5154e-05, 7.8380e-06, ..., 2.7969e-05, 3.0503e-05, 3.4362e-05]], device='cuda:0') Epoch 24, bias, value: tensor([-0.0378, 0.0497, -0.0652, -0.0239, 0.0594, -0.0413, 0.0638], device='cuda:0'), grad: tensor([ 1.0443e-04, 2.2441e-05, 2.6131e-04, 3.2115e-04, -3.3417e-03, 2.4929e-03, 1.3912e-04], device='cuda:0') 306 0.007564496387029534 changing lr epoch 23, time 429.30, cls_loss 0.0200 cls_loss_mapping 0.1184 cls_loss_causal 0.7899 re_mapping 0.0647 re_causal 0.0658 /// teacc 85.58 lr 0.00736934 Epoch 25, weight, value: tensor([[ 0.2588, 0.2577, 0.2946, ..., -0.0171, -0.0030, -0.0160], [-0.0635, -0.0567, -0.0819, ..., 0.0609, 0.0619, 0.0534], [-0.0968, -0.0871, -0.1049, ..., 0.0503, 0.0291, 0.0212], ..., [-0.1225, -0.1247, -0.0735, ..., 0.0152, 0.0745, 0.0544], [-0.0159, -0.0147, -0.0414, ..., -0.1544, -0.1725, -0.1874], [ 0.1051, 0.1059, 0.0938, ..., -0.0290, -0.0201, 0.0078]], device='cuda:0'), grad: tensor([[-7.3481e-04, -4.4537e-04, -4.3368e-04, ..., -7.9334e-05, -8.5056e-05, -9.8109e-05], [ 3.8147e-05, 1.6361e-05, 1.4454e-05, ..., 9.4771e-06, 1.0312e-05, 1.1541e-05], [ 2.7347e-04, 1.6248e-04, 1.5748e-04, ..., 3.1322e-05, 3.3647e-05, 3.8832e-05], ..., [ 1.1367e-04, 6.1572e-05, 5.8204e-05, ..., 1.6898e-05, 1.8463e-05, 2.0921e-05], [ 1.5259e-04, 1.1063e-04, 1.1152e-04, ..., 3.8184e-06, 4.9807e-06, 5.8860e-06], [ 6.7055e-05, 4.6581e-05, 4.6998e-05, ..., 3.6433e-06, 2.3656e-06, 3.5577e-06]], device='cuda:0') Epoch 25, bias, value: tensor([-0.0299, 0.0463, -0.0686, -0.0247, 0.0534, -0.0402, 0.0682], device='cuda:0'), grad: tensor([-9.9564e-04, 7.6473e-05, 3.8290e-04, 1.4734e-04, 1.8203e-04, 1.3793e-04, 6.9320e-05], device='cuda:0') 306 0.007369343312364995 changing lr epoch 24, time 426.07, cls_loss 0.0251 cls_loss_mapping 0.1051 cls_loss_causal 0.7912 re_mapping 0.0639 re_causal 0.0653 /// teacc 86.06 lr 0.00716942 Epoch 26, weight, value: tensor([[ 0.2537, 0.2548, 0.2915, ..., -0.0177, -0.0040, -0.0170], [-0.0597, -0.0541, -0.0789, ..., 0.0608, 0.0620, 0.0536], [-0.0957, -0.0868, -0.1045, ..., 0.0499, 0.0289, 0.0212], ..., [-0.1195, -0.1230, -0.0726, ..., 0.0159, 0.0747, 0.0549], [-0.0175, -0.0150, -0.0414, ..., -0.1536, -0.1717, -0.1865], [ 0.1010, 0.1035, 0.0916, ..., -0.0302, -0.0214, 0.0061]], device='cuda:0'), grad: tensor([[ 1.2159e-04, 5.0753e-05, 4.7743e-05, ..., 3.1084e-05, 3.1054e-05, 3.8862e-05], [ 2.5213e-05, 1.1273e-05, 1.0602e-05, ..., 5.8673e-06, 5.9754e-06, 7.4059e-06], [ 1.0826e-05, 5.6848e-06, 5.4725e-06, ..., 2.8498e-06, 3.1032e-06, 3.4459e-06], ..., [-3.8218e-04, -1.7190e-04, -1.6272e-04, ..., -9.3102e-05, -9.3400e-05, -1.1605e-04], [ 5.5507e-06, 4.1053e-06, 4.2394e-06, ..., 1.8217e-06, 1.5423e-06, 1.8962e-06], [ 3.5197e-05, 1.7121e-05, 1.6272e-05, ..., 7.8604e-06, 7.9349e-06, 9.8199e-06]], device='cuda:0') Epoch 26, bias, value: tensor([-0.0365, 0.0497, -0.0662, -0.0146, 0.0565, -0.0447, 0.0602], device='cuda:0'), grad: tensor([ 2.1684e-04, 4.3213e-05, 1.8209e-05, 3.1137e-04, -6.5041e-04, 4.7274e-06, 5.6416e-05], device='cuda:0') 306 0.0071694186955877925 changing lr epoch 25, time 425.98, cls_loss 0.0207 cls_loss_mapping 0.1023 cls_loss_causal 0.7798 re_mapping 0.0635 re_causal 0.0652 /// teacc 85.10 lr 0.00696513 Epoch 27, weight, value: tensor([[ 0.2559, 0.2547, 0.2909, ..., -0.0169, -0.0032, -0.0158], [-0.0587, -0.0531, -0.0774, ..., 0.0610, 0.0621, 0.0538], [-0.0948, -0.0862, -0.1037, ..., 0.0492, 0.0285, 0.0209], ..., [-0.1254, -0.1251, -0.0754, ..., 0.0145, 0.0725, 0.0526], [-0.0178, -0.0154, -0.0415, ..., -0.1522, -0.1701, -0.1848], [ 0.1037, 0.1041, 0.0924, ..., -0.0293, -0.0205, 0.0069]], device='cuda:0'), grad: tensor([[ 0.0110, 0.0050, 0.0045, ..., 0.0032, 0.0036, 0.0042], [ 0.0048, 0.0021, 0.0020, ..., 0.0013, 0.0015, 0.0018], [-0.0494, -0.0227, -0.0208, ..., -0.0142, -0.0161, -0.0184], ..., [-0.0014, -0.0005, -0.0005, ..., -0.0004, -0.0004, -0.0005], [ 0.0221, 0.0103, 0.0094, ..., 0.0062, 0.0070, 0.0081], [ 0.0116, 0.0054, 0.0049, ..., 0.0033, 0.0037, 0.0043]], device='cuda:0') Epoch 27, bias, value: tensor([-0.0291, 0.0499, -0.0656, -0.0168, 0.0440, -0.0445, 0.0666], device='cuda:0'), grad: tensor([ 0.0222, 0.0096, -0.0975, 0.0028, -0.0031, 0.0432, 0.0229], device='cuda:0') 306 0.0069651251582696205 changing lr epoch 26, time 430.16, cls_loss 0.0187 cls_loss_mapping 0.0969 cls_loss_causal 0.7636 re_mapping 0.0622 re_causal 0.0637 /// teacc 81.25 lr 0.00675687 Epoch 28, weight, value: tensor([[ 0.2514, 0.2523, 0.2883, ..., -0.0173, -0.0038, -0.0165], [-0.0548, -0.0502, -0.0744, ..., 0.0596, 0.0609, 0.0527], [-0.0947, -0.0856, -0.1030, ..., 0.0477, 0.0272, 0.0195], ..., [-0.1235, -0.1248, -0.0757, ..., 0.0164, 0.0739, 0.0545], [-0.0161, -0.0152, -0.0411, ..., -0.1507, -0.1689, -0.1832], [ 0.1032, 0.1029, 0.0916, ..., -0.0280, -0.0190, 0.0081]], device='cuda:0'), grad: tensor([[-6.0415e-04, -3.0947e-04, -3.3808e-04, ..., -1.3041e-04, -1.2791e-04, -1.2898e-04], [ 4.1664e-05, 2.1353e-05, 2.3395e-05, ..., 9.0152e-06, 8.7842e-06, 8.8438e-06], [ 1.7655e-04, 9.0182e-05, 9.8526e-05, ..., 3.8236e-05, 3.7462e-05, 3.7819e-05], ..., [ 1.4520e-04, 7.4148e-05, 8.1003e-05, ..., 3.1382e-05, 3.0756e-05, 3.1054e-05], [ 1.5426e-04, 7.8619e-05, 8.5890e-05, ..., 3.3408e-05, 3.2693e-05, 3.3081e-05], [ 3.6776e-05, 1.9848e-05, 2.1681e-05, ..., 7.6070e-06, 7.5847e-06, 7.5288e-06]], device='cuda:0') Epoch 28, bias, value: tensor([-0.0358, 0.0521, -0.0681, -0.0243, 0.0515, -0.0394, 0.0684], device='cuda:0'), grad: tensor([-9.4175e-04, 6.4969e-05, 2.7657e-04, 7.8797e-05, 2.2757e-04, 2.4247e-04, 5.2065e-05], device='cuda:0') 306 0.006756874120406716 changing lr epoch 27, time 426.89, cls_loss 0.0093 cls_loss_mapping 0.0824 cls_loss_causal 0.7365 re_mapping 0.0618 re_causal 0.0635 /// teacc 83.17 lr 0.00654508 Epoch 29, weight, value: tensor([[ 0.2525, 0.2524, 0.2881, ..., -0.0166, -0.0033, -0.0158], [-0.0550, -0.0497, -0.0737, ..., 0.0594, 0.0608, 0.0527], [-0.0931, -0.0852, -0.1024, ..., 0.0472, 0.0270, 0.0195], ..., [-0.1213, -0.1236, -0.0752, ..., 0.0161, 0.0732, 0.0540], [-0.0171, -0.0156, -0.0414, ..., -0.1496, -0.1678, -0.1820], [ 0.0988, 0.1004, 0.0894, ..., -0.0284, -0.0196, 0.0071]], device='cuda:0'), grad: tensor([[ 8.8453e-04, 5.9783e-05, 7.9036e-05, ..., 3.7479e-04, 4.0603e-04, 4.2248e-04], [ 8.6021e-04, 1.4782e-04, 1.5628e-04, ..., 3.4332e-04, 3.6836e-04, 3.8695e-04], [-3.3932e-03, -5.7173e-04, -6.1417e-04, ..., -1.2398e-03, -1.3380e-03, -1.4076e-03], ..., [ 7.9203e-04, 1.7011e-04, 1.7536e-04, ..., 2.8706e-04, 3.0875e-04, 3.2568e-04], [ 9.2268e-04, 1.6189e-04, 1.6904e-04, ..., 3.9434e-04, 4.2653e-04, 4.4417e-04], [ 6.0749e-04, 1.1957e-04, 1.2189e-04, ..., 2.7061e-04, 2.9230e-04, 3.0398e-04]], device='cuda:0') Epoch 29, bias, value: tensor([-0.0325, 0.0499, -0.0641, -0.0240, 0.0539, -0.0409, 0.0621], device='cuda:0'), grad: tensor([ 0.0026, 0.0023, -0.0089, -0.0022, 0.0020, 0.0025, 0.0017], device='cuda:0') 306 0.00654508497187474 changing lr epoch 28, time 428.99, cls_loss 0.0094 cls_loss_mapping 0.0876 cls_loss_causal 0.7353 re_mapping 0.0604 re_causal 0.0621 /// teacc 80.77 lr 0.00633018 Epoch 30, weight, value: tensor([[ 0.2499, 0.2511, 0.2864, ..., -0.0170, -0.0040, -0.0164], [-0.0549, -0.0491, -0.0729, ..., 0.0583, 0.0598, 0.0516], [-0.0926, -0.0852, -0.1023, ..., 0.0469, 0.0268, 0.0194], ..., [-0.1186, -0.1221, -0.0742, ..., 0.0167, 0.0734, 0.0544], [-0.0186, -0.0162, -0.0417, ..., -0.1486, -0.1666, -0.1807], [ 0.0981, 0.0995, 0.0886, ..., -0.0280, -0.0192, 0.0073]], device='cuda:0'), grad: tensor([[-1.0252e-04, -6.0618e-05, -6.3598e-05, ..., -1.8939e-05, -2.0102e-05, -2.1175e-05], [ 1.9372e-07, 9.1270e-07, 1.0356e-06, ..., -1.5274e-07, -1.6391e-07, -1.3039e-07], [ 8.8155e-05, 4.9591e-05, 5.1677e-05, ..., 1.5959e-05, 1.6943e-05, 1.8016e-05], ..., [ 2.2665e-05, 5.7630e-06, 4.2319e-06, ..., 5.5581e-06, 6.2585e-06, 7.0035e-06], [-1.6429e-06, 1.9372e-07, 3.2037e-07, ..., 1.7881e-07, 2.0862e-07, -2.9802e-08], [-8.9109e-06, 3.0212e-06, 5.1446e-06, ..., -2.9318e-06, -3.5129e-06, -4.0941e-06]], device='cuda:0') Epoch 30, bias, value: tensor([-0.0365, 0.0475, -0.0623, -0.0214, 0.0576, -0.0432, 0.0627], device='cuda:0'), grad: tensor([-1.2743e-04, -6.9290e-07, 1.1796e-04, 3.1106e-06, 5.4836e-05, -1.0312e-05, -3.7611e-05], device='cuda:0') 306 0.006330184227833378 changing lr ---------------------saving model at epoch 29---------------------------------------------------- epoch 29, time 434.47, cls_loss 0.0094 cls_loss_mapping 0.0785 cls_loss_causal 0.7117 re_mapping 0.0606 re_causal 0.0626 /// teacc 87.98 lr 0.00611260 Epoch 31, weight, value: tensor([[ 0.2503, 0.2510, 0.2860, ..., -0.0165, -0.0035, -0.0157], [-0.0539, -0.0484, -0.0720, ..., 0.0578, 0.0594, 0.0514], [-0.0942, -0.0859, -0.1026, ..., 0.0460, 0.0261, 0.0188], ..., [-0.1188, -0.1221, -0.0747, ..., 0.0166, 0.0727, 0.0539], [-0.0162, -0.0152, -0.0408, ..., -0.1465, -0.1646, -0.1787], [ 0.0972, 0.0984, 0.0877, ..., -0.0281, -0.0194, 0.0069]], device='cuda:0'), grad: tensor([[-4.4346e-04, -2.6488e-04, -2.6155e-04, ..., -9.3997e-05, -1.1146e-04, -1.1992e-04], [-1.3721e-04, -4.3064e-05, -4.4137e-05, ..., -5.4270e-05, -5.3287e-05, -5.9724e-05], [ 3.0375e-04, 1.5330e-04, 1.5008e-04, ..., 8.3685e-05, 9.2566e-05, 1.0157e-04], ..., [ 1.5867e-04, 9.5129e-05, 9.6679e-05, ..., 3.1501e-05, 3.5971e-05, 3.8385e-05], [ 3.0577e-05, 1.7941e-05, 1.7643e-05, ..., 7.5772e-06, 8.9332e-06, 9.7007e-06], [ 5.4836e-05, 2.5928e-05, 2.5794e-05, ..., 1.6108e-05, 1.7121e-05, 1.8835e-05]], device='cuda:0') Epoch 31, bias, value: tensor([-0.0348, 0.0479, -0.0655, -0.0245, 0.0561, -0.0370, 0.0620], device='cuda:0'), grad: tensor([-5.3549e-04, -4.0984e-04, 5.2595e-04, 6.1989e-05, 2.0742e-04, 4.1455e-05, 1.0896e-04], device='cuda:0') 306 0.006112604669781575 changing lr epoch 30, time 430.27, cls_loss 0.0094 cls_loss_mapping 0.0758 cls_loss_causal 0.7057 re_mapping 0.0586 re_causal 0.0606 /// teacc 87.98 lr 0.00589278 Epoch 32, weight, value: tensor([[ 0.2509, 0.2513, 0.2860, ..., -0.0162, -0.0034, -0.0155], [-0.0541, -0.0483, -0.0717, ..., 0.0573, 0.0590, 0.0510], [-0.0944, -0.0854, -0.1020, ..., 0.0452, 0.0255, 0.0182], ..., [-0.1200, -0.1226, -0.0757, ..., 0.0161, 0.0717, 0.0530], [-0.0130, -0.0142, -0.0394, ..., -0.1452, -0.1632, -0.1771], [ 0.0928, 0.0963, 0.0856, ..., -0.0284, -0.0198, 0.0063]], device='cuda:0'), grad: tensor([[ 1.1826e-03, 5.4789e-04, 4.1938e-04, ..., 3.5620e-04, 4.4441e-04, 4.4823e-04], [-3.4008e-03, -1.2341e-03, -9.7752e-04, ..., -9.1982e-04, -1.2970e-03, -1.2903e-03], [ 1.4651e-04, 4.3333e-05, 2.9847e-05, ..., 3.4630e-05, 6.0350e-05, 5.6565e-05], ..., [ 8.8120e-04, 2.8610e-04, 2.3878e-04, ..., 1.9145e-04, 3.2616e-04, 3.1114e-04], [ 3.8791e-04, 9.6798e-05, 7.8917e-05, ..., 1.2290e-04, 1.5807e-04, 1.6606e-04], [ 5.0402e-04, 1.7214e-04, 1.4174e-04, ..., 1.1754e-04, 1.8585e-04, 1.8072e-04]], device='cuda:0') Epoch 32, bias, value: tensor([-0.0336, 0.0461, -0.0671, -0.0183, 0.0533, -0.0309, 0.0549], device='cuda:0'), grad: tensor([ 0.0021, -0.0077, 0.0004, 0.0007, 0.0023, 0.0010, 0.0012], device='cuda:0') 306 0.005892784473993186 changing lr ---------------------saving model at epoch 31---------------------------------------------------- epoch 31, time 433.10, cls_loss 0.0076 cls_loss_mapping 0.0626 cls_loss_causal 0.6597 re_mapping 0.0570 re_causal 0.0588 /// teacc 88.46 lr 0.00567117 Epoch 33, weight, value: tensor([[ 0.2517, 0.2513, 0.2859, ..., -0.0157, -0.0029, -0.0148], [-0.0529, -0.0478, -0.0712, ..., 0.0578, 0.0593, 0.0514], [-0.0946, -0.0857, -0.1020, ..., 0.0441, 0.0248, 0.0176], ..., [-0.1194, -0.1216, -0.0752, ..., 0.0161, 0.0713, 0.0527], [-0.0172, -0.0160, -0.0411, ..., -0.1449, -0.1629, -0.1767], [ 0.0949, 0.0966, 0.0861, ..., -0.0278, -0.0193, 0.0065]], device='cuda:0'), grad: tensor([[-1.3514e-03, -7.8011e-04, -7.8249e-04, ..., -2.5344e-04, -2.6321e-04, -3.0375e-04], [ 2.7561e-04, 1.6153e-04, 1.6010e-04, ..., 5.0634e-05, 5.1320e-05, 6.0827e-05], [ 2.5606e-04, 1.3661e-04, 1.3423e-04, ..., 5.1290e-05, 5.2512e-05, 6.0558e-05], ..., [ 3.1090e-04, 1.7726e-04, 1.7977e-04, ..., 5.7518e-05, 6.1095e-05, 6.9439e-05], [ 5.1171e-05, 4.8816e-05, 5.0664e-05, ..., 1.1854e-05, 1.3553e-05, 1.4424e-05], [ 2.8563e-04, 1.6415e-04, 1.6391e-04, ..., 5.2303e-05, 5.2512e-05, 6.1929e-05]], device='cuda:0') Epoch 33, bias, value: tensor([-0.0312, 0.0491, -0.0674, -0.0199, 0.0518, -0.0381, 0.0600], device='cuda:0'), grad: tensor([-1.9341e-03, 3.7980e-04, 4.0054e-04, 2.6917e-04, 4.5681e-04, 2.0683e-05, 4.0674e-04], device='cuda:0') 306 0.00567116632908828 changing lr epoch 32, time 426.99, cls_loss 0.0106 cls_loss_mapping 0.0678 cls_loss_causal 0.6951 re_mapping 0.0560 re_causal 0.0581 /// teacc 82.69 lr 0.00544820 Epoch 34, weight, value: tensor([[ 0.2521, 0.2514, 0.2857, ..., -0.0157, -0.0028, -0.0148], [-0.0508, -0.0470, -0.0703, ..., 0.0579, 0.0594, 0.0517], [-0.0956, -0.0860, -0.1020, ..., 0.0434, 0.0242, 0.0169], ..., [-0.1187, -0.1210, -0.0750, ..., 0.0163, 0.0710, 0.0527], [-0.0184, -0.0165, -0.0415, ..., -0.1441, -0.1620, -0.1757], [ 0.0944, 0.0958, 0.0854, ..., -0.0276, -0.0192, 0.0065]], device='cuda:0'), grad: tensor([[ 5.7459e-04, 2.7680e-04, 2.7418e-04, ..., 5.6207e-05, 6.7353e-05, 1.0335e-04], [ 1.3304e-04, 6.0856e-05, 5.9932e-05, ..., 1.6704e-05, 1.9088e-05, 2.7254e-05], [-1.2275e-06, 2.3041e-06, 2.5667e-06, ..., -1.2163e-06, -2.1253e-06, -1.9837e-06], ..., [ 2.1264e-05, 1.0088e-05, 1.0043e-05, ..., 2.2147e-06, 2.6375e-06, 3.8743e-06], [ 1.0014e-04, 4.5031e-05, 4.4435e-05, ..., 1.4551e-05, 1.6063e-05, 2.1860e-05], [-8.6927e-04, -4.2367e-04, -4.2057e-04, ..., -8.1062e-05, -9.7871e-05, -1.5199e-04]], device='cuda:0') Epoch 34, bias, value: tensor([-0.0304, 0.0528, -0.0694, -0.0211, 0.0513, -0.0399, 0.0607], device='cuda:0'), grad: tensor([ 1.0033e-03, 2.4414e-04, -6.8992e-06, 3.7402e-05, 3.8028e-05, 1.8930e-04, -1.5049e-03], device='cuda:0') 306 0.00544819654451717 changing lr epoch 33, time 427.05, cls_loss 0.0080 cls_loss_mapping 0.0678 cls_loss_causal 0.7349 re_mapping 0.0560 re_causal 0.0582 /// teacc 86.54 lr 0.00522432 Epoch 35, weight, value: tensor([[ 0.2496, 0.2498, 0.2841, ..., -0.0159, -0.0032, -0.0152], [-0.0509, -0.0468, -0.0699, ..., 0.0572, 0.0588, 0.0512], [-0.0940, -0.0851, -0.1011, ..., 0.0432, 0.0242, 0.0171], ..., [-0.1182, -0.1201, -0.0745, ..., 0.0161, 0.0704, 0.0523], [-0.0183, -0.0166, -0.0416, ..., -0.1430, -0.1609, -0.1745], [ 0.0933, 0.0948, 0.0846, ..., -0.0276, -0.0192, 0.0062]], device='cuda:0'), grad: tensor([[-2.4395e-03, -1.4315e-03, -1.4219e-03, ..., -6.6614e-04, -6.5660e-04, -7.2098e-04], [ 4.0746e-04, 1.6570e-04, 1.5414e-04, ..., 8.9586e-05, 9.1910e-05, 1.0550e-04], [ 4.8685e-04, 2.4652e-04, 2.3901e-04, ..., 1.1039e-04, 1.1253e-04, 1.2529e-04], ..., [ 8.7929e-04, 5.0354e-04, 4.9829e-04, ..., 2.5129e-04, 2.4438e-04, 2.7108e-04], [-3.9077e-04, 3.8408e-06, 4.0323e-05, ..., -4.2140e-05, -4.7147e-05, -6.9320e-05], [ 8.9645e-04, 4.3797e-04, 4.2105e-04, ..., 2.1875e-04, 2.1732e-04, 2.4557e-04]], device='cuda:0') Epoch 35, bias, value: tensor([-0.0333, 0.0510, -0.0670, -0.0178, 0.0498, -0.0383, 0.0596], device='cuda:0'), grad: tensor([-0.0039, 0.0008, 0.0008, 0.0003, 0.0015, -0.0012, 0.0016], device='cuda:0') 306 0.005224324151752577 changing lr epoch 34, time 428.33, cls_loss 0.0042 cls_loss_mapping 0.0575 cls_loss_causal 0.6963 re_mapping 0.0553 re_causal 0.0577 /// teacc 85.58 lr 0.00500000 Epoch 36, weight, value: tensor([[ 0.2492, 0.2493, 0.2833, ..., -0.0158, -0.0032, -0.0151], [-0.0521, -0.0472, -0.0701, ..., 0.0566, 0.0582, 0.0505], [-0.0929, -0.0847, -0.1006, ..., 0.0430, 0.0242, 0.0171], ..., [-0.1164, -0.1190, -0.0738, ..., 0.0165, 0.0704, 0.0524], [-0.0172, -0.0164, -0.0414, ..., -0.1417, -0.1595, -0.1728], [ 0.0920, 0.0938, 0.0838, ..., -0.0276, -0.0193, 0.0059]], device='cuda:0'), grad: tensor([[-2.1362e-04, -1.1706e-04, -1.1629e-04, ..., -5.4300e-05, -5.6475e-05, -5.9038e-05], [ 2.7761e-05, 1.4283e-05, 1.4096e-05, ..., 6.8471e-06, 7.1414e-06, 7.5735e-06], [ 1.2493e-04, 6.5327e-05, 6.4611e-05, ..., 3.1233e-05, 3.2574e-05, 3.4422e-05], ..., [ 1.7971e-05, 9.3132e-06, 9.1493e-06, ..., 3.6974e-06, 3.9116e-06, 4.1649e-06], [ 4.7088e-05, 2.2233e-05, 2.1875e-05, ..., 1.0654e-05, 1.1109e-05, 1.2226e-05], [-1.9550e-05, -2.3656e-06, -1.5628e-06, ..., -1.5832e-06, -1.8850e-06, -3.2447e-06]], device='cuda:0') Epoch 36, bias, value: tensor([-0.0330, 0.0474, -0.0645, -0.0211, 0.0514, -0.0336, 0.0575], device='cuda:0'), grad: tensor([-3.4666e-04, 4.8816e-05, 2.1541e-04, 2.4825e-05, 3.2306e-05, 9.2864e-05, -6.8188e-05], device='cuda:0') 306 0.005000000000000003 changing lr epoch 35, time 428.68, cls_loss 0.0057 cls_loss_mapping 0.0564 cls_loss_causal 0.6790 re_mapping 0.0534 re_causal 0.0557 /// teacc 86.54 lr 0.00477568 Epoch 37, weight, value: tensor([[ 0.2490, 0.2492, 0.2831, ..., -0.0157, -0.0032, -0.0150], [-0.0533, -0.0474, -0.0702, ..., 0.0559, 0.0575, 0.0498], [-0.0930, -0.0845, -0.1003, ..., 0.0425, 0.0239, 0.0169], ..., [-0.1153, -0.1182, -0.0733, ..., 0.0167, 0.0703, 0.0524], [-0.0177, -0.0169, -0.0417, ..., -0.1408, -0.1588, -0.1719], [ 0.0928, 0.0935, 0.0835, ..., -0.0270, -0.0188, 0.0063]], device='cuda:0'), grad: tensor([[-1.9026e-04, -1.2153e-04, -1.2040e-04, ..., -2.5526e-05, -2.8968e-05, -3.4362e-05], [ 1.2696e-04, 4.9949e-05, 4.9144e-05, ..., 3.4660e-05, 4.0740e-05, 4.1753e-05], [ 8.4221e-05, 4.2409e-05, 4.1664e-05, ..., 1.6063e-05, 1.8761e-05, 2.1011e-05], ..., [-4.0221e-04, -1.1349e-04, -1.1182e-04, ..., -1.4126e-04, -1.6749e-04, -1.6463e-04], [ 1.6600e-05, 1.8686e-05, 2.0117e-05, ..., 3.7812e-07, 2.9206e-06, 2.1532e-06], [ 1.7178e-04, 6.0230e-05, 5.8681e-05, ..., 5.3525e-05, 6.0827e-05, 6.0886e-05]], device='cuda:0') Epoch 37, bias, value: tensor([-0.0331, 0.0438, -0.0652, -0.0216, 0.0517, -0.0324, 0.0609], device='cuda:0'), grad: tensor([-2.2066e-04, 2.5630e-04, 1.3638e-04, 4.3583e-04, -9.7466e-04, -8.4341e-06, 3.7503e-04], device='cuda:0') 306 0.004775675848247429 changing lr epoch 36, time 426.09, cls_loss 0.0056 cls_loss_mapping 0.0548 cls_loss_causal 0.6659 re_mapping 0.0529 re_causal 0.0553 /// teacc 84.62 lr 0.00455180 Epoch 38, weight, value: tensor([[ 0.2494, 0.2492, 0.2830, ..., -0.0157, -0.0032, -0.0149], [-0.0519, -0.0468, -0.0695, ..., 0.0557, 0.0575, 0.0499], [-0.0936, -0.0847, -0.1003, ..., 0.0419, 0.0234, 0.0164], ..., [-0.1138, -0.1172, -0.0726, ..., 0.0171, 0.0703, 0.0526], [-0.0186, -0.0172, -0.0419, ..., -0.1402, -0.1581, -0.1712], [ 0.0912, 0.0921, 0.0822, ..., -0.0268, -0.0187, 0.0063]], device='cuda:0'), grad: tensor([[ 3.4750e-05, 1.2450e-05, 1.1384e-05, ..., 8.5309e-06, 9.8050e-06, 1.1593e-05], [-1.0271e-03, -4.1318e-04, -3.9172e-04, ..., -1.4448e-04, -1.7440e-04, -2.1589e-04], [ 5.4985e-06, 2.5854e-06, 2.5835e-06, ..., 2.0489e-08, 1.8626e-08, 2.0862e-07], ..., [-4.6492e-06, -3.4589e-06, -3.4049e-06, ..., -4.2692e-06, -5.1111e-06, -5.2936e-06], [-1.3888e-05, -8.7917e-07, -6.4634e-07, ..., -3.5372e-06, -2.5574e-06, -3.6340e-06], [ 9.9373e-04, 3.9911e-04, 3.7861e-04, ..., 1.4126e-04, 1.6975e-04, 2.0993e-04]], device='cuda:0') Epoch 38, bias, value: tensor([-0.0323, 0.0466, -0.0664, -0.0230, 0.0530, -0.0339, 0.0600], device='cuda:0'), grad: tensor([ 7.5400e-05, -1.8406e-03, 8.1286e-06, 3.1173e-05, -1.1809e-05, -4.4286e-05, 1.7853e-03], device='cuda:0') 306 0.004551803455482836 changing lr epoch 37, time 425.12, cls_loss 0.0061 cls_loss_mapping 0.0590 cls_loss_causal 0.6774 re_mapping 0.0525 re_causal 0.0551 /// teacc 83.17 lr 0.00432883 Epoch 39, weight, value: tensor([[ 0.2483, 0.2485, 0.2821, ..., -0.0158, -0.0035, -0.0151], [-0.0507, -0.0462, -0.0687, ..., 0.0556, 0.0575, 0.0499], [-0.0933, -0.0845, -0.1000, ..., 0.0415, 0.0232, 0.0162], ..., [-0.1130, -0.1164, -0.0722, ..., 0.0172, 0.0700, 0.0525], [-0.0192, -0.0175, -0.0421, ..., -0.1396, -0.1574, -0.1704], [ 0.0900, 0.0911, 0.0812, ..., -0.0268, -0.0187, 0.0061]], device='cuda:0'), grad: tensor([[-1.6487e-04, -1.0222e-04, -1.0294e-04, ..., -1.8775e-05, -1.5110e-05, -1.9088e-05], [ 2.0981e-05, 3.2540e-06, 2.7195e-06, ..., 5.4128e-06, 6.1393e-06, 7.4096e-06], [ 1.8418e-04, 1.0049e-04, 1.0014e-04, ..., 2.5466e-05, 2.3037e-05, 2.8580e-05], ..., [ 4.7654e-05, 9.2238e-06, 7.8902e-06, ..., 1.1139e-05, 1.2316e-05, 1.5184e-05], [-1.4281e-04, -1.4052e-05, -9.4771e-06, ..., -3.4928e-05, -3.9577e-05, -4.9472e-05], [ 8.9034e-06, -2.5947e-06, -2.8443e-06, ..., 2.9244e-07, 4.1351e-07, 1.5832e-06]], device='cuda:0') Epoch 39, bias, value: tensor([-0.0338, 0.0484, -0.0660, -0.0216, 0.0525, -0.0344, 0.0589], device='cuda:0'), grad: tensor([-1.9264e-04, 6.2644e-05, 2.6774e-04, 1.4007e-04, 1.3340e-04, -4.5323e-04, 4.2528e-05], device='cuda:0') 306 0.004328833670911726 changing lr ---------------------saving model at epoch 38---------------------------------------------------- epoch 38, time 430.89, cls_loss 0.0048 cls_loss_mapping 0.0498 cls_loss_causal 0.6565 re_mapping 0.0502 re_causal 0.0527 /// teacc 90.38 lr 0.00410722 Epoch 40, weight, value: tensor([[ 0.2478, 0.2481, 0.2815, ..., -0.0158, -0.0035, -0.0151], [-0.0505, -0.0459, -0.0683, ..., 0.0551, 0.0570, 0.0494], [-0.0923, -0.0841, -0.0996, ..., 0.0414, 0.0233, 0.0164], ..., [-0.1132, -0.1161, -0.0721, ..., 0.0171, 0.0696, 0.0521], [-0.0200, -0.0178, -0.0422, ..., -0.1390, -0.1568, -0.1697], [ 0.0894, 0.0902, 0.0805, ..., -0.0267, -0.0188, 0.0059]], device='cuda:0'), grad: tensor([[-3.8218e-04, -2.5368e-04, -2.5129e-04, ..., -3.4750e-05, -4.4465e-05, -5.1945e-05], [ 9.3281e-05, 5.9783e-05, 5.9187e-05, ..., 1.0341e-05, 1.3016e-05, 1.4789e-05], [ 1.3657e-05, 1.0043e-05, 9.9838e-06, ..., 1.3318e-06, 1.8254e-06, 1.9278e-06], ..., [ 5.6207e-05, 3.2425e-05, 3.2097e-05, ..., 8.9929e-06, 1.1019e-05, 1.1913e-05], [ 4.3213e-06, 8.9258e-06, 9.3654e-06, ..., -5.5321e-07, 8.8103e-07, 4.0978e-07], [ 2.1148e-04, 1.3483e-04, 1.3328e-04, ..., 2.1800e-05, 2.6941e-05, 3.1203e-05]], device='cuda:0') Epoch 40, bias, value: tensor([-0.0341, 0.0476, -0.0636, -0.0197, 0.0505, -0.0361, 0.0592], device='cuda:0'), grad: tensor([-3.9268e-04, 1.0526e-04, 1.1511e-05, -2.8744e-05, 7.9453e-05, -1.2942e-05, 2.3806e-04], device='cuda:0') 306 0.0041072155260068206 changing lr epoch 39, time 431.70, cls_loss 0.0051 cls_loss_mapping 0.0535 cls_loss_causal 0.7021 re_mapping 0.0495 re_causal 0.0520 /// teacc 86.06 lr 0.00388740 Epoch 41, weight, value: tensor([[ 0.2513, 0.2500, 0.2832, ..., -0.0149, -0.0026, -0.0140], [-0.0511, -0.0459, -0.0682, ..., 0.0546, 0.0565, 0.0489], [-0.0920, -0.0841, -0.0995, ..., 0.0412, 0.0232, 0.0164], ..., [-0.1122, -0.1155, -0.0718, ..., 0.0172, 0.0693, 0.0519], [-0.0206, -0.0180, -0.0424, ..., -0.1384, -0.1561, -0.1690], [ 0.0860, 0.0879, 0.0783, ..., -0.0271, -0.0193, 0.0051]], device='cuda:0'), grad: tensor([[-9.6977e-05, -7.2122e-05, -7.0632e-05, ..., -4.6343e-06, -6.8955e-06, -6.5528e-06], [ 7.1883e-05, 1.8924e-05, 1.8641e-05, ..., 2.4423e-05, 2.8446e-05, 3.0786e-05], [-1.3530e-04, -5.0068e-06, -4.9546e-06, ..., -5.1588e-05, -5.1260e-05, -5.7846e-05], ..., [ 1.7941e-04, 4.3571e-05, 4.2886e-05, ..., 5.6475e-05, 6.2644e-05, 6.8307e-05], [-4.5598e-05, 6.7055e-08, 9.5367e-07, ..., -1.6466e-05, -2.1920e-05, -2.2978e-05], [-3.1441e-05, 5.8487e-07, -5.1036e-07, ..., -2.8342e-05, -3.4481e-05, -3.7223e-05]], device='cuda:0') Epoch 41, bias, value: tensor([-0.0283, 0.0453, -0.0618, -0.0206, 0.0508, -0.0370, 0.0554], device='cuda:0'), grad: tensor([-7.4744e-05, 1.7619e-04, -4.0269e-04, 1.4830e-04, 4.4298e-04, -2.0444e-04, -8.5533e-05], device='cuda:0') 306 0.0038873953302184317 changing lr epoch 40, time 429.47, cls_loss 0.0052 cls_loss_mapping 0.0468 cls_loss_causal 0.6742 re_mapping 0.0483 re_causal 0.0508 /// teacc 87.02 lr 0.00366982 Epoch 42, weight, value: tensor([[ 0.2490, 0.2490, 0.2821, ..., -0.0153, -0.0030, -0.0145], [-0.0508, -0.0456, -0.0678, ..., 0.0542, 0.0562, 0.0486], [-0.0907, -0.0838, -0.0991, ..., 0.0412, 0.0233, 0.0167], ..., [-0.1121, -0.1149, -0.0715, ..., 0.0172, 0.0689, 0.0516], [-0.0202, -0.0181, -0.0423, ..., -0.1376, -0.1552, -0.1681], [ 0.0861, 0.0873, 0.0777, ..., -0.0269, -0.0191, 0.0052]], device='cuda:0'), grad: tensor([[ 7.0524e-04, 2.4354e-04, 2.1636e-04, ..., 9.5367e-05, 1.0043e-04, 1.0616e-04], [ 8.6203e-06, -1.9744e-06, -2.6431e-06, ..., -6.4969e-06, -6.2473e-06, -7.8455e-06], [ 1.5199e-04, 4.8757e-05, 4.3005e-05, ..., 3.3259e-05, 3.6716e-05, 3.8743e-05], ..., [ 9.5427e-05, 3.2485e-05, 2.9102e-05, ..., 1.6078e-05, 1.7956e-05, 1.9237e-05], [ 7.1466e-05, 2.4527e-05, 2.2009e-05, ..., 1.1332e-05, 1.2673e-05, 1.3590e-05], [-9.4128e-04, -3.2496e-04, -2.8896e-04, ..., -1.0651e-04, -1.1295e-04, -1.1897e-04]], device='cuda:0') Epoch 42, bias, value: tensor([-0.0322, 0.0448, -0.0585, -0.0213, 0.0490, -0.0356, 0.0576], device='cuda:0'), grad: tensor([ 1.4572e-03, 3.4839e-05, 3.1948e-04, -2.0826e-04, 1.9515e-04, 1.4567e-04, -1.9464e-03], device='cuda:0') 306 0.003669815772166629 changing lr epoch 41, time 427.88, cls_loss 0.0065 cls_loss_mapping 0.0534 cls_loss_causal 0.6753 re_mapping 0.0489 re_causal 0.0515 /// teacc 86.54 lr 0.00345492 Epoch 43, weight, value: tensor([[ 0.2492, 0.2492, 0.2821, ..., -0.0153, -0.0032, -0.0146], [-0.0510, -0.0455, -0.0676, ..., 0.0537, 0.0557, 0.0482], [-0.0909, -0.0837, -0.0989, ..., 0.0408, 0.0231, 0.0165], ..., [-0.1110, -0.1143, -0.0712, ..., 0.0175, 0.0690, 0.0519], [-0.0209, -0.0183, -0.0424, ..., -0.1370, -0.1546, -0.1674], [ 0.0852, 0.0864, 0.0768, ..., -0.0269, -0.0192, 0.0050]], device='cuda:0'), grad: tensor([[ 1.2779e-04, 4.0859e-05, 3.5703e-05, ..., 3.8385e-05, 4.5091e-05, 5.1558e-05], [ 7.3612e-06, 2.3488e-06, 2.0191e-06, ..., 2.3954e-06, 2.9132e-06, 3.2187e-06], [-6.0797e-05, -1.8999e-05, -1.8507e-05, ..., -4.4629e-06, -9.8720e-07, -7.0706e-06], ..., [-1.0753e-04, -3.7402e-05, -3.1054e-05, ..., -4.6730e-05, -5.9545e-05, -6.1870e-05], [-4.8988e-06, 8.9221e-07, 1.3467e-06, ..., 1.0617e-07, -4.2282e-07, -1.7136e-07], [ 1.0051e-05, 3.7905e-06, 3.2373e-06, ..., 1.7025e-06, 2.3823e-06, 2.7604e-06]], device='cuda:0') Epoch 43, bias, value: tensor([-0.0323, 0.0438, -0.0593, -0.0198, 0.0506, -0.0368, 0.0577], device='cuda:0'), grad: tensor([ 2.9945e-04, 1.6332e-05, -1.7369e-04, 6.4611e-05, -2.1088e-04, -1.8358e-05, 2.2486e-05], device='cuda:0') 306 0.0034549150281252667 changing lr epoch 42, time 431.83, cls_loss 0.0043 cls_loss_mapping 0.0461 cls_loss_causal 0.6401 re_mapping 0.0486 re_causal 0.0514 /// teacc 84.62 lr 0.00324313 Epoch 44, weight, value: tensor([[ 0.2495, 0.2492, 0.2820, ..., -0.0153, -0.0032, -0.0145], [-0.0501, -0.0451, -0.0671, ..., 0.0536, 0.0556, 0.0481], [-0.0910, -0.0837, -0.0989, ..., 0.0405, 0.0229, 0.0163], ..., [-0.1111, -0.1140, -0.0711, ..., 0.0175, 0.0687, 0.0517], [-0.0213, -0.0185, -0.0426, ..., -0.1364, -0.1540, -0.1667], [ 0.0847, 0.0858, 0.0763, ..., -0.0268, -0.0192, 0.0050]], device='cuda:0'), grad: tensor([[-1.4976e-05, -1.8388e-05, -1.7866e-05, ..., 1.8664e-06, 2.2594e-06, 2.4643e-06], [ 1.0014e-04, 9.9018e-06, 6.0499e-06, ..., 3.9488e-05, 4.5002e-05, 4.9442e-05], [ 2.6536e-04, 2.3797e-05, 1.4089e-05, ..., 1.0502e-04, 1.2076e-04, 1.3196e-04], ..., [-6.3515e-04, -4.6462e-05, -2.4155e-05, ..., -2.5392e-04, -2.9373e-04, -3.2043e-04], [ 5.5671e-05, 5.0776e-06, 3.2317e-06, ..., 2.2560e-05, 2.5466e-05, 2.7850e-05], [ 1.9002e-04, 2.3529e-05, 1.6689e-05, ..., 7.2062e-05, 8.2552e-05, 9.0301e-05]], device='cuda:0') Epoch 44, bias, value: tensor([-0.0314, 0.0445, -0.0594, -0.0199, 0.0491, -0.0369, 0.0577], device='cuda:0'), grad: tensor([ 1.7658e-05, 3.0422e-04, 8.1301e-04, 1.1760e-04, -1.9779e-03, 1.6999e-04, 5.5742e-04], device='cuda:0') 306 0.0032431258795932905 changing lr epoch 43, time 427.82, cls_loss 0.0039 cls_loss_mapping 0.0450 cls_loss_causal 0.6249 re_mapping 0.0474 re_causal 0.0502 /// teacc 86.54 lr 0.00303487 Epoch 45, weight, value: tensor([[ 0.2480, 0.2484, 0.2811, ..., -0.0155, -0.0035, -0.0148], [-0.0493, -0.0446, -0.0666, ..., 0.0534, 0.0555, 0.0480], [-0.0911, -0.0836, -0.0987, ..., 0.0401, 0.0227, 0.0161], ..., [-0.1105, -0.1136, -0.0709, ..., 0.0176, 0.0687, 0.0517], [-0.0208, -0.0185, -0.0425, ..., -0.1358, -0.1533, -0.1659], [ 0.0842, 0.0853, 0.0759, ..., -0.0268, -0.0191, 0.0049]], device='cuda:0'), grad: tensor([[ 3.8862e-04, 1.0622e-04, 1.0890e-04, ..., 1.5783e-04, 1.7059e-04, 1.8311e-04], [ 3.5614e-05, 1.0513e-05, 8.5682e-06, ..., 1.9521e-05, 2.2337e-05, 2.3574e-05], [-2.2268e-04, -3.7193e-05, -4.0859e-05, ..., -7.3195e-05, -7.8142e-05, -9.1970e-05], ..., [-3.6907e-04, -1.2648e-04, -1.1647e-04, ..., -1.8442e-04, -2.0373e-04, -2.0933e-04], [ 4.9770e-05, 1.2524e-05, 1.2390e-05, ..., 2.1741e-05, 2.4214e-05, 2.6256e-05], [ 3.8534e-05, 1.1489e-05, 9.8050e-06, ..., 1.9610e-05, 2.1860e-05, 2.2933e-05]], device='cuda:0') Epoch 45, bias, value: tensor([-0.0337, 0.0451, -0.0600, -0.0192, 0.0491, -0.0350, 0.0576], device='cuda:0'), grad: tensor([ 8.6308e-04, 7.6950e-05, -5.8699e-04, 1.7869e-04, -7.2908e-04, 1.1402e-04, 8.3148e-05], device='cuda:0') 306 0.0030348748417303863 changing lr epoch 44, time 426.53, cls_loss 0.0031 cls_loss_mapping 0.0422 cls_loss_causal 0.6510 re_mapping 0.0467 re_causal 0.0495 /// teacc 86.06 lr 0.00283058 Epoch 46, weight, value: tensor([[ 0.2471, 0.2479, 0.2805, ..., -0.0157, -0.0037, -0.0151], [-0.0492, -0.0445, -0.0663, ..., 0.0532, 0.0554, 0.0479], [-0.0904, -0.0834, -0.0984, ..., 0.0401, 0.0227, 0.0162], ..., [-0.1096, -0.1130, -0.0704, ..., 0.0179, 0.0687, 0.0519], [-0.0199, -0.0182, -0.0422, ..., -0.1351, -0.1525, -0.1650], [ 0.0826, 0.0844, 0.0751, ..., -0.0269, -0.0193, 0.0046]], device='cuda:0'), grad: tensor([[ 2.2233e-05, 4.0531e-06, 3.6173e-06, ..., 7.3127e-06, 8.8513e-06, 9.5591e-06], [ 2.4855e-05, 4.6380e-06, 4.7497e-06, ..., 1.1235e-05, 1.2487e-05, 1.3188e-05], [-1.9088e-05, -3.9786e-06, -2.7493e-06, ..., 3.8594e-06, 1.6009e-06, -4.5635e-07], ..., [ 7.0989e-05, 8.3148e-06, 9.3728e-06, ..., 3.5614e-05, 3.9279e-05, 4.1932e-05], [ 1.1116e-04, 1.2904e-05, 1.5192e-05, ..., 6.3360e-05, 6.8307e-05, 7.0989e-05], [-3.1441e-05, -1.5251e-05, -1.2942e-05, ..., -6.5006e-07, -3.1181e-06, -3.5092e-06]], device='cuda:0') Epoch 46, bias, value: tensor([-0.0350, 0.0450, -0.0580, -0.0206, 0.0497, -0.0327, 0.0553], device='cuda:0'), grad: tensor([ 6.0409e-05, 6.4909e-05, -5.6982e-05, -5.0735e-04, 1.9813e-04, 3.0398e-04, -6.2764e-05], device='cuda:0') 306 0.0028305813044122124 changing lr epoch 45, time 428.43, cls_loss 0.0033 cls_loss_mapping 0.0372 cls_loss_causal 0.6389 re_mapping 0.0461 re_causal 0.0489 /// teacc 86.06 lr 0.00263066 Epoch 47, weight, value: tensor([[ 0.2468, 0.2476, 0.2801, ..., -0.0156, -0.0037, -0.0151], [-0.0488, -0.0443, -0.0660, ..., 0.0530, 0.0552, 0.0478], [-0.0904, -0.0833, -0.0983, ..., 0.0398, 0.0225, 0.0160], ..., [-0.1086, -0.1124, -0.0699, ..., 0.0182, 0.0688, 0.0521], [-0.0199, -0.0184, -0.0423, ..., -0.1346, -0.1520, -0.1645], [ 0.0815, 0.0837, 0.0744, ..., -0.0270, -0.0195, 0.0043]], device='cuda:0'), grad: tensor([[ 2.9951e-05, -2.6792e-05, -3.1501e-05, ..., 8.2031e-06, 8.5086e-06, 1.1154e-05], [-1.4961e-04, -6.3956e-05, -6.0052e-05, ..., -5.1737e-05, -6.2943e-05, -6.9201e-05], [-9.4604e-04, -1.0973e-04, -5.1677e-05, ..., -1.7416e-04, -1.8346e-04, -2.0885e-04], ..., [ 3.8338e-04, 8.6546e-05, 6.5982e-05, ..., 9.0778e-05, 1.0258e-04, 1.1402e-04], [ 4.9114e-04, 6.0260e-05, 3.3230e-05, ..., 8.3089e-05, 8.5413e-05, 9.7692e-05], [ 1.1992e-04, 3.7611e-05, 3.2216e-05, ..., 2.8685e-05, 3.3110e-05, 3.6567e-05]], device='cuda:0') Epoch 47, bias, value: tensor([-0.0350, 0.0456, -0.0582, -0.0206, 0.0506, -0.0320, 0.0536], device='cuda:0'), grad: tensor([ 0.0002, -0.0003, -0.0027, 0.0002, 0.0010, 0.0014, 0.0003], device='cuda:0') 306 0.0026306566876350096 changing lr epoch 46, time 470.73, cls_loss 0.0037 cls_loss_mapping 0.0379 cls_loss_causal 0.6401 re_mapping 0.0456 re_causal 0.0483 /// teacc 86.06 lr 0.00243550 Epoch 48, weight, value: tensor([[ 0.2481, 0.2483, 0.2807, ..., -0.0153, -0.0034, -0.0146], [-0.0491, -0.0443, -0.0660, ..., 0.0528, 0.0550, 0.0475], [-0.0905, -0.0833, -0.0982, ..., 0.0396, 0.0224, 0.0159], ..., [-0.1091, -0.1123, -0.0700, ..., 0.0180, 0.0684, 0.0517], [-0.0202, -0.0186, -0.0425, ..., -0.1343, -0.1516, -0.1641], [ 0.0814, 0.0831, 0.0739, ..., -0.0267, -0.0193, 0.0045]], device='cuda:0'), grad: tensor([[-6.1356e-06, -1.0031e-04, -1.0329e-04, ..., -3.7737e-06, 4.6641e-06, -4.1816e-07], [ 1.4558e-05, 8.7991e-06, 8.2701e-06, ..., 1.1362e-07, 1.1306e-06, 1.6019e-06], [-2.8872e-04, -2.8953e-05, -2.7329e-05, ..., -4.5776e-05, -5.9903e-05, -6.5863e-05], ..., [ 1.9741e-04, 9.6142e-05, 9.7275e-05, ..., 3.3945e-05, 3.6329e-05, 4.4256e-05], [ 5.7846e-05, 1.5557e-05, 1.5691e-05, ..., 1.1273e-05, 1.2152e-05, 1.4022e-05], [ 1.4775e-05, 4.5337e-06, 5.0813e-06, ..., 3.1181e-06, 4.2319e-06, 4.7013e-06]], device='cuda:0') Epoch 48, bias, value: tensor([-0.0331, 0.0446, -0.0586, -0.0211, 0.0488, -0.0323, 0.0554], device='cuda:0'), grad: tensor([ 2.4533e-04, 1.8641e-05, -8.0538e-04, 1.8939e-05, 3.4881e-04, 1.3614e-04, 3.7193e-05], device='cuda:0') 306 0.0024355036129704724 changing lr epoch 47, time 429.21, cls_loss 0.0029 cls_loss_mapping 0.0359 cls_loss_causal 0.6213 re_mapping 0.0454 re_causal 0.0483 /// teacc 85.58 lr 0.00224552 Epoch 49, weight, value: tensor([[ 0.2480, 0.2482, 0.2805, ..., -0.0153, -0.0034, -0.0146], [-0.0491, -0.0442, -0.0658, ..., 0.0525, 0.0547, 0.0473], [-0.0902, -0.0832, -0.0981, ..., 0.0395, 0.0223, 0.0158], ..., [-0.1089, -0.1120, -0.0699, ..., 0.0180, 0.0682, 0.0516], [-0.0203, -0.0188, -0.0426, ..., -0.1339, -0.1512, -0.1637], [ 0.0814, 0.0828, 0.0736, ..., -0.0265, -0.0191, 0.0047]], device='cuda:0'), grad: tensor([[-2.2233e-04, -1.4102e-04, -1.4091e-04, ..., -3.0786e-05, -3.1620e-05, -3.5584e-05], [ 7.0989e-05, 1.6347e-05, 1.2673e-05, ..., 1.3053e-05, 1.4298e-05, 1.8820e-05], [ 9.7334e-05, 4.0352e-05, 3.7849e-05, ..., 1.6138e-05, 1.6898e-05, 2.0996e-05], ..., [ 1.5485e-04, 8.2076e-05, 8.0466e-05, ..., 2.3916e-05, 2.5123e-05, 2.9176e-05], [-1.0288e-04, -8.0243e-06, 7.8324e-07, ..., -2.0772e-05, -2.2560e-05, -3.1799e-05], [-3.6329e-05, 1.5283e-06, 2.0489e-06, ..., -9.4473e-06, -1.0073e-05, -1.0923e-05]], device='cuda:0') Epoch 49, bias, value: tensor([-0.0331, 0.0438, -0.0579, -0.0215, 0.0480, -0.0320, 0.0564], device='cuda:0'), grad: tensor([-0.0002, 0.0002, 0.0002, 0.0001, 0.0002, -0.0004, -0.0001], device='cuda:0') 306 0.00224551509273949 changing lr epoch 48, time 427.84, cls_loss 0.0030 cls_loss_mapping 0.0376 cls_loss_causal 0.6181 re_mapping 0.0457 re_causal 0.0488 /// teacc 87.98 lr 0.00206107 Epoch 50, weight, value: tensor([[ 0.2472, 0.2477, 0.2799, ..., -0.0154, -0.0036, -0.0148], [-0.0493, -0.0442, -0.0657, ..., 0.0523, 0.0545, 0.0471], [-0.0901, -0.0832, -0.0980, ..., 0.0394, 0.0223, 0.0158], ..., [-0.1086, -0.1115, -0.0695, ..., 0.0177, 0.0678, 0.0513], [-0.0209, -0.0190, -0.0428, ..., -0.1336, -0.1509, -0.1634], [ 0.0821, 0.0828, 0.0736, ..., -0.0262, -0.0188, 0.0050]], device='cuda:0'), grad: tensor([[ 1.4400e-04, 4.0770e-05, 3.5912e-05, ..., 4.9591e-05, 5.7667e-05, 5.9694e-05], [-2.9826e-04, -7.5936e-05, -6.6042e-05, ..., -1.1307e-04, -1.3459e-04, -1.3673e-04], [ 1.9684e-05, 1.6894e-06, 8.9593e-07, ..., 1.1876e-05, 1.4775e-05, 1.4365e-05], ..., [ 2.1636e-05, 5.9158e-06, 5.1521e-06, ..., 7.8231e-06, 9.0152e-06, 9.3728e-06], [ 7.5512e-06, -1.0384e-06, -7.2550e-07, ..., 5.0589e-06, 8.4639e-06, 7.0296e-06], [ 9.3520e-05, 2.5243e-05, 2.1860e-05, ..., 3.4422e-05, 3.9756e-05, 4.1187e-05]], device='cuda:0') Epoch 50, bias, value: tensor([-0.0340, 0.0430, -0.0577, -0.0202, 0.0473, -0.0331, 0.0585], device='cuda:0'), grad: tensor([ 3.0828e-04, -6.5708e-04, 5.1737e-05, 2.4900e-05, 4.6939e-05, 2.0772e-05, 2.0397e-04], device='cuda:0') 306 0.002061073738537637 changing lr epoch 49, time 427.43, cls_loss 0.0027 cls_loss_mapping 0.0358 cls_loss_causal 0.6541 re_mapping 0.0448 re_causal 0.0479 /// teacc 88.46 lr 0.00188255 Epoch 51, weight, value: tensor([[ 0.2476, 0.2479, 0.2800, ..., -0.0152, -0.0035, -0.0146], [-0.0483, -0.0439, -0.0654, ..., 0.0525, 0.0547, 0.0474], [-0.0902, -0.0832, -0.0979, ..., 0.0392, 0.0221, 0.0157], ..., [-0.1085, -0.1113, -0.0694, ..., 0.0177, 0.0677, 0.0511], [-0.0210, -0.0191, -0.0428, ..., -0.1333, -0.1506, -0.1629], [ 0.0808, 0.0820, 0.0729, ..., -0.0264, -0.0190, 0.0046]], device='cuda:0'), grad: tensor([[-7.3761e-06, -6.8903e-05, -6.1393e-05, ..., 3.7342e-05, 4.6521e-05, 4.6879e-05], [-4.7827e-04, -1.3089e-04, -1.4806e-04, ..., -1.6093e-04, -1.8084e-04, -1.7262e-04], [ 9.6321e-05, 4.4078e-05, 4.3571e-05, ..., 2.0713e-05, 2.2292e-05, 2.2203e-05], ..., [ 1.0103e-04, 4.2081e-05, 4.1813e-05, ..., 2.4498e-05, 2.7284e-05, 2.7090e-05], [-7.2420e-05, -5.0105e-06, -3.0901e-06, ..., -3.3885e-05, -3.8385e-05, -3.9697e-05], [ 3.3355e-04, 1.0943e-04, 1.1837e-04, ..., 1.0401e-04, 1.1396e-04, 1.0717e-04]], device='cuda:0') Epoch 51, bias, value: tensor([-0.0331, 0.0450, -0.0578, -0.0209, 0.0469, -0.0330, 0.0567], device='cuda:0'), grad: tensor([ 1.2612e-04, -9.6321e-04, 1.7142e-04, 6.3956e-05, 1.8120e-04, -2.7323e-04, 6.9284e-04], device='cuda:0') 306 0.0018825509907063344 changing lr epoch 50, time 426.95, cls_loss 0.0033 cls_loss_mapping 0.0344 cls_loss_causal 0.6313 re_mapping 0.0445 re_causal 0.0478 /// teacc 86.54 lr 0.00171031 Epoch 52, weight, value: tensor([[ 0.2476, 0.2479, 0.2798, ..., -0.0152, -0.0035, -0.0146], [-0.0481, -0.0439, -0.0653, ..., 0.0525, 0.0548, 0.0474], [-0.0902, -0.0832, -0.0979, ..., 0.0390, 0.0220, 0.0156], ..., [-0.1074, -0.1107, -0.0689, ..., 0.0179, 0.0678, 0.0512], [-0.0213, -0.0192, -0.0429, ..., -0.1330, -0.1503, -0.1626], [ 0.0802, 0.0816, 0.0724, ..., -0.0264, -0.0191, 0.0045]], device='cuda:0'), grad: tensor([[ 1.1891e-04, 3.1561e-05, 2.2739e-05, ..., 3.7551e-05, 2.8744e-05, 3.4332e-05], [ 2.6679e-04, 6.5207e-05, 4.4525e-05, ..., 7.9095e-05, 5.8681e-05, 7.3135e-05], [ 7.6234e-05, 1.8701e-05, 1.0729e-05, ..., 2.2352e-05, 1.4529e-05, 1.7613e-05], ..., [ 3.4004e-05, 5.2080e-06, 1.2452e-06, ..., 9.4548e-06, 6.0275e-06, 6.8285e-06], [-1.1311e-03, -2.8086e-04, -1.7083e-04, ..., -3.4928e-04, -2.4164e-04, -2.8658e-04], [ 3.1185e-04, 7.9513e-05, 4.2439e-05, ..., 1.0026e-04, 6.4373e-05, 7.1526e-05]], device='cuda:0') Epoch 52, bias, value: tensor([-0.0330, 0.0449, -0.0578, -0.0217, 0.0485, -0.0332, 0.0560], device='cuda:0'), grad: tensor([ 0.0003, 0.0008, 0.0002, 0.0009, 0.0001, -0.0033, 0.0009], device='cuda:0') 306 0.0017103063703014388 changing lr epoch 51, time 429.18, cls_loss 0.0033 cls_loss_mapping 0.0337 cls_loss_causal 0.6361 re_mapping 0.0436 re_causal 0.0467 /// teacc 83.65 lr 0.00154469 Epoch 53, weight, value: tensor([[ 0.2471, 0.2476, 0.2795, ..., -0.0152, -0.0035, -0.0147], [-0.0487, -0.0439, -0.0653, ..., 0.0522, 0.0545, 0.0471], [-0.0902, -0.0831, -0.0978, ..., 0.0388, 0.0219, 0.0155], ..., [-0.1067, -0.1102, -0.0686, ..., 0.0180, 0.0678, 0.0513], [-0.0216, -0.0193, -0.0429, ..., -0.1328, -0.1500, -0.1623], [ 0.0808, 0.0815, 0.0723, ..., -0.0261, -0.0189, 0.0047]], device='cuda:0'), grad: tensor([[-2.8858e-03, -1.5879e-03, -1.5993e-03, ..., -4.1032e-04, -5.5933e-04, -5.7220e-04], [ 2.7132e-04, 1.4031e-04, 1.3936e-04, ..., 5.5403e-05, 6.7294e-05, 6.7949e-05], [ 7.6151e-04, 4.0030e-04, 4.0221e-04, ..., 1.1492e-04, 1.5497e-04, 1.5962e-04], ..., [ 1.3828e-03, 7.1383e-04, 7.1812e-04, ..., 1.9825e-04, 2.7800e-04, 2.9016e-04], [-2.2149e-04, 8.4192e-06, 1.9401e-05, ..., -1.2338e-04, -1.3256e-04, -1.3959e-04], [ 5.5552e-04, 2.9540e-04, 2.9325e-04, ..., 1.1337e-04, 1.3423e-04, 1.3423e-04]], device='cuda:0') Epoch 53, bias, value: tensor([-0.0338, 0.0429, -0.0577, -0.0219, 0.0494, -0.0337, 0.0586], device='cuda:0'), grad: tensor([-0.0040, 0.0004, 0.0011, 0.0004, 0.0021, -0.0009, 0.0009], device='cuda:0') 306 0.0015446867550656784 changing lr epoch 52, time 430.64, cls_loss 0.0035 cls_loss_mapping 0.0352 cls_loss_causal 0.6378 re_mapping 0.0435 re_causal 0.0466 /// teacc 85.58 lr 0.00138603 Epoch 54, weight, value: tensor([[ 0.2474, 0.2477, 0.2796, ..., -0.0150, -0.0034, -0.0145], [-0.0488, -0.0439, -0.0652, ..., 0.0520, 0.0543, 0.0469], [-0.0901, -0.0832, -0.0979, ..., 0.0387, 0.0218, 0.0155], ..., [-0.1063, -0.1099, -0.0683, ..., 0.0181, 0.0678, 0.0514], [-0.0212, -0.0193, -0.0429, ..., -0.1324, -0.1496, -0.1619], [ 0.0799, 0.0810, 0.0718, ..., -0.0262, -0.0191, 0.0045]], device='cuda:0'), grad: tensor([[ 4.4417e-04, 1.6558e-04, 1.5855e-04, ..., 2.1517e-04, 2.1636e-04, 2.3675e-04], [ 9.7871e-05, 3.1441e-05, 2.6315e-05, ..., 5.4896e-05, 5.1409e-05, 5.7578e-05], [ 5.0247e-05, 2.3901e-05, 2.0757e-05, ..., 3.8058e-05, 3.3408e-05, 3.6001e-05], ..., [-2.1362e-04, -1.0002e-04, -1.0782e-04, ..., -8.4937e-05, -9.6142e-05, -1.0109e-04], [ 1.8752e-04, 5.8442e-05, 4.9949e-05, ..., 9.8109e-05, 9.3877e-05, 1.0526e-04], [ 1.3046e-03, 4.1962e-04, 3.4881e-04, ..., 7.3671e-04, 6.8760e-04, 7.7057e-04]], device='cuda:0') Epoch 54, bias, value: tensor([-0.0333, 0.0422, -0.0572, -0.0223, 0.0495, -0.0324, 0.0572], device='cuda:0'), grad: tensor([ 0.0009, 0.0002, 0.0001, -0.0045, -0.0004, 0.0004, 0.0031], device='cuda:0') 306 0.001386025680863044 changing lr epoch 53, time 427.28, cls_loss 0.0021 cls_loss_mapping 0.0327 cls_loss_causal 0.5944 re_mapping 0.0434 re_causal 0.0465 /// teacc 88.94 lr 0.00123464 Epoch 55, weight, value: tensor([[ 0.2475, 0.2476, 0.2795, ..., -0.0149, -0.0033, -0.0144], [-0.0489, -0.0439, -0.0652, ..., 0.0518, 0.0542, 0.0468], [-0.0900, -0.0831, -0.0978, ..., 0.0387, 0.0218, 0.0155], ..., [-0.1068, -0.1099, -0.0684, ..., 0.0180, 0.0676, 0.0511], [-0.0210, -0.0193, -0.0429, ..., -0.1321, -0.1493, -0.1615], [ 0.0799, 0.0808, 0.0717, ..., -0.0261, -0.0190, 0.0045]], device='cuda:0'), grad: tensor([[ 3.1680e-05, 1.2154e-06, -2.8219e-07, ..., 1.5661e-05, 1.7688e-05, 1.8716e-05], [ 2.3949e-04, 2.4959e-05, 1.8969e-05, ..., 1.3852e-04, 1.4877e-04, 1.5247e-04], [ 1.6904e-04, 3.0249e-05, 1.9073e-05, ..., 8.3625e-05, 9.3639e-05, 9.9063e-05], ..., [ 3.6716e-04, 5.6565e-05, 3.5137e-05, ..., 1.9920e-04, 2.1935e-04, 2.2936e-04], [ 4.8161e-05, 3.6024e-06, -6.8396e-06, ..., 5.0902e-05, 5.4926e-05, 5.8174e-05], [ 2.0003e-04, 3.6567e-05, 2.5868e-05, ..., 9.4771e-05, 1.0562e-04, 1.1134e-04]], device='cuda:0') Epoch 55, bias, value: tensor([-0.0327, 0.0418, -0.0569, -0.0223, 0.0481, -0.0316, 0.0574], device='cuda:0'), grad: tensor([ 9.8288e-05, 6.6566e-04, 4.4632e-04, -2.8362e-03, 9.7656e-04, 1.2803e-04, 5.2500e-04], device='cuda:0') 306 0.0012346426699819469 changing lr epoch 54, time 429.05, cls_loss 0.0017 cls_loss_mapping 0.0296 cls_loss_causal 0.6024 re_mapping 0.0432 re_causal 0.0464 /// teacc 87.50 lr 0.00109084 Epoch 56, weight, value: tensor([[ 0.2472, 0.2475, 0.2793, ..., -0.0149, -0.0033, -0.0144], [-0.0489, -0.0439, -0.0651, ..., 0.0517, 0.0541, 0.0467], [-0.0899, -0.0831, -0.0977, ..., 0.0386, 0.0218, 0.0155], ..., [-0.1066, -0.1098, -0.0684, ..., 0.0180, 0.0675, 0.0511], [-0.0209, -0.0193, -0.0429, ..., -0.1319, -0.1491, -0.1613], [ 0.0798, 0.0807, 0.0716, ..., -0.0261, -0.0190, 0.0045]], device='cuda:0'), grad: tensor([[-3.3474e-03, -2.1229e-03, -2.0885e-03, ..., -2.0981e-04, -3.1662e-04, -3.3784e-04], [ 2.4605e-04, 1.4138e-04, 1.3793e-04, ..., 3.1292e-05, 3.9667e-05, 4.2975e-05], [ 1.2693e-03, 7.1526e-04, 6.9714e-04, ..., 1.6069e-04, 2.0373e-04, 2.1863e-04], ..., [ 6.1178e-04, 3.5381e-04, 3.4547e-04, ..., 7.1764e-05, 9.2328e-05, 9.9242e-05], [ 6.0844e-04, 2.5725e-04, 2.4486e-04, ..., 1.1837e-04, 1.4007e-04, 1.4651e-04], [ 1.3056e-03, 7.2861e-04, 7.1001e-04, ..., 1.7309e-04, 2.1768e-04, 2.3365e-04]], device='cuda:0') Epoch 56, bias, value: tensor([-0.0329, 0.0416, -0.0569, -0.0224, 0.0480, -0.0312, 0.0576], device='cuda:0'), grad: tensor([-0.0040, 0.0004, 0.0019, -0.0025, 0.0009, 0.0013, 0.0020], device='cuda:0') 306 0.0010908425876598518 changing lr epoch 55, time 426.80, cls_loss 0.0022 cls_loss_mapping 0.0329 cls_loss_causal 0.6184 re_mapping 0.0427 re_causal 0.0458 /// teacc 86.06 lr 0.00095492 Epoch 57, weight, value: tensor([[ 0.2475, 0.2476, 0.2794, ..., -0.0149, -0.0033, -0.0143], [-0.0490, -0.0439, -0.0651, ..., 0.0516, 0.0539, 0.0466], [-0.0899, -0.0830, -0.0976, ..., 0.0385, 0.0217, 0.0154], ..., [-0.1065, -0.1096, -0.0683, ..., 0.0180, 0.0674, 0.0511], [-0.0207, -0.0193, -0.0429, ..., -0.1316, -0.1488, -0.1610], [ 0.0791, 0.0803, 0.0713, ..., -0.0262, -0.0191, 0.0043]], device='cuda:0'), grad: tensor([[ 4.2391e-04, 1.2326e-04, 1.0896e-04, ..., 1.5497e-04, 1.8597e-04, 2.0492e-04], [-1.5807e-04, -5.4806e-05, -4.8727e-05, ..., -3.1233e-05, -3.6001e-05, -4.9263e-05], [ 2.2995e-04, 7.9930e-05, 6.9261e-05, ..., 8.5711e-05, 1.0091e-04, 9.5487e-05], ..., [-5.3787e-04, -1.6129e-04, -1.4102e-04, ..., -2.2840e-04, -2.7609e-04, -2.7633e-04], [ 2.2441e-05, 1.0282e-05, 1.1332e-05, ..., 7.6592e-06, 1.0222e-05, 1.0461e-05], [ 9.7454e-06, 8.0746e-07, -2.0936e-06, ..., 9.2760e-06, 1.2159e-05, 1.1526e-05]], device='cuda:0') Epoch 57, bias, value: tensor([-0.0323, 0.0411, -0.0570, -0.0218, 0.0477, -0.0305, 0.0565], device='cuda:0'), grad: tensor([ 1.1177e-03, -3.0947e-04, 5.4264e-04, 2.3723e-05, -1.4601e-03, 4.5985e-05, 4.0382e-05], device='cuda:0') 306 0.000954915028125264 changing lr epoch 56, time 428.67, cls_loss 0.0014 cls_loss_mapping 0.0284 cls_loss_causal 0.5831 re_mapping 0.0422 re_causal 0.0451 /// teacc 89.42 lr 0.00082713 Epoch 58, weight, value: tensor([[ 0.2476, 0.2476, 0.2793, ..., -0.0148, -0.0033, -0.0143], [-0.0488, -0.0438, -0.0650, ..., 0.0515, 0.0539, 0.0466], [-0.0900, -0.0830, -0.0976, ..., 0.0384, 0.0216, 0.0153], ..., [-0.1065, -0.1095, -0.0683, ..., 0.0180, 0.0674, 0.0510], [-0.0208, -0.0194, -0.0429, ..., -0.1315, -0.1486, -0.1608], [ 0.0788, 0.0800, 0.0710, ..., -0.0262, -0.0191, 0.0043]], device='cuda:0'), grad: tensor([[ 1.0282e-05, -2.3227e-06, -4.9472e-06, ..., 5.3830e-06, 5.2825e-06, 6.8247e-06], [-1.4150e-04, -3.4779e-05, -2.1845e-05, ..., -4.9174e-05, -5.9396e-05, -6.2883e-05], [ 5.2303e-05, 1.5765e-05, 1.3009e-05, ..., 1.6302e-05, 1.9863e-05, 2.0280e-05], ..., [ 2.0757e-05, 5.6326e-06, 4.7274e-06, ..., 6.5118e-06, 7.2829e-06, 7.7114e-06], [-4.5560e-06, 1.7434e-06, 8.5402e-07, ..., -4.2468e-06, -4.0457e-06, -3.2615e-06], [ 5.4955e-05, 1.2606e-05, 7.1153e-06, ..., 2.2396e-05, 2.7567e-05, 2.8089e-05]], device='cuda:0') Epoch 58, bias, value: tensor([-0.0320, 0.0412, -0.0571, -0.0216, 0.0475, -0.0304, 0.0562], device='cuda:0'), grad: tensor([ 5.3912e-05, -3.7479e-04, 1.2082e-04, 2.0117e-05, 5.1171e-05, -1.5542e-05, 1.4389e-04], device='cuda:0') 306 0.0008271337313934874 changing lr epoch 57, time 423.71, cls_loss 0.0016 cls_loss_mapping 0.0269 cls_loss_causal 0.5705 re_mapping 0.0421 re_causal 0.0451 /// teacc 87.98 lr 0.00070776 Epoch 59, weight, value: tensor([[ 0.2474, 0.2474, 0.2791, ..., -0.0149, -0.0033, -0.0143], [-0.0487, -0.0437, -0.0649, ..., 0.0515, 0.0538, 0.0465], [-0.0899, -0.0830, -0.0975, ..., 0.0383, 0.0216, 0.0153], ..., [-0.1063, -0.1094, -0.0682, ..., 0.0181, 0.0674, 0.0510], [-0.0207, -0.0194, -0.0429, ..., -0.1313, -0.1485, -0.1606], [ 0.0785, 0.0799, 0.0709, ..., -0.0262, -0.0192, 0.0042]], device='cuda:0'), grad: tensor([[-9.9182e-05, -6.8128e-05, -6.8247e-05, ..., -9.6485e-06, -9.5665e-06, -1.1303e-05], [-1.1706e-04, -3.6120e-05, -3.3170e-05, ..., -3.1769e-05, -3.9518e-05, -4.3839e-05], [ 1.0276e-04, 3.3408e-05, 3.0905e-05, ..., 3.1918e-05, 3.6389e-05, 3.8296e-05], ..., [-8.3864e-05, 3.0044e-06, 7.2084e-06, ..., -4.6194e-05, -5.0426e-05, -5.1051e-05], [ 8.7619e-05, 2.3827e-05, 2.1413e-05, ..., 3.0056e-05, 3.3259e-05, 3.5048e-05], [ 8.0526e-05, 3.5167e-05, 3.3796e-05, ..., 1.6659e-05, 1.9476e-05, 2.1666e-05]], device='cuda:0') Epoch 59, bias, value: tensor([-0.0322, 0.0413, -0.0569, -0.0217, 0.0476, -0.0303, 0.0558], device='cuda:0'), grad: tensor([-1.0520e-04, -2.8086e-04, 2.3329e-04, 6.8605e-05, -2.8443e-04, 2.1255e-04, 1.5569e-04], device='cuda:0') 306 0.00070775603199067 changing lr epoch 58, time 421.70, cls_loss 0.0019 cls_loss_mapping 0.0311 cls_loss_causal 0.5978 re_mapping 0.0416 re_causal 0.0446 /// teacc 88.46 lr 0.00059702 Epoch 60, weight, value: tensor([[ 0.2471, 0.2473, 0.2789, ..., -0.0149, -0.0034, -0.0144], [-0.0483, -0.0436, -0.0647, ..., 0.0515, 0.0539, 0.0466], [-0.0900, -0.0830, -0.0975, ..., 0.0382, 0.0215, 0.0152], ..., [-0.1060, -0.1092, -0.0680, ..., 0.0181, 0.0674, 0.0510], [-0.0210, -0.0194, -0.0429, ..., -0.1313, -0.1484, -0.1605], [ 0.0782, 0.0797, 0.0707, ..., -0.0263, -0.0192, 0.0042]], device='cuda:0'), grad: tensor([[ 9.6798e-05, 3.5226e-05, 3.2395e-05, ..., 2.2635e-05, 2.4319e-05, 2.8834e-05], [-1.2958e-04, -2.9504e-05, -2.5034e-05, ..., -5.7399e-05, -6.4015e-05, -7.1168e-05], [ 2.7716e-05, 7.5512e-06, 6.8322e-06, ..., 8.3372e-06, 8.8662e-06, 1.0163e-05], ..., [ 1.3828e-05, 3.9861e-06, 3.5781e-06, ..., 7.3947e-06, 5.8748e-06, 6.3442e-06], [-9.2566e-05, -2.2456e-05, -2.1636e-05, ..., -2.6435e-05, -1.8954e-05, -2.1636e-05], [-9.8124e-06, -1.6779e-05, -1.5572e-05, ..., 9.1493e-06, 7.4469e-06, 6.7241e-06]], device='cuda:0') Epoch 60, bias, value: tensor([-0.0326, 0.0420, -0.0573, -0.0211, 0.0478, -0.0307, 0.0556], device='cuda:0'), grad: tensor([ 2.0659e-04, -3.7932e-04, 6.7472e-05, 2.5487e-04, 4.3511e-05, -2.1946e-04, 2.6584e-05], device='cuda:0') 306 0.0005970223407163104 changing lr epoch 59, time 419.85, cls_loss 0.0019 cls_loss_mapping 0.0281 cls_loss_causal 0.6170 re_mapping 0.0414 re_causal 0.0446 /// teacc 86.54 lr 0.00049516 Epoch 61, weight, value: tensor([[ 0.2471, 0.2472, 0.2789, ..., -0.0149, -0.0034, -0.0144], [-0.0482, -0.0435, -0.0646, ..., 0.0515, 0.0538, 0.0465], [-0.0900, -0.0830, -0.0975, ..., 0.0382, 0.0215, 0.0152], ..., [-0.1060, -0.1092, -0.0680, ..., 0.0181, 0.0673, 0.0510], [-0.0211, -0.0195, -0.0430, ..., -0.1312, -0.1483, -0.1604], [ 0.0781, 0.0796, 0.0706, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[ 3.0851e-04, 8.7559e-05, 8.5950e-05, ..., 8.6188e-05, 9.1970e-05, 9.9599e-05], [ 9.4950e-05, 2.9132e-05, 2.7582e-05, ..., 3.2783e-05, 3.2723e-05, 3.5316e-05], [ 2.3806e-04, 6.7055e-05, 6.4909e-05, ..., 6.5923e-05, 6.9439e-05, 7.5936e-05], ..., [-6.3276e-04, -1.7273e-04, -1.7107e-04, ..., -1.5140e-04, -1.6677e-04, -1.8263e-04], [ 4.5270e-05, 1.0677e-05, 1.3418e-05, ..., 2.3127e-05, 2.5794e-05, 2.4602e-05], [ 1.1426e-04, 3.2634e-05, 3.1173e-05, ..., 3.8356e-05, 4.0770e-05, 4.3154e-05]], device='cuda:0') Epoch 61, bias, value: tensor([-0.0325, 0.0420, -0.0571, -0.0209, 0.0476, -0.0309, 0.0554], device='cuda:0'), grad: tensor([ 6.9857e-04, 2.1350e-04, 5.4169e-04, -3.6597e-04, -1.4334e-03, 8.5652e-05, 2.5988e-04], device='cuda:0') 306 0.0004951556604879052 changing lr epoch 60, time 418.06, cls_loss 0.0022 cls_loss_mapping 0.0259 cls_loss_causal 0.6041 re_mapping 0.0412 re_causal 0.0443 /// teacc 86.54 lr 0.00040236 Epoch 62, weight, value: tensor([[ 0.2470, 0.2472, 0.2788, ..., -0.0149, -0.0034, -0.0144], [-0.0483, -0.0435, -0.0646, ..., 0.0514, 0.0538, 0.0465], [-0.0900, -0.0830, -0.0975, ..., 0.0382, 0.0215, 0.0152], ..., [-0.1060, -0.1091, -0.0680, ..., 0.0181, 0.0673, 0.0510], [-0.0210, -0.0195, -0.0430, ..., -0.1311, -0.1482, -0.1603], [ 0.0780, 0.0795, 0.0705, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[-1.7226e-04, -1.2827e-04, -1.2863e-04, ..., -2.8118e-05, -2.7418e-05, -3.1918e-05], [ 7.6115e-05, 2.7344e-05, 2.4825e-05, ..., 2.2009e-05, 1.9506e-05, 2.3991e-05], [-1.7416e-04, -4.3243e-05, -3.6567e-05, ..., -5.5730e-05, -6.6042e-05, -7.8082e-05], ..., [ 1.4949e-04, 5.6535e-05, 5.2422e-05, ..., 3.9339e-05, 4.3750e-05, 5.2392e-05], [ 3.4302e-05, 1.2308e-05, 1.1645e-05, ..., -4.5598e-06, 4.0010e-06, 5.4576e-06], [ 2.7195e-05, 5.8621e-05, 6.1750e-05, ..., 8.6352e-06, 8.2105e-06, 5.7817e-06]], device='cuda:0') Epoch 62, bias, value: tensor([-0.0326, 0.0417, -0.0571, -0.0207, 0.0475, -0.0307, 0.0555], device='cuda:0'), grad: tensor([-1.1986e-04, 1.8978e-04, -5.1594e-04, 1.7297e-04, 3.4547e-04, 1.6943e-05, -8.9169e-05], device='cuda:0') 306 0.00040236113724274745 changing lr epoch 61, time 417.60, cls_loss 0.0014 cls_loss_mapping 0.0258 cls_loss_causal 0.5725 re_mapping 0.0409 re_causal 0.0440 /// teacc 87.50 lr 0.00031883 Epoch 63, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0144], [-0.0483, -0.0435, -0.0645, ..., 0.0514, 0.0538, 0.0465], [-0.0899, -0.0830, -0.0975, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1060, -0.1090, -0.0679, ..., 0.0181, 0.0672, 0.0509], [-0.0210, -0.0195, -0.0430, ..., -0.1310, -0.1482, -0.1602], [ 0.0780, 0.0794, 0.0704, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[-3.8743e-06, -7.7784e-06, -8.5309e-06, ..., 9.4250e-07, 1.3597e-06, 1.5721e-06], [ 1.4007e-05, 4.0494e-06, 3.5055e-06, ..., 4.4778e-06, 4.9956e-06, 5.0962e-06], [-1.5117e-05, -8.3074e-07, 8.9593e-07, ..., -2.8815e-06, -3.7625e-06, -4.4890e-06], ..., [ 3.1926e-06, 2.3656e-07, -2.8312e-07, ..., 2.5332e-07, 3.2783e-07, 6.7800e-07], [-1.6674e-05, -3.1460e-06, -2.5742e-06, ..., -6.2101e-06, -7.0110e-06, -7.0184e-06], [ 1.7866e-05, 6.9812e-06, 6.5118e-06, ..., 4.3884e-06, 4.9882e-06, 5.0478e-06]], device='cuda:0') Epoch 63, bias, value: tensor([-0.0326, 0.0417, -0.0570, -0.0206, 0.0473, -0.0306, 0.0555], device='cuda:0'), grad: tensor([ 9.9242e-06, 3.3975e-05, -4.4495e-05, 3.4831e-07, 9.3728e-06, -4.6074e-05, 3.7163e-05], device='cuda:0') 306 0.00031882564680131423 changing lr epoch 62, time 418.99, cls_loss 0.0020 cls_loss_mapping 0.0323 cls_loss_causal 0.5851 re_mapping 0.0406 re_causal 0.0436 /// teacc 88.46 lr 0.00024472 Epoch 64, weight, value: tensor([[ 0.2469, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0144], [-0.0482, -0.0434, -0.0645, ..., 0.0514, 0.0538, 0.0465], [-0.0899, -0.0829, -0.0975, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1060, -0.1090, -0.0679, ..., 0.0181, 0.0672, 0.0509], [-0.0210, -0.0195, -0.0430, ..., -0.1310, -0.1481, -0.1602], [ 0.0778, 0.0793, 0.0704, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[ 1.2884e-03, 1.9336e-04, 1.0115e-04, ..., 3.2020e-04, 3.5286e-04, 3.9339e-04], [ 6.8521e-04, 1.0777e-04, 5.8651e-05, ..., 1.6892e-04, 1.8585e-04, 2.0766e-04], [ 1.1377e-03, 1.6427e-04, 7.5102e-05, ..., 2.8133e-04, 3.1304e-04, 3.5000e-04], ..., [ 2.3975e-03, 3.2902e-04, 1.3816e-04, ..., 5.9986e-04, 6.6805e-04, 7.4720e-04], [-8.9111e-03, -1.2150e-03, -5.0020e-04, ..., -2.2259e-03, -2.4834e-03, -2.7771e-03], [ 3.9291e-04, 1.2249e-05, -3.6687e-05, ..., 1.0353e-04, 1.2201e-04, 1.3936e-04]], device='cuda:0') Epoch 64, bias, value: tensor([-0.0327, 0.0418, -0.0569, -0.0207, 0.0473, -0.0304, 0.0553], device='cuda:0'), grad: tensor([ 0.0043, 0.0023, 0.0039, 0.0104, 0.0083, -0.0308, 0.0017], device='cuda:0') 306 0.0002447174185242325 changing lr epoch 63, time 414.73, cls_loss 0.0017 cls_loss_mapping 0.0244 cls_loss_causal 0.6037 re_mapping 0.0406 re_causal 0.0438 /// teacc 86.06 lr 0.00018019 Epoch 65, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0144], [-0.0482, -0.0434, -0.0645, ..., 0.0514, 0.0538, 0.0465], [-0.0899, -0.0829, -0.0975, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1059, -0.1090, -0.0679, ..., 0.0181, 0.0672, 0.0509], [-0.0209, -0.0195, -0.0429, ..., -0.1309, -0.1480, -0.1601], [ 0.0777, 0.0792, 0.0703, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[-4.4964e-06, -1.0490e-05, -1.0796e-05, ..., 1.2871e-06, 1.4622e-06, 2.0340e-06], [ 7.4387e-05, 1.5885e-05, 1.3448e-05, ..., 1.7881e-05, 1.8552e-05, 2.2933e-05], [ 9.9063e-05, 2.2203e-05, 1.9148e-05, ..., 2.7061e-05, 2.4825e-05, 3.1203e-05], ..., [ 2.1532e-05, 5.4277e-06, 4.3362e-06, ..., -1.9260e-06, -1.6876e-06, -8.2701e-07], [ 1.1361e-04, 2.1651e-05, 1.7896e-05, ..., 3.0607e-05, 2.9683e-05, 3.7044e-05], [-7.2420e-05, -2.1517e-05, -1.9461e-05, ..., -1.2890e-05, -1.8656e-05, -2.1428e-05]], device='cuda:0') Epoch 65, bias, value: tensor([-0.0326, 0.0418, -0.0570, -0.0208, 0.0472, -0.0302, 0.0552], device='cuda:0'), grad: tensor([ 2.3663e-05, 1.9288e-04, 2.5082e-04, -6.5517e-04, 5.3644e-05, 3.0255e-04, -1.6916e-04], device='cuda:0') 306 0.0001801856965207339 changing lr epoch 64, time 413.22, cls_loss 0.0016 cls_loss_mapping 0.0248 cls_loss_causal 0.6161 re_mapping 0.0405 re_causal 0.0437 /// teacc 88.94 lr 0.00012536 Epoch 66, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0144], [-0.0481, -0.0434, -0.0645, ..., 0.0513, 0.0538, 0.0465], [-0.0898, -0.0829, -0.0974, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1059, -0.1090, -0.0679, ..., 0.0180, 0.0672, 0.0509], [-0.0209, -0.0195, -0.0430, ..., -0.1309, -0.1480, -0.1601], [ 0.0777, 0.0792, 0.0703, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[-7.0512e-05, -4.7058e-05, -5.0157e-05, ..., -8.2478e-06, -8.2850e-06, -9.5665e-06], [ 2.0787e-05, 9.5218e-06, 9.8273e-06, ..., 3.1665e-06, 3.2280e-06, 4.0270e-06], [-4.8801e-06, 1.4380e-06, 2.3991e-06, ..., -1.5441e-06, -1.7043e-06, -3.0026e-06], ..., [ 7.8917e-05, 3.0667e-05, 3.0696e-05, ..., 1.4104e-05, 1.4886e-05, 1.8507e-05], [ 4.3333e-05, 1.1683e-05, 1.1191e-05, ..., 9.6634e-06, 9.5069e-06, 1.1683e-05], [-9.7632e-05, -2.0340e-05, -1.8418e-05, ..., -2.1979e-05, -2.2665e-05, -2.7910e-05]], device='cuda:0') Epoch 66, bias, value: tensor([-0.0326, 0.0419, -0.0568, -0.0208, 0.0471, -0.0302, 0.0551], device='cuda:0'), grad: tensor([-8.1718e-05, 3.6746e-05, -2.0251e-05, 5.1826e-05, 1.5628e-04, 1.0252e-04, -2.4605e-04], device='cuda:0') 306 0.000125360439090882 changing lr epoch 65, time 415.02, cls_loss 0.0015 cls_loss_mapping 0.0254 cls_loss_causal 0.5788 re_mapping 0.0404 re_causal 0.0434 /// teacc 87.50 lr 0.00008035 Epoch 67, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0144], [-0.0481, -0.0434, -0.0645, ..., 0.0513, 0.0537, 0.0465], [-0.0898, -0.0829, -0.0974, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1060, -0.1090, -0.0679, ..., 0.0180, 0.0672, 0.0509], [-0.0209, -0.0195, -0.0430, ..., -0.1309, -0.1480, -0.1600], [ 0.0777, 0.0792, 0.0702, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[ 4.1676e-04, 1.7440e-04, 1.6701e-04, ..., 7.9632e-05, 8.6963e-05, 9.7990e-05], [ 1.0794e-04, 4.8518e-05, 4.6700e-05, ..., 2.0698e-05, 2.1353e-05, 2.3842e-05], [ 1.4484e-04, 6.3241e-05, 6.0648e-05, ..., 2.8014e-05, 3.1024e-05, 3.4124e-05], ..., [-2.1820e-03, -9.4128e-04, -9.0265e-04, ..., -4.2129e-04, -4.6349e-04, -5.1355e-04], [ 8.1873e-04, 3.5405e-04, 3.3951e-04, ..., 1.5831e-04, 1.7440e-04, 1.9288e-04], [ 6.5470e-04, 2.8372e-04, 2.7227e-04, ..., 1.2708e-04, 1.4138e-04, 1.5545e-04]], device='cuda:0') Epoch 67, bias, value: tensor([-0.0325, 0.0418, -0.0568, -0.0208, 0.0470, -0.0302, 0.0551], device='cuda:0'), grad: tensor([ 9.2602e-04, 2.3615e-04, 3.1257e-04, 8.7500e-05, -4.7569e-03, 1.7815e-03, 1.4172e-03], device='cuda:0') 306 8.03520570068517e-05 changing lr epoch 66, time 411.65, cls_loss 0.0014 cls_loss_mapping 0.0245 cls_loss_causal 0.5859 re_mapping 0.0403 re_causal 0.0433 /// teacc 88.46 lr 0.00004525 Epoch 68, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0143], [-0.0482, -0.0434, -0.0645, ..., 0.0513, 0.0537, 0.0464], [-0.0898, -0.0829, -0.0974, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1060, -0.1089, -0.0679, ..., 0.0180, 0.0671, 0.0509], [-0.0209, -0.0195, -0.0430, ..., -0.1309, -0.1480, -0.1600], [ 0.0776, 0.0792, 0.0702, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[ 6.9320e-05, 2.5973e-05, 2.5213e-05, ..., 1.3977e-05, 1.4402e-05, 1.9848e-05], [ 9.1970e-05, 1.9833e-05, 1.6674e-05, ..., 5.2363e-05, 5.4508e-05, 6.0260e-05], [ 1.2094e-04, 4.5180e-05, 4.3631e-05, ..., 2.4304e-05, 2.4974e-05, 3.4660e-05], ..., [ 1.0721e-05, 3.2503e-06, 2.9113e-06, ..., 4.1090e-06, 4.3735e-06, 5.1148e-06], [-1.9260e-06, 8.3819e-08, 8.9779e-07, ..., 8.9593e-07, 7.9162e-07, 1.0133e-06], [-2.2829e-04, -8.7917e-05, -8.5890e-05, ..., -4.2439e-05, -4.3660e-05, -6.2108e-05]], device='cuda:0') Epoch 68, bias, value: tensor([-0.0325, 0.0418, -0.0568, -0.0208, 0.0470, -0.0302, 0.0551], device='cuda:0'), grad: tensor([ 1.3232e-04, 2.3067e-04, 2.3127e-04, -1.8311e-04, 2.3663e-05, -9.8050e-06, -4.2439e-04], device='cuda:0') 306 4.5251191160326525e-05 changing lr epoch 67, time 411.01, cls_loss 0.0014 cls_loss_mapping 0.0238 cls_loss_causal 0.6263 re_mapping 0.0403 re_causal 0.0434 /// teacc 86.06 lr 0.00002013 Epoch 69, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0143], [-0.0482, -0.0434, -0.0645, ..., 0.0513, 0.0537, 0.0464], [-0.0898, -0.0829, -0.0974, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1060, -0.1089, -0.0679, ..., 0.0180, 0.0671, 0.0509], [-0.0209, -0.0195, -0.0430, ..., -0.1309, -0.1479, -0.1600], [ 0.0776, 0.0792, 0.0702, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[-1.0309e-03, -6.2466e-04, -6.1512e-04, ..., -9.6560e-05, -1.5414e-04, -1.9073e-04], [ 5.1165e-04, 2.3520e-04, 2.2519e-04, ..., 1.5378e-04, 1.6558e-04, 1.8859e-04], [ 3.4738e-04, 1.6952e-04, 1.6379e-04, ..., 7.6950e-05, 8.8632e-05, 1.0359e-04], ..., [-6.6471e-04, -1.8024e-04, -1.5926e-04, ..., -3.7694e-04, -3.6836e-04, -4.0412e-04], [ 2.2173e-04, 7.6950e-05, 7.1943e-05, ..., 7.9215e-05, 8.2910e-05, 9.3460e-05], [ 4.4799e-04, 2.6631e-04, 2.6083e-04, ..., 8.7559e-05, 1.0908e-04, 1.2445e-04]], device='cuda:0') Epoch 69, bias, value: tensor([-0.0325, 0.0418, -0.0568, -0.0208, 0.0470, -0.0301, 0.0551], device='cuda:0'), grad: tensor([-0.0012, 0.0010, 0.0006, 0.0004, -0.0018, 0.0005, 0.0005], device='cuda:0') 306 2.0128530023804673e-05 changing lr epoch 68, time 410.85, cls_loss 0.0015 cls_loss_mapping 0.0246 cls_loss_causal 0.5761 re_mapping 0.0403 re_causal 0.0433 /// teacc 89.42 lr 0.00000503 Epoch 70, weight, value: tensor([[ 0.2470, 0.2471, 0.2787, ..., -0.0149, -0.0034, -0.0143], [-0.0482, -0.0434, -0.0645, ..., 0.0513, 0.0537, 0.0464], [-0.0898, -0.0829, -0.0974, ..., 0.0381, 0.0214, 0.0152], ..., [-0.1059, -0.1089, -0.0679, ..., 0.0180, 0.0671, 0.0509], [-0.0209, -0.0195, -0.0430, ..., -0.1309, -0.1479, -0.1600], [ 0.0776, 0.0791, 0.0702, ..., -0.0262, -0.0192, 0.0041]], device='cuda:0'), grad: tensor([[ 3.0696e-05, 8.4490e-06, 8.9705e-06, ..., 7.9796e-06, 8.0392e-06, 9.2313e-06], [ 2.1577e-05, 5.4277e-06, 5.2042e-06, ..., 7.7486e-06, 7.9274e-06, 8.6129e-06], [ 1.7853e-06, 1.0766e-06, 1.6764e-08, ..., 1.8803e-06, 2.0973e-06, 1.9260e-06], ..., [ 7.2084e-06, 2.0005e-06, 1.8878e-06, ..., 1.7714e-06, 1.9129e-06, 2.2277e-06], [-7.9811e-05, -2.2769e-05, -2.2128e-05, ..., -2.4080e-05, -2.4781e-05, -2.7478e-05], [ 1.5169e-05, 3.6657e-06, 3.6526e-06, ..., 5.2936e-06, 5.4389e-06, 5.9977e-06]], device='cuda:0') Epoch 70, bias, value: tensor([-0.0325, 0.0418, -0.0568, -0.0208, 0.0470, -0.0301, 0.0551], device='cuda:0'), grad: tensor([ 8.1778e-05, 6.2466e-05, 9.5889e-06, 1.1146e-05, 2.0608e-05, -2.3401e-04, 4.8488e-05], device='cuda:0') 306 5.034667293427056e-06 changing lr epoch 69, time 414.33, cls_loss 0.0018 cls_loss_mapping 0.0269 cls_loss_causal 0.6085 re_mapping 0.0402 re_causal 0.0433 /// teacc 88.46 lr 0.00000000 ---------------------saving last model at epoch 69---------------------------------------------------- /home/yuqian_fu {'gpu': '0', 'svroot': '/data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/saved-PACS//art_painting/CA_multiple_16fa_v2_ep70_lr0.01_cosine_base0.01_bs6_lamCa_1_lamRe1_adt4_cls1_EW2_70_rmTrue_rnTrue_str5_WithStyleAttackExp1', 'source_domain': 'art_painting', 'svpath': '/data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/saved-PACS//art_painting/CA_multiple_16fa_v2_ep70_lr0.01_cosine_base0.01_bs6_lamCa_1_lamRe1_adt4_cls1_EW2_70_rmTrue_rnTrue_str5_WithStyleAttackExp1/art_painting_16factor_best_test_check.csv', 'factor_num': 16, 'epoch': 'best', 'stride': 5, 'eval_mapping': False, 'network': 'resnet18'} -------------------------------------loading pretrain weights---------------------------------- loading weight of best randm: False stride: 5 loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best loading weight of best columns: ['art_painting', 'cartoon', 'photo', 'sketch'] /data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/data/PACS/art_painting_test.hdf5 torch.Size([2048, 3, 227, 227]) torch.Size([2048]) /data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/data/PACS/cartoon_test.hdf5 torch.Size([2344, 3, 227, 227]) torch.Size([2344]) /data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/data/PACS/photo_test.hdf5 torch.Size([1670, 3, 227, 227]) torch.Size([1670]) /data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/data/PACS/sketch_test.hdf5 torch.Size([3929, 3, 227, 227]) torch.Size([3929]) art_painting cartoon photo sketch Avg w/o do (original x) 99.023438 69.795222 90.778443 73.199287 77.924317 art_painting cartoon photo sketch Avg do 99.023438 72.1843 92.035928 72.410283 78.876837