| /home/yuqian_fu |
| here1 |
| here2 |
| {'gpu': '0', 'data': 'mnist', 'ntr': None, 'translate': None, 'autoaug': 'CA_multiple', 'n': 3, 'stride': 3, 'factor_num': 14, 'epochs': 250, 'nbatch': 100, 'batchsize': 32, 'lr': 0.0001, 'lr_scheduler': 'Step', 'svroot': '/data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/saved-digit/CA_multiple_14fa_all_ep250_lr1e-4_lr_schedulerStep0.8_bs32_lamCa_1_lamRe_1_cls1_adt2_EW2_100_rmTrue_rnTrue_str3_WithStyleAttackExp1_epoch250', 'clsadapt': True, 'lambda_causal': 1.0, 'lambda_re': 1.0, 'randm': True, 'randn': True, 'network': 'resnet18'} |
| stride: 3 |
| --------------------------CA_multiple-------------------------- |
| ---------------------------14 factors----------------- |
| randm: True |
| randn: True |
| n: 3 |
| randm: False |
| Epoch 1, weight, value: tensor([[-0.0118, 0.0228, -0.0024, ..., -0.0263, 0.0294, 0.0225], |
| [ 0.0117, 0.0125, -0.0312, ..., 0.0006, -0.0133, 0.0132], |
| [ 0.0204, -0.0145, -0.0123, ..., -0.0209, 0.0127, -0.0073], |
| ..., |
| [-0.0015, -0.0013, 0.0121, ..., 0.0043, 0.0199, 0.0109], |
| [ 0.0096, 0.0208, -0.0142, ..., 0.0160, -0.0174, 0.0288], |
| [-0.0055, -0.0161, 0.0242, ..., -0.0260, -0.0230, 0.0003]], |
| device='cuda:0'), grad: None |
| Epoch 1, bias, value: tensor([ 0.0002, 0.0236, 0.0062, 0.0051, 0.0040, -0.0094, 0.0250, -0.0272, |
| 0.0250, 0.0046], device='cuda:0'), grad: None |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 0---------------------------------------------------- |
| epoch 0, time 221.75, cls_loss 1.6842 cls_loss_mapping 2.0049 cls_loss_causal 2.2496 re_mapping 0.0767 re_causal 0.0774 /// teacc 84.37 lr 0.00010000 |
| Epoch 2, weight, value: tensor([[-0.0126, 0.0228, 0.0004, ..., -0.0319, 0.0314, 0.0219], |
| [ 0.0086, 0.0125, -0.0421, ..., 0.0074, -0.0105, 0.0126], |
| [ 0.0192, -0.0145, -0.0187, ..., -0.0247, 0.0152, -0.0079], |
| ..., |
| [ 0.0011, -0.0013, 0.0107, ..., 0.0045, 0.0185, 0.0114], |
| [ 0.0097, 0.0208, -0.0128, ..., 0.0117, -0.0165, 0.0295], |
| [-0.0047, -0.0161, 0.0288, ..., -0.0276, -0.0288, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 0.0072, 0.0000, 0.0089, ..., 0.0002, 0.0204, 0.0000], |
| [ 0.0124, 0.0000, 0.0125, ..., 0.0015, 0.0347, 0.0000], |
| [-0.0183, 0.0000, 0.0107, ..., 0.0003, -0.0353, 0.0000], |
| ..., |
| [-0.0045, 0.0000, -0.0565, ..., -0.0074, -0.0013, 0.0000], |
| [-0.0344, 0.0000, -0.0058, ..., -0.0060, -0.0693, 0.0000], |
| [ 0.0173, 0.0000, 0.1266, ..., 0.0083, 0.0386, 0.0000]], |
| device='cuda:0') |
| Epoch 2, bias, value: tensor([-0.0037, 0.0263, 0.0055, 0.0059, 0.0041, -0.0085, 0.0241, -0.0262, |
| 0.0237, 0.0050], device='cuda:0'), grad: tensor([ 0.0261, 0.0339, -0.0185, 0.0468, -0.0235, -0.0387, -0.0398, -0.0343, |
| -0.0600, 0.1082], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 1---------------------------------------------------- |
| epoch 1, time 222.81, cls_loss 0.5751 cls_loss_mapping 0.8903 cls_loss_causal 1.9459 re_mapping 0.2095 re_causal 0.2474 /// teacc 90.15 lr 0.00010000 |
| Epoch 3, weight, value: tensor([[-0.0133, 0.0228, -0.0016, ..., -0.0345, 0.0322, 0.0219], |
| [ 0.0073, 0.0125, -0.0446, ..., 0.0099, -0.0113, 0.0126], |
| [ 0.0210, -0.0145, -0.0211, ..., -0.0259, 0.0167, -0.0079], |
| ..., |
| [ 0.0014, -0.0013, 0.0108, ..., 0.0048, 0.0192, 0.0114], |
| [ 0.0087, 0.0208, -0.0127, ..., 0.0093, -0.0153, 0.0295], |
| [-0.0047, -0.0161, 0.0294, ..., -0.0308, -0.0306, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 0.0004, 0.0000, 0.0062, ..., 0.0003, 0.0099, 0.0000], |
| [ 0.0026, 0.0000, 0.0009, ..., 0.0002, 0.0028, 0.0000], |
| [-0.0282, 0.0000, -0.0045, ..., 0.0004, -0.0375, 0.0000], |
| ..., |
| [-0.0017, 0.0000, 0.0044, ..., 0.0007, -0.0014, 0.0000], |
| [ 0.0175, 0.0000, 0.0013, ..., 0.0011, 0.0173, 0.0000], |
| [ 0.0029, 0.0000, 0.0058, ..., 0.0027, 0.0027, 0.0000]], |
| device='cuda:0') |
| Epoch 3, bias, value: tensor([-0.0044, 0.0265, 0.0052, 0.0057, 0.0041, -0.0070, 0.0236, -0.0264, |
| 0.0234, 0.0052], device='cuda:0'), grad: tensor([ 0.0199, 0.0046, -0.0508, -0.0047, 0.0078, -0.0029, -0.0100, 0.0019, |
| 0.0240, 0.0102], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 2---------------------------------------------------- |
| epoch 2, time 220.88, cls_loss 0.3477 cls_loss_mapping 0.5253 cls_loss_causal 1.7447 re_mapping 0.1589 re_causal 0.2344 /// teacc 93.42 lr 0.00010000 |
| Epoch 4, weight, value: tensor([[-0.0136, 0.0228, -0.0035, ..., -0.0363, 0.0330, 0.0219], |
| [ 0.0068, 0.0125, -0.0460, ..., 0.0108, -0.0103, 0.0126], |
| [ 0.0225, -0.0145, -0.0216, ..., -0.0257, 0.0167, -0.0079], |
| ..., |
| [ 0.0025, -0.0013, 0.0104, ..., 0.0053, 0.0195, 0.0114], |
| [ 0.0083, 0.0208, -0.0141, ..., 0.0069, -0.0150, 0.0295], |
| [-0.0057, -0.0161, 0.0304, ..., -0.0329, -0.0310, -0.0004]], |
| device='cuda:0'), grad: tensor([[-0.0008, 0.0000, 0.0006, ..., 0.0000, -0.0026, 0.0000], |
| [ 0.0009, 0.0000, 0.0005, ..., 0.0000, -0.0022, 0.0000], |
| [-0.0275, 0.0000, 0.0009, ..., 0.0000, -0.0058, 0.0000], |
| ..., |
| [ 0.0023, 0.0000, 0.0018, ..., 0.0000, 0.0020, 0.0000], |
| [ 0.0611, 0.0000, 0.0223, ..., 0.0000, 0.0236, 0.0000], |
| [ 0.0045, 0.0000, -0.0014, ..., 0.0000, 0.0022, 0.0000]], |
| device='cuda:0') |
| Epoch 4, bias, value: tensor([-0.0041, 0.0269, 0.0052, 0.0057, 0.0037, -0.0064, 0.0235, -0.0267, |
| 0.0230, 0.0052], device='cuda:0'), grad: tensor([-0.0077, -0.0008, -0.0145, -0.0323, 0.0042, -0.0293, 0.0009, 0.0038, |
| 0.0704, 0.0052], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 3---------------------------------------------------- |
| epoch 3, time 218.72, cls_loss 0.2539 cls_loss_mapping 0.3669 cls_loss_causal 1.5234 re_mapping 0.1287 re_causal 0.2210 /// teacc 94.90 lr 0.00010000 |
| Epoch 5, weight, value: tensor([[-0.0134, 0.0228, -0.0046, ..., -0.0408, 0.0336, 0.0219], |
| [ 0.0065, 0.0125, -0.0470, ..., 0.0115, -0.0087, 0.0126], |
| [ 0.0238, -0.0145, -0.0223, ..., -0.0229, 0.0165, -0.0079], |
| ..., |
| [ 0.0047, -0.0013, 0.0098, ..., 0.0070, 0.0199, 0.0114], |
| [ 0.0074, 0.0208, -0.0146, ..., 0.0038, -0.0147, 0.0295], |
| [-0.0073, -0.0161, 0.0309, ..., -0.0338, -0.0308, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 2.4948e-03, 0.0000e+00, 2.3232e-03, ..., 5.9843e-05, |
| 8.5449e-03, 0.0000e+00], |
| [-3.3360e-03, 0.0000e+00, -6.3744e-03, ..., -1.0071e-02, |
| -2.1877e-03, 0.0000e+00], |
| [ 2.0355e-02, 0.0000e+00, 1.5106e-03, ..., 8.4114e-04, |
| 1.0719e-02, 0.0000e+00], |
| ..., |
| [ 1.0414e-02, 0.0000e+00, 7.0229e-03, ..., -1.8120e-04, |
| 6.8588e-03, 0.0000e+00], |
| [-4.2389e-02, 0.0000e+00, 1.3069e-02, ..., 2.7485e-03, |
| -1.3840e-02, 0.0000e+00], |
| [ 4.2953e-03, 0.0000e+00, -3.7781e-02, ..., 6.1703e-04, |
| -8.6594e-03, 0.0000e+00]], device='cuda:0') |
| Epoch 5, bias, value: tensor([-0.0039, 0.0276, 0.0055, 0.0055, 0.0038, -0.0067, 0.0232, -0.0267, |
| 0.0230, 0.0051], device='cuda:0'), grad: tensor([ 0.0161, -0.0300, 0.0190, 0.0335, -0.0035, -0.0140, -0.0066, 0.0149, |
| -0.0064, -0.0230], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 4---------------------------------------------------- |
| epoch 4, time 219.21, cls_loss 0.2005 cls_loss_mapping 0.2738 cls_loss_causal 1.4344 re_mapping 0.1058 re_causal 0.2003 /// teacc 95.39 lr 0.00010000 |
| Epoch 6, weight, value: tensor([[-0.0138, 0.0228, -0.0060, ..., -0.0444, 0.0334, 0.0219], |
| [ 0.0058, 0.0125, -0.0481, ..., 0.0117, -0.0080, 0.0126], |
| [ 0.0248, -0.0145, -0.0229, ..., -0.0204, 0.0163, -0.0079], |
| ..., |
| [ 0.0057, -0.0013, 0.0094, ..., 0.0076, 0.0200, 0.0114], |
| [ 0.0074, 0.0208, -0.0151, ..., 0.0024, -0.0143, 0.0295], |
| [-0.0078, -0.0161, 0.0315, ..., -0.0345, -0.0306, -0.0004]], |
| device='cuda:0'), grad: tensor([[-2.9697e-03, 0.0000e+00, 6.4507e-03, ..., 2.4348e-05, |
| -9.5177e-04, 0.0000e+00], |
| [ 1.1158e-03, 0.0000e+00, 4.9877e-04, ..., 7.8022e-05, |
| 9.6560e-04, 0.0000e+00], |
| [ 1.1053e-03, 0.0000e+00, 1.8187e-03, ..., 1.6415e-04, |
| -8.0795e-03, 0.0000e+00], |
| ..., |
| [-1.9119e-02, 0.0000e+00, -5.7793e-04, ..., 5.4866e-05, |
| -1.5533e-02, 0.0000e+00], |
| [ 2.9640e-03, 0.0000e+00, -6.6185e-03, ..., 3.2568e-04, |
| 4.6883e-03, 0.0000e+00], |
| [ 1.2794e-02, 0.0000e+00, 8.0338e-03, ..., 2.3575e-03, |
| 1.3672e-02, 0.0000e+00]], device='cuda:0') |
| Epoch 6, bias, value: tensor([-0.0039, 0.0276, 0.0055, 0.0052, 0.0041, -0.0066, 0.0229, -0.0268, |
| 0.0231, 0.0053], device='cuda:0'), grad: tensor([ 0.0018, 0.0014, -0.0032, 0.0058, -0.0147, 0.0065, 0.0040, -0.0266, |
| -0.0003, 0.0253], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 5---------------------------------------------------- |
| epoch 5, time 219.32, cls_loss 0.1731 cls_loss_mapping 0.2329 cls_loss_causal 1.3032 re_mapping 0.0907 re_causal 0.1857 /// teacc 95.98 lr 0.00010000 |
| Epoch 7, weight, value: tensor([[-0.0144, 0.0228, -0.0067, ..., -0.0481, 0.0336, 0.0219], |
| [ 0.0057, 0.0125, -0.0497, ..., 0.0112, -0.0073, 0.0126], |
| [ 0.0258, -0.0145, -0.0238, ..., -0.0200, 0.0160, -0.0079], |
| ..., |
| [ 0.0062, -0.0013, 0.0088, ..., 0.0078, 0.0200, 0.0114], |
| [ 0.0069, 0.0208, -0.0155, ..., 0.0013, -0.0143, 0.0295], |
| [-0.0090, -0.0161, 0.0321, ..., -0.0352, -0.0308, -0.0004]], |
| device='cuda:0'), grad: tensor([[-5.1041e-03, 0.0000e+00, 1.3676e-03, ..., 2.3358e-06, |
| -3.9597e-03, 0.0000e+00], |
| [ 1.2379e-03, 0.0000e+00, 9.1839e-04, ..., 1.4175e-06, |
| 6.3438e-03, 0.0000e+00], |
| [ 5.2166e-04, 0.0000e+00, 2.2125e-03, ..., 9.5833e-07, |
| -5.6763e-03, 0.0000e+00], |
| ..., |
| [ 1.5430e-03, 0.0000e+00, 1.4477e-03, ..., -1.9759e-05, |
| 4.8304e-04, 0.0000e+00], |
| [ 4.6120e-03, 0.0000e+00, 2.7122e-03, ..., 3.3388e-07, |
| 3.3855e-03, 0.0000e+00], |
| [ 1.3089e-04, 0.0000e+00, -1.5572e-02, ..., 1.0930e-05, |
| -2.2449e-03, 0.0000e+00]], device='cuda:0') |
| Epoch 7, bias, value: tensor([-0.0034, 0.0276, 0.0053, 0.0051, 0.0045, -0.0070, 0.0225, -0.0269, |
| 0.0232, 0.0054], device='cuda:0'), grad: tensor([-0.0089, 0.0094, -0.0060, -0.0251, 0.0048, 0.0266, -0.0009, 0.0030, |
| 0.0094, -0.0122], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 6---------------------------------------------------- |
| epoch 6, time 219.21, cls_loss 0.1524 cls_loss_mapping 0.2100 cls_loss_causal 1.2993 re_mapping 0.0762 re_causal 0.1680 /// teacc 96.07 lr 0.00010000 |
| Epoch 8, weight, value: tensor([[-0.0149, 0.0228, -0.0079, ..., -0.0489, 0.0333, 0.0219], |
| [ 0.0052, 0.0125, -0.0507, ..., 0.0111, -0.0067, 0.0126], |
| [ 0.0266, -0.0145, -0.0243, ..., -0.0201, 0.0152, -0.0079], |
| ..., |
| [ 0.0067, -0.0013, 0.0084, ..., 0.0079, 0.0200, 0.0114], |
| [ 0.0065, 0.0208, -0.0156, ..., 0.0013, -0.0144, 0.0295], |
| [-0.0094, -0.0161, 0.0328, ..., -0.0354, -0.0297, -0.0004]], |
| device='cuda:0'), grad: tensor([[-0.0003, 0.0000, 0.0005, ..., 0.0000, -0.0020, 0.0000], |
| [ 0.0006, 0.0000, -0.0210, ..., 0.0000, -0.0089, 0.0000], |
| [-0.0083, 0.0000, 0.0005, ..., 0.0000, -0.0036, 0.0000], |
| ..., |
| [ 0.0036, 0.0000, 0.0120, ..., 0.0000, 0.0046, 0.0000], |
| [ 0.0022, 0.0000, 0.0033, ..., 0.0000, 0.0036, 0.0000], |
| [-0.0108, 0.0000, -0.0057, ..., 0.0000, -0.0091, 0.0000]], |
| device='cuda:0') |
| Epoch 8, bias, value: tensor([-0.0035, 0.0279, 0.0055, 0.0049, 0.0042, -0.0072, 0.0225, -0.0270, |
| 0.0231, 0.0058], device='cuda:0'), grad: tensor([-0.0048, -0.0293, -0.0114, 0.0277, 0.0001, 0.0094, -0.0057, 0.0235, |
| 0.0070, -0.0167], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 7---------------------------------------------------- |
| epoch 7, time 219.46, cls_loss 0.1335 cls_loss_mapping 0.1743 cls_loss_causal 1.2317 re_mapping 0.0690 re_causal 0.1570 /// teacc 96.80 lr 0.00010000 |
| Epoch 9, weight, value: tensor([[-0.0148, 0.0228, -0.0085, ..., -0.0495, 0.0332, 0.0219], |
| [ 0.0047, 0.0125, -0.0512, ..., 0.0108, -0.0057, 0.0126], |
| [ 0.0269, -0.0145, -0.0251, ..., -0.0202, 0.0141, -0.0079], |
| ..., |
| [ 0.0080, -0.0013, 0.0080, ..., 0.0082, 0.0203, 0.0114], |
| [ 0.0060, 0.0208, -0.0156, ..., 0.0017, -0.0143, 0.0295], |
| [-0.0108, -0.0161, 0.0333, ..., -0.0362, -0.0297, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 2.0733e-03, 0.0000e+00, 2.9349e-04, ..., 0.0000e+00, |
| -7.9453e-05, 0.0000e+00], |
| [ 1.6060e-03, 0.0000e+00, 8.5402e-04, ..., 0.0000e+00, |
| 9.1362e-04, 0.0000e+00], |
| [ 5.8055e-05, 0.0000e+00, 3.3522e-04, ..., 0.0000e+00, |
| 9.9277e-04, 0.0000e+00], |
| ..., |
| [ 4.6692e-03, 0.0000e+00, 2.4109e-03, ..., 0.0000e+00, |
| 8.7357e-04, 0.0000e+00], |
| [-1.6518e-03, 0.0000e+00, 1.9779e-03, ..., 0.0000e+00, |
| -1.9665e-03, 0.0000e+00], |
| [ 2.2774e-03, 0.0000e+00, 4.0283e-03, ..., 0.0000e+00, |
| -9.0265e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 9, bias, value: tensor([-0.0035, 0.0284, 0.0051, 0.0052, 0.0044, -0.0077, 0.0223, -0.0268, |
| 0.0235, 0.0056], device='cuda:0'), grad: tensor([ 0.0006, 0.0042, 0.0017, 0.0190, -0.0242, -0.0178, -0.0019, 0.0090, |
| 0.0014, 0.0079], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 8---------------------------------------------------- |
| epoch 8, time 219.68, cls_loss 0.1251 cls_loss_mapping 0.1543 cls_loss_causal 1.1917 re_mapping 0.0646 re_causal 0.1443 /// teacc 97.04 lr 0.00010000 |
| Epoch 10, weight, value: tensor([[-0.0155, 0.0228, -0.0092, ..., -0.0506, 0.0329, 0.0219], |
| [ 0.0038, 0.0125, -0.0519, ..., 0.0105, -0.0049, 0.0126], |
| [ 0.0274, -0.0145, -0.0256, ..., -0.0202, 0.0133, -0.0079], |
| ..., |
| [ 0.0088, -0.0013, 0.0078, ..., 0.0089, 0.0203, 0.0114], |
| [ 0.0060, 0.0208, -0.0159, ..., 0.0020, -0.0140, 0.0295], |
| [-0.0118, -0.0161, 0.0339, ..., -0.0364, -0.0292, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 3.0828e-04, 0.0000e+00, 4.2892e-04, ..., 4.5970e-06, |
| 4.0579e-04, 0.0000e+00], |
| [ 6.8617e-04, 0.0000e+00, 8.0824e-05, ..., 1.7239e-06, |
| -4.3917e-04, 0.0000e+00], |
| [ 1.7424e-03, 0.0000e+00, 5.7983e-04, ..., 1.4137e-06, |
| 2.6398e-03, 0.0000e+00], |
| ..., |
| [ 1.7726e-04, 0.0000e+00, 2.1839e-04, ..., -1.5914e-05, |
| 5.3644e-04, 0.0000e+00], |
| [-4.1962e-03, 0.0000e+00, -5.4979e-04, ..., 1.0096e-06, |
| 2.3484e-05, 0.0000e+00], |
| [ 9.2316e-04, 0.0000e+00, 4.6015e-04, ..., 4.0270e-06, |
| 1.1492e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 10, bias, value: tensor([-0.0037, 0.0280, 0.0050, 0.0049, 0.0042, -0.0077, 0.0222, -0.0264, |
| 0.0240, 0.0057], device='cuda:0'), grad: tensor([ 0.0023, 0.0010, 0.0195, -0.0186, 0.0017, 0.0015, -0.0014, 0.0021, |
| -0.0106, 0.0024], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 9, time 218.62, cls_loss 0.1113 cls_loss_mapping 0.1363 cls_loss_causal 1.1221 re_mapping 0.0603 re_causal 0.1350 /// teacc 96.99 lr 0.00010000 |
| Epoch 11, weight, value: tensor([[-0.0164, 0.0228, -0.0096, ..., -0.0521, 0.0328, 0.0219], |
| [ 0.0033, 0.0125, -0.0523, ..., 0.0103, -0.0042, 0.0126], |
| [ 0.0280, -0.0145, -0.0261, ..., -0.0204, 0.0129, -0.0079], |
| ..., |
| [ 0.0089, -0.0013, 0.0071, ..., 0.0104, 0.0199, 0.0114], |
| [ 0.0057, 0.0208, -0.0162, ..., 0.0015, -0.0138, 0.0295], |
| [-0.0123, -0.0161, 0.0344, ..., -0.0364, -0.0289, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 1.7178e-04, 0.0000e+00, 1.3912e-04, ..., 2.2531e-05, |
| 3.3593e-04, 0.0000e+00], |
| [ 3.1853e-04, 0.0000e+00, 1.2231e-04, ..., 5.9724e-05, |
| -3.1519e-04, 0.0000e+00], |
| [ 1.0300e-03, 0.0000e+00, 1.1057e-04, ..., 3.7074e-04, |
| 4.4799e-04, 0.0000e+00], |
| ..., |
| [-2.0523e-03, 0.0000e+00, 4.9686e-04, ..., -3.4976e-04, |
| -1.0376e-03, 0.0000e+00], |
| [ 2.0905e-03, 0.0000e+00, 3.2921e-03, ..., 2.2125e-04, |
| 2.3098e-03, 0.0000e+00], |
| [-1.5465e-02, 0.0000e+00, -5.7793e-03, ..., 1.4937e-04, |
| -1.5442e-02, 0.0000e+00]], device='cuda:0') |
| Epoch 11, bias, value: tensor([-0.0038, 0.0280, 0.0052, 0.0048, 0.0044, -0.0077, 0.0219, -0.0265, |
| 0.0243, 0.0056], device='cuda:0'), grad: tensor([ 6.6853e-04, -5.5701e-05, 1.0920e-03, 1.9730e-02, 7.5483e-04, |
| -5.8441e-03, -1.4508e-04, -1.8644e-03, 5.8098e-03, -2.0142e-02], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 10---------------------------------------------------- |
| epoch 10, time 221.32, cls_loss 0.0999 cls_loss_mapping 0.1227 cls_loss_causal 1.1094 re_mapping 0.0549 re_causal 0.1267 /// teacc 97.24 lr 0.00010000 |
| Epoch 12, weight, value: tensor([[-0.0171, 0.0228, -0.0102, ..., -0.0525, 0.0324, 0.0219], |
| [ 0.0031, 0.0125, -0.0536, ..., 0.0116, -0.0032, 0.0126], |
| [ 0.0284, -0.0145, -0.0262, ..., -0.0203, 0.0123, -0.0079], |
| ..., |
| [ 0.0094, -0.0013, 0.0069, ..., 0.0110, 0.0197, 0.0114], |
| [ 0.0054, 0.0208, -0.0166, ..., 0.0004, -0.0138, 0.0295], |
| [-0.0129, -0.0161, 0.0346, ..., -0.0370, -0.0284, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 2.4557e-04, 0.0000e+00, -2.1343e-03, ..., 1.0300e-06, |
| -1.6947e-03, 0.0000e+00], |
| [ 2.1577e-04, 0.0000e+00, 3.1209e-04, ..., 1.6671e-07, |
| -8.0013e-04, 0.0000e+00], |
| [-1.9207e-03, 0.0000e+00, 4.3845e-04, ..., 1.9372e-07, |
| 2.2519e-04, 0.0000e+00], |
| ..., |
| [ 2.8992e-04, 0.0000e+00, 1.5678e-03, ..., 1.7462e-07, |
| 1.4315e-03, 0.0000e+00], |
| [-1.1641e-04, 0.0000e+00, -3.7231e-03, ..., 4.0419e-06, |
| -3.1757e-03, 0.0000e+00], |
| [ 1.1234e-03, 0.0000e+00, 7.3776e-03, ..., 1.6848e-06, |
| 2.5864e-03, 0.0000e+00]], device='cuda:0') |
| Epoch 12, bias, value: tensor([-0.0040, 0.0282, 0.0049, 0.0049, 0.0042, -0.0078, 0.0220, -0.0264, |
| 0.0244, 0.0056], device='cuda:0'), grad: tensor([-1.0231e-02, -1.9920e-04, -1.6049e-05, 4.3144e-03, -4.3869e-03, |
| 3.5877e-03, 1.2624e-04, 5.4092e-03, -1.2894e-02, 1.4290e-02], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 11---------------------------------------------------- |
| epoch 11, time 218.79, cls_loss 0.0856 cls_loss_mapping 0.1040 cls_loss_causal 1.0154 re_mapping 0.0538 re_causal 0.1211 /// teacc 97.46 lr 0.00010000 |
| Epoch 13, weight, value: tensor([[-1.7175e-02, 2.2769e-02, -1.0488e-02, ..., -5.3541e-02, |
| 3.2494e-02, 2.1893e-02], |
| [ 2.9271e-03, 1.2500e-02, -5.4046e-02, ..., 1.1676e-02, |
| -2.6134e-03, 1.2576e-02], |
| [ 2.8715e-02, -1.4480e-02, -2.6587e-02, ..., -2.0364e-02, |
| 1.1720e-02, -7.9110e-03], |
| ..., |
| [ 1.0327e-02, -1.2941e-03, 6.8708e-03, ..., 1.1770e-02, |
| 1.9598e-02, 1.1425e-02], |
| [ 4.7974e-03, 2.0836e-02, -1.6912e-02, ..., 4.7806e-05, |
| -1.3859e-02, 2.9502e-02], |
| [-1.3850e-02, -1.6141e-02, 3.5095e-02, ..., -3.7387e-02, |
| -2.7884e-02, -3.8447e-04]], device='cuda:0'), grad: tensor([[ 4.3821e-04, 0.0000e+00, 6.7174e-05, ..., 5.4501e-06, |
| 9.9778e-05, 0.0000e+00], |
| [ 1.0386e-03, 0.0000e+00, 1.5581e-04, ..., 1.2964e-05, |
| -3.2783e-04, 0.0000e+00], |
| [ 1.9627e-03, 0.0000e+00, 2.3377e-04, ..., 1.3493e-05, |
| 1.2789e-03, 0.0000e+00], |
| ..., |
| [-1.0918e-02, 0.0000e+00, -2.1954e-03, ..., -1.4901e-04, |
| -1.3571e-03, 0.0000e+00], |
| [ 2.9602e-03, 0.0000e+00, 9.1553e-05, ..., 1.5095e-05, |
| -7.1478e-04, 0.0000e+00], |
| [-3.8795e-03, 0.0000e+00, -1.3718e-02, ..., 4.4942e-05, |
| 7.5340e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 13, bias, value: tensor([-0.0035, 0.0282, 0.0049, 0.0050, 0.0041, -0.0079, 0.0217, -0.0261, |
| 0.0245, 0.0054], device='cuda:0'), grad: tensor([-0.0006, 0.0010, -0.0040, 0.0029, 0.0187, 0.0026, -0.0006, -0.0105, |
| 0.0057, -0.0152], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 12---------------------------------------------------- |
| epoch 12, time 219.25, cls_loss 0.0909 cls_loss_mapping 0.1133 cls_loss_causal 1.0595 re_mapping 0.0484 re_causal 0.1174 /// teacc 97.57 lr 0.00010000 |
| Epoch 14, weight, value: tensor([[-0.0174, 0.0223, -0.0110, ..., -0.0539, 0.0324, 0.0219], |
| [ 0.0026, 0.0108, -0.0538, ..., 0.0116, -0.0016, 0.0126], |
| [ 0.0294, -0.0126, -0.0267, ..., -0.0198, 0.0113, -0.0079], |
| ..., |
| [ 0.0107, -0.0024, 0.0064, ..., 0.0118, 0.0192, 0.0114], |
| [ 0.0041, 0.0201, -0.0172, ..., -0.0004, -0.0138, 0.0295], |
| [-0.0141, -0.0164, 0.0356, ..., -0.0377, -0.0274, -0.0004]], |
| device='cuda:0'), grad: tensor([[-1.1490e-02, 0.0000e+00, -4.9531e-05, ..., 0.0000e+00, |
| -1.7609e-02, 0.0000e+00], |
| [ 8.0109e-03, 0.0000e+00, 3.1352e-04, ..., 0.0000e+00, |
| 1.1238e-02, 0.0000e+00], |
| [-5.4665e-03, 0.0000e+00, 1.9860e-04, ..., 0.0000e+00, |
| 4.8375e-04, 0.0000e+00], |
| ..., |
| [ 3.0017e-04, 0.0000e+00, 1.4007e-04, ..., 0.0000e+00, |
| 4.7827e-04, 0.0000e+00], |
| [ 5.2948e-03, 0.0000e+00, 6.5279e-04, ..., 0.0000e+00, |
| 1.3485e-03, 0.0000e+00], |
| [ 3.5019e-03, 0.0000e+00, 1.9302e-03, ..., 0.0000e+00, |
| 2.7733e-03, 0.0000e+00]], device='cuda:0') |
| Epoch 14, bias, value: tensor([-0.0034, 0.0282, 0.0052, 0.0048, 0.0041, -0.0080, 0.0213, -0.0263, |
| 0.0245, 0.0056], device='cuda:0'), grad: tensor([-0.0307, 0.0211, -0.0127, -0.0020, -0.0030, -0.0025, 0.0077, 0.0009, |
| 0.0143, 0.0070], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 13, time 218.77, cls_loss 0.0798 cls_loss_mapping 0.0973 cls_loss_causal 1.0558 re_mapping 0.0448 re_causal 0.1108 /// teacc 97.50 lr 0.00010000 |
| Epoch 15, weight, value: tensor([[-0.0175, 0.0202, -0.0115, ..., -0.0566, 0.0324, 0.0219], |
| [ 0.0023, 0.0079, -0.0546, ..., 0.0113, -0.0007, 0.0126], |
| [ 0.0293, -0.0095, -0.0275, ..., -0.0198, 0.0108, -0.0079], |
| ..., |
| [ 0.0112, -0.0049, 0.0064, ..., 0.0129, 0.0189, 0.0114], |
| [ 0.0036, 0.0175, -0.0176, ..., -0.0007, -0.0135, 0.0295], |
| [-0.0153, -0.0178, 0.0362, ..., -0.0381, -0.0271, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 1.3185e-04, 0.0000e+00, -1.1963e-04, ..., 1.0490e-05, |
| 2.2864e-04, 0.0000e+00], |
| [-1.4439e-03, 0.0000e+00, 6.2943e-05, ..., 1.4651e-04, |
| -2.2297e-03, 0.0000e+00], |
| [ 2.1629e-03, 0.0000e+00, 7.2241e-05, ..., 7.5197e-04, |
| 1.1454e-03, 0.0000e+00], |
| ..., |
| [ 3.1403e-02, 0.0000e+00, 3.8028e-04, ..., 1.6312e-02, |
| 4.7922e-04, 0.0000e+00], |
| [ 4.9973e-04, 0.0000e+00, 1.9255e-03, ..., 1.0657e-04, |
| 3.2711e-04, 0.0000e+00], |
| [ 1.1492e-03, 0.0000e+00, 1.2932e-03, ..., 5.6684e-05, |
| 3.3975e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 15, bias, value: tensor([-0.0033, 0.0283, 0.0051, 0.0051, 0.0043, -0.0082, 0.0211, -0.0261, |
| 0.0245, 0.0054], device='cuda:0'), grad: tensor([-0.0007, -0.0028, 0.0026, -0.0197, -0.0010, -0.0102, 0.0006, 0.0254, |
| 0.0029, 0.0028], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 14, time 217.64, cls_loss 0.0826 cls_loss_mapping 0.1031 cls_loss_causal 1.0196 re_mapping 0.0416 re_causal 0.0998 /// teacc 97.44 lr 0.00010000 |
| Epoch 16, weight, value: tensor([[-1.7718e-02, 1.2063e-02, -1.2141e-02, ..., -5.6871e-02, |
| 3.2121e-02, 2.1893e-02], |
| [ 1.5121e-03, -5.6718e-04, -5.4833e-02, ..., 1.1570e-02, |
| -2.3863e-05, 1.2576e-02], |
| [ 2.9192e-02, -1.5483e-03, -2.7864e-02, ..., -2.0202e-02, |
| 1.0512e-02, -7.9110e-03], |
| ..., |
| [ 1.2028e-02, -1.3074e-02, 5.7202e-03, ..., 1.3618e-02, |
| 1.8653e-02, 1.1425e-02], |
| [ 4.3767e-03, 1.0639e-02, -1.7824e-02, ..., -8.9181e-04, |
| -1.3371e-02, 2.9502e-02], |
| [-1.6387e-02, -2.5177e-02, 3.6844e-02, ..., -3.8102e-02, |
| -2.6650e-02, -3.8447e-04]], device='cuda:0'), grad: tensor([[ 5.1260e-05, 0.0000e+00, -8.0913e-06, ..., 3.9674e-07, |
| -1.6284e-04, 0.0000e+00], |
| [ 3.3450e-04, 0.0000e+00, 2.2471e-04, ..., 2.7921e-06, |
| -3.1233e-04, 0.0000e+00], |
| [ 1.9503e-04, 0.0000e+00, 1.0413e-04, ..., 1.9163e-05, |
| 1.1146e-04, 0.0000e+00], |
| ..., |
| [-9.7370e-04, 0.0000e+00, 1.0544e-04, ..., 2.6613e-05, |
| -6.1214e-05, 0.0000e+00], |
| [ 3.0208e-04, 0.0000e+00, 2.2686e-04, ..., 1.3120e-05, |
| 1.8597e-04, 0.0000e+00], |
| [ 2.8443e-04, 0.0000e+00, -5.9414e-04, ..., 8.5756e-06, |
| -4.8470e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 16, bias, value: tensor([-0.0035, 0.0285, 0.0049, 0.0049, 0.0045, -0.0085, 0.0211, -0.0261, |
| 0.0251, 0.0052], device='cuda:0'), grad: tensor([-1.1375e-02, 1.7452e-04, 6.6805e-04, 1.0559e-02, 7.7844e-05, |
| -3.0689e-03, 3.1452e-03, -7.8106e-04, 9.4700e-04, -3.5429e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 15---------------------------------------------------- |
| epoch 15, time 217.96, cls_loss 0.0682 cls_loss_mapping 0.0822 cls_loss_causal 0.9587 re_mapping 0.0406 re_causal 0.0976 /// teacc 97.77 lr 0.00010000 |
| Epoch 17, weight, value: tensor([[-0.0181, 0.0045, -0.0123, ..., -0.0573, 0.0318, 0.0219], |
| [ 0.0012, 0.0009, -0.0553, ..., 0.0113, 0.0007, 0.0126], |
| [ 0.0296, -0.0034, -0.0283, ..., -0.0206, 0.0101, -0.0079], |
| ..., |
| [ 0.0129, -0.0164, 0.0054, ..., 0.0138, 0.0187, 0.0114], |
| [ 0.0039, 0.0084, -0.0182, ..., -0.0005, -0.0135, 0.0295], |
| [-0.0172, -0.0323, 0.0373, ..., -0.0381, -0.0264, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 1.4687e-04, 0.0000e+00, 1.2720e-04, ..., 2.6543e-07, |
| 1.8859e-04, 0.0000e+00], |
| [ 1.4913e-04, 0.0000e+00, 1.7488e-04, ..., 1.1977e-06, |
| -1.1253e-03, 0.0000e+00], |
| [-6.3324e-04, 0.0000e+00, 1.1492e-04, ..., 3.8277e-07, |
| 1.6022e-04, 0.0000e+00], |
| ..., |
| [ 2.5234e-03, 0.0000e+00, 5.4646e-04, ..., -6.7838e-06, |
| 8.2111e-04, 0.0000e+00], |
| [ 3.8171e-04, 0.0000e+00, 1.2569e-03, ..., 3.0734e-07, |
| 7.5245e-04, 0.0000e+00], |
| [ 3.7074e-04, 0.0000e+00, -4.2114e-03, ..., 2.6487e-06, |
| -5.6601e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 17, bias, value: tensor([-0.0036, 0.0284, 0.0050, 0.0047, 0.0045, -0.0085, 0.0211, -0.0256, |
| 0.0248, 0.0053], device='cuda:0'), grad: tensor([ 0.0004, -0.0008, -0.0018, 0.0008, 0.0044, -0.0139, 0.0076, 0.0042, |
| 0.0035, -0.0042], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 16, time 217.24, cls_loss 0.0656 cls_loss_mapping 0.0820 cls_loss_causal 0.9977 re_mapping 0.0380 re_causal 0.0964 /// teacc 97.29 lr 0.00010000 |
| Epoch 18, weight, value: tensor([[-0.0176, -0.0005, -0.0122, ..., -0.0576, 0.0316, 0.0219], |
| [ 0.0006, 0.0024, -0.0561, ..., 0.0109, 0.0013, 0.0126], |
| [ 0.0301, -0.0043, -0.0290, ..., -0.0205, 0.0099, -0.0079], |
| ..., |
| [ 0.0131, -0.0207, 0.0057, ..., 0.0136, 0.0186, 0.0114], |
| [ 0.0035, 0.0050, -0.0187, ..., -0.0001, -0.0133, 0.0295], |
| [-0.0181, -0.0383, 0.0374, ..., -0.0383, -0.0263, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 8.4162e-05, 0.0000e+00, 6.0856e-05, ..., 2.8871e-07, |
| 5.6237e-05, 0.0000e+00], |
| [-8.5354e-04, 0.0000e+00, 8.1241e-05, ..., 4.4331e-07, |
| -1.0977e-03, 0.0000e+00], |
| [-2.4509e-04, 0.0000e+00, 7.7248e-05, ..., 7.7300e-07, |
| 1.1319e-04, 0.0000e+00], |
| ..., |
| [-6.0678e-05, 0.0000e+00, 8.0490e-04, ..., -5.2229e-06, |
| 2.4199e-04, 0.0000e+00], |
| [ 5.5408e-04, 0.0000e+00, 1.1454e-03, ..., 2.2631e-07, |
| 2.9325e-04, 0.0000e+00], |
| [-8.4305e-04, 0.0000e+00, -4.2801e-03, ..., 1.4668e-06, |
| 7.2896e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 18, bias, value: tensor([-0.0032, 0.0283, 0.0051, 0.0047, 0.0047, -0.0088, 0.0209, -0.0256, |
| 0.0248, 0.0053], device='cuda:0'), grad: tensor([-0.0002, -0.0039, 0.0004, 0.0022, 0.0046, -0.0015, 0.0001, 0.0017, |
| 0.0029, -0.0064], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 17, time 217.25, cls_loss 0.0763 cls_loss_mapping 0.0965 cls_loss_causal 0.9589 re_mapping 0.0377 re_causal 0.0918 /// teacc 97.72 lr 0.00010000 |
| Epoch 19, weight, value: tensor([[-1.8015e-02, -1.0963e-02, -1.2383e-02, ..., -5.7778e-02, |
| 3.1339e-02, 2.1893e-02], |
| [ 8.5515e-05, 4.4174e-03, -5.6827e-02, ..., 1.0723e-02, |
| 2.0767e-03, 1.2576e-02], |
| [ 3.0601e-02, -3.0445e-03, -2.8312e-02, ..., -2.0565e-02, |
| 9.7862e-03, -7.9110e-03], |
| ..., |
| [ 1.2963e-02, -2.1107e-02, 5.4822e-03, ..., 1.4224e-02, |
| 1.8137e-02, 1.1425e-02], |
| [ 3.4132e-03, -6.8826e-03, -1.9315e-02, ..., -2.9777e-04, |
| -1.3033e-02, 2.9502e-02], |
| [-1.8927e-02, -5.1199e-02, 3.7845e-02, ..., -3.8537e-02, |
| -2.5991e-02, -3.8447e-04]], device='cuda:0'), grad: tensor([[ 3.1853e-04, 2.1532e-06, 1.5855e-04, ..., 2.0117e-07, |
| 1.8072e-04, 0.0000e+00], |
| [ 2.7370e-04, 7.7114e-06, 6.9857e-05, ..., -1.7434e-05, |
| -6.2704e-05, 0.0000e+00], |
| [-6.3896e-04, -2.4572e-05, 1.4186e-04, ..., 1.6643e-06, |
| 3.7342e-05, 0.0000e+00], |
| ..., |
| [-8.5115e-04, 2.7381e-06, 1.6773e-04, ..., 7.1079e-06, |
| -7.5245e-04, 0.0000e+00], |
| [ 3.0422e-04, 3.9712e-06, 5.8651e-04, ..., 1.0869e-06, |
| 8.0168e-05, 0.0000e+00], |
| [ 7.1096e-04, 3.1386e-07, 2.6274e-04, ..., 7.3947e-07, |
| 4.4203e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 19, bias, value: tensor([-0.0031, 0.0282, 0.0061, 0.0049, 0.0044, -0.0090, 0.0207, -0.0260, |
| 0.0248, 0.0051], device='cuda:0'), grad: tensor([ 1.1177e-03, 3.6025e-04, -2.2340e-04, -1.9180e-02, 8.4519e-05, |
| 1.4404e-02, 1.5745e-03, -2.0638e-03, 1.7977e-03, 2.1400e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 18---------------------------------------------------- |
| epoch 18, time 218.21, cls_loss 0.0601 cls_loss_mapping 0.0775 cls_loss_causal 0.9205 re_mapping 0.0346 re_causal 0.0893 /// teacc 97.79 lr 0.00010000 |
| Epoch 20, weight, value: tensor([[-0.0183, -0.0115, -0.0128, ..., -0.0579, 0.0312, 0.0219], |
| [-0.0007, 0.0045, -0.0574, ..., 0.0107, 0.0024, 0.0126], |
| [ 0.0308, -0.0015, -0.0285, ..., -0.0207, 0.0096, -0.0079], |
| ..., |
| [ 0.0134, -0.0245, 0.0053, ..., 0.0145, 0.0179, 0.0114], |
| [ 0.0032, -0.0086, -0.0197, ..., -0.0002, -0.0130, 0.0295], |
| [-0.0196, -0.0530, 0.0381, ..., -0.0390, -0.0255, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 1.0884e-04, 0.0000e+00, 1.6153e-05, ..., 1.6764e-08, |
| 6.9022e-05, 0.0000e+00], |
| [ 1.8191e-04, 0.0000e+00, 2.9355e-05, ..., 1.6857e-07, |
| 1.8477e-04, 0.0000e+00], |
| [-1.3094e-03, 0.0000e+00, -3.0130e-05, ..., 1.5832e-07, |
| 1.6534e-04, 0.0000e+00], |
| ..., |
| [ 1.2052e-04, 0.0000e+00, 2.3365e-05, ..., -1.0924e-06, |
| -1.6861e-03, 0.0000e+00], |
| [-1.8060e-05, 0.0000e+00, 1.0139e-04, ..., 1.0245e-07, |
| 6.1655e-04, 0.0000e+00], |
| [-9.4414e-05, 0.0000e+00, 6.0320e-04, ..., 2.3562e-07, |
| -2.7871e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 20, bias, value: tensor([-0.0034, 0.0280, 0.0061, 0.0053, 0.0048, -0.0089, 0.0204, -0.0259, |
| 0.0248, 0.0049], device='cuda:0'), grad: tensor([ 4.3273e-04, 2.6550e-03, -2.4376e-03, 1.8759e-03, 7.3075e-05, |
| 1.4553e-03, 1.5039e-03, -1.5472e-02, 1.0910e-02, -1.0080e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 19---------------------------------------------------- |
| epoch 19, time 218.04, cls_loss 0.0640 cls_loss_mapping 0.0817 cls_loss_causal 0.8950 re_mapping 0.0349 re_causal 0.0851 /// teacc 98.11 lr 0.00010000 |
| Epoch 21, weight, value: tensor([[-0.0184, -0.0115, -0.0130, ..., -0.0580, 0.0308, 0.0219], |
| [-0.0009, 0.0049, -0.0574, ..., 0.0107, 0.0033, 0.0126], |
| [ 0.0312, -0.0017, -0.0292, ..., -0.0208, 0.0092, -0.0079], |
| ..., |
| [ 0.0143, -0.0247, 0.0050, ..., 0.0148, 0.0179, 0.0114], |
| [ 0.0026, -0.0089, -0.0199, ..., -0.0002, -0.0128, 0.0295], |
| [-0.0204, -0.0534, 0.0383, ..., -0.0392, -0.0253, -0.0004]], |
| device='cuda:0'), grad: tensor([[-5.2547e-04, 1.7667e-06, 5.3942e-05, ..., 5.7742e-08, |
| 4.2856e-05, 0.0000e+00], |
| [ 3.6860e-04, 1.1669e-06, 1.4949e-04, ..., 1.7779e-06, |
| -6.7472e-05, 0.0000e+00], |
| [-7.4244e-04, -1.0341e-05, 3.2067e-05, ..., 3.9022e-07, |
| 2.6047e-05, 0.0000e+00], |
| ..., |
| [-3.2663e-04, 2.5127e-06, 1.7512e-04, ..., -5.9791e-06, |
| 1.5795e-04, 0.0000e+00], |
| [ 2.0874e-04, 1.8450e-06, 1.1164e-04, ..., 2.2445e-07, |
| -2.4092e-04, 0.0000e+00], |
| [-1.0830e-04, 3.3621e-07, 3.2806e-04, ..., 1.4734e-06, |
| -2.7418e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 21, bias, value: tensor([-0.0038, 0.0283, 0.0060, 0.0048, 0.0048, -0.0084, 0.0200, -0.0256, |
| 0.0249, 0.0049], device='cuda:0'), grad: tensor([-0.0028, 0.0007, -0.0006, 0.0019, -0.0031, 0.0019, 0.0008, 0.0007, |
| -0.0004, 0.0009], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 20, time 217.31, cls_loss 0.0542 cls_loss_mapping 0.0705 cls_loss_causal 0.9315 re_mapping 0.0321 re_causal 0.0833 /// teacc 98.00 lr 0.00010000 |
| Epoch 22, weight, value: tensor([[-0.0184, -0.0117, -0.0132, ..., -0.0585, 0.0304, 0.0219], |
| [-0.0010, 0.0058, -0.0580, ..., 0.0102, 0.0037, 0.0126], |
| [ 0.0317, -0.0022, -0.0298, ..., -0.0210, 0.0090, -0.0079], |
| ..., |
| [ 0.0145, -0.0249, 0.0048, ..., 0.0151, 0.0174, 0.0114], |
| [ 0.0020, -0.0101, -0.0203, ..., -0.0005, -0.0128, 0.0295], |
| [-0.0208, -0.0541, 0.0386, ..., -0.0394, -0.0246, -0.0004]], |
| device='cuda:0'), grad: tensor([[-9.7454e-05, 3.9972e-06, -3.6925e-05, ..., 2.0768e-07, |
| 3.5435e-05, 0.0000e+00], |
| [ 1.3280e-04, 1.9222e-05, 2.7806e-05, ..., 1.4547e-06, |
| -6.2227e-04, 0.0000e+00], |
| [-6.7043e-04, -1.7178e-04, 6.2108e-05, ..., 1.5860e-06, |
| 1.1045e-04, 0.0000e+00], |
| ..., |
| [ 6.2799e-04, 1.4037e-05, 9.3579e-05, ..., -7.2643e-06, |
| 3.0923e-04, 0.0000e+00], |
| [ 7.3671e-04, 8.6010e-05, 4.7803e-04, ..., 5.4855e-07, |
| -2.6751e-04, 0.0000e+00], |
| [ 5.8937e-04, 1.4156e-06, -2.1660e-04, ..., 1.5246e-06, |
| -2.0134e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 22, bias, value: tensor([-0.0040, 0.0282, 0.0062, 0.0045, 0.0049, -0.0084, 0.0202, -0.0260, |
| 0.0250, 0.0051], device='cuda:0'), grad: tensor([-0.0005, -0.0007, -0.0005, 0.0026, 0.0002, -0.0050, 0.0004, 0.0015, |
| 0.0013, 0.0006], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 21---------------------------------------------------- |
| epoch 21, time 218.45, cls_loss 0.0558 cls_loss_mapping 0.0664 cls_loss_causal 0.8972 re_mapping 0.0329 re_causal 0.0796 /// teacc 98.12 lr 0.00010000 |
| Epoch 23, weight, value: tensor([[-0.0186, -0.0118, -0.0133, ..., -0.0587, 0.0301, 0.0219], |
| [-0.0009, 0.0076, -0.0593, ..., 0.0102, 0.0041, 0.0126], |
| [ 0.0318, -0.0036, -0.0302, ..., -0.0211, 0.0086, -0.0079], |
| ..., |
| [ 0.0148, -0.0253, 0.0048, ..., 0.0162, 0.0172, 0.0114], |
| [ 0.0021, -0.0110, -0.0203, ..., -0.0007, -0.0126, 0.0295], |
| [-0.0218, -0.0545, 0.0391, ..., -0.0396, -0.0243, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 3.8981e-04, 0.0000e+00, 1.4699e-04, ..., 2.1756e-06, |
| 2.6536e-04, 0.0000e+00], |
| [ 6.7186e-04, 0.0000e+00, 3.4070e-04, ..., 6.9365e-06, |
| -9.4461e-04, 0.0000e+00], |
| [-3.8490e-03, 0.0000e+00, 8.8835e-04, ..., 6.3591e-06, |
| -6.7091e-04, 0.0000e+00], |
| ..., |
| [-5.2261e-04, 0.0000e+00, 3.4733e-03, ..., 1.2529e-04, |
| 5.5164e-05, 0.0000e+00], |
| [ 3.1433e-03, 0.0000e+00, 1.5602e-03, ..., 4.8876e-05, |
| 1.4668e-03, 0.0000e+00], |
| [ 7.0763e-04, 0.0000e+00, 1.0300e-03, ..., 2.1428e-05, |
| 9.9838e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 23, bias, value: tensor([-0.0037, 0.0282, 0.0060, 0.0042, 0.0050, -0.0085, 0.0200, -0.0257, |
| 0.0254, 0.0050], device='cuda:0'), grad: tensor([ 0.0011, 0.0006, -0.0045, 0.0013, -0.0178, 0.0050, -0.0032, 0.0054, |
| 0.0092, 0.0029], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 22, time 217.54, cls_loss 0.0515 cls_loss_mapping 0.0723 cls_loss_causal 0.9218 re_mapping 0.0304 re_causal 0.0818 /// teacc 98.03 lr 0.00010000 |
| Epoch 24, weight, value: tensor([[-0.0190, -0.0121, -0.0134, ..., -0.0590, 0.0299, 0.0219], |
| [-0.0009, 0.0083, -0.0598, ..., 0.0101, 0.0048, 0.0126], |
| [ 0.0321, -0.0038, -0.0306, ..., -0.0207, 0.0081, -0.0079], |
| ..., |
| [ 0.0150, -0.0254, 0.0046, ..., 0.0170, 0.0171, 0.0114], |
| [ 0.0021, -0.0132, -0.0203, ..., -0.0009, -0.0127, 0.0295], |
| [-0.0229, -0.0565, 0.0390, ..., -0.0399, -0.0241, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 8.8394e-05, 4.2655e-06, 3.4682e-06, ..., 5.5693e-06, |
| 2.7239e-05, 0.0000e+00], |
| [-1.1295e-04, -7.3290e-04, -4.3720e-05, ..., -3.6359e-05, |
| -5.7220e-04, 0.0000e+00], |
| [ 4.0527e-02, 4.5562e-04, 1.8632e-04, ..., 6.4278e-03, |
| 1.5032e-04, 0.0000e+00], |
| ..., |
| [ 6.4468e-04, 7.9751e-05, 3.1877e-04, ..., 3.8838e-04, |
| 3.8862e-04, 0.0000e+00], |
| [-6.1512e-04, 7.6592e-05, -3.2444e-03, ..., 3.0696e-05, |
| -7.5388e-04, 0.0000e+00], |
| [ 2.3174e-03, 1.2986e-05, 2.5196e-03, ..., 1.5192e-05, |
| 5.9557e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 24, bias, value: tensor([-0.0039, 0.0285, 0.0059, 0.0042, 0.0053, -0.0083, 0.0200, -0.0256, |
| 0.0251, 0.0047], device='cuda:0'), grad: tensor([-0.0012, -0.0017, 0.0298, -0.0299, 0.0003, 0.0004, 0.0009, 0.0002, |
| -0.0079, 0.0092], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 23, time 217.11, cls_loss 0.0458 cls_loss_mapping 0.0581 cls_loss_causal 0.8614 re_mapping 0.0302 re_causal 0.0783 /// teacc 98.01 lr 0.00010000 |
| Epoch 25, weight, value: tensor([[-0.0196, -0.0126, -0.0135, ..., -0.0592, 0.0296, 0.0219], |
| [-0.0011, 0.0091, -0.0608, ..., 0.0096, 0.0052, 0.0126], |
| [ 0.0319, -0.0037, -0.0306, ..., -0.0209, 0.0079, -0.0079], |
| ..., |
| [ 0.0156, -0.0260, 0.0052, ..., 0.0189, 0.0168, 0.0114], |
| [ 0.0015, -0.0175, -0.0203, ..., -0.0011, -0.0129, 0.0295], |
| [-0.0235, -0.0598, 0.0393, ..., -0.0400, -0.0235, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 4.8935e-05, 0.0000e+00, 2.1607e-05, ..., 1.3225e-07, |
| 3.8505e-05, 0.0000e+00], |
| [ 1.8549e-04, 0.0000e+00, 1.1384e-05, ..., 1.1250e-06, |
| -2.6122e-05, 0.0000e+00], |
| [-8.5258e-04, 0.0000e+00, 8.4117e-06, ..., 1.3690e-06, |
| 5.0306e-05, 0.0000e+00], |
| ..., |
| [ 4.9067e-04, 0.0000e+00, 1.0548e-03, ..., -1.5140e-05, |
| 1.9670e-04, 0.0000e+00], |
| [ 2.7752e-04, 0.0000e+00, 5.5969e-05, ..., 8.2888e-07, |
| 1.3232e-04, 0.0000e+00], |
| [ 1.8609e-04, 0.0000e+00, -1.1787e-03, ..., 6.0052e-06, |
| -9.8109e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 25, bias, value: tensor([-0.0035, 0.0285, 0.0059, 0.0044, 0.0050, -0.0086, 0.0200, -0.0250, |
| 0.0248, 0.0046], device='cuda:0'), grad: tensor([ 8.8573e-05, 2.4557e-04, -1.3609e-03, -4.1275e-03, 2.1636e-04, |
| 3.8681e-03, -4.5490e-04, 2.5425e-03, 5.7983e-04, -1.5926e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 24---------------------------------------------------- |
| epoch 24, time 218.00, cls_loss 0.0482 cls_loss_mapping 0.0591 cls_loss_causal 0.9073 re_mapping 0.0287 re_causal 0.0757 /// teacc 98.21 lr 0.00010000 |
| Epoch 26, weight, value: tensor([[-0.0200, -0.0126, -0.0137, ..., -0.0600, 0.0294, 0.0219], |
| [-0.0016, 0.0096, -0.0609, ..., 0.0104, 0.0054, 0.0126], |
| [ 0.0320, -0.0038, -0.0307, ..., -0.0209, 0.0078, -0.0079], |
| ..., |
| [ 0.0163, -0.0261, 0.0048, ..., 0.0181, 0.0170, 0.0114], |
| [ 0.0009, -0.0183, -0.0208, ..., -0.0012, -0.0128, 0.0295], |
| [-0.0241, -0.0604, 0.0394, ..., -0.0390, -0.0238, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 9.5725e-05, 0.0000e+00, 8.0466e-05, ..., 1.1489e-05, |
| 4.8697e-05, 0.0000e+00], |
| [-2.5034e-04, 0.0000e+00, 1.3340e-04, ..., -2.1994e-04, |
| -6.4373e-04, 0.0000e+00], |
| [ 8.7690e-04, 0.0000e+00, 5.6028e-04, ..., 9.1791e-05, |
| 7.7188e-05, 0.0000e+00], |
| ..., |
| [-3.2005e-03, 0.0000e+00, -2.3712e-02, ..., -4.6501e-03, |
| 3.2640e-04, 0.0000e+00], |
| [-3.3665e-04, 0.0000e+00, -1.1625e-03, ..., 1.3635e-05, |
| 9.6321e-05, 0.0000e+00], |
| [-9.8944e-05, 0.0000e+00, -9.7084e-04, ..., 2.3186e-05, |
| 8.5235e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 26, bias, value: tensor([-0.0037, 0.0280, 0.0060, 0.0041, 0.0053, -0.0081, 0.0202, -0.0248, |
| 0.0245, 0.0045], device='cuda:0'), grad: tensor([-0.0005, -0.0022, 0.0017, 0.0016, 0.0291, 0.0009, -0.0002, -0.0281, |
| -0.0015, -0.0009], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 25, time 217.30, cls_loss 0.0344 cls_loss_mapping 0.0494 cls_loss_causal 0.8657 re_mapping 0.0287 re_causal 0.0770 /// teacc 98.14 lr 0.00010000 |
| Epoch 27, weight, value: tensor([[-0.0204, -0.0127, -0.0142, ..., -0.0601, 0.0290, 0.0219], |
| [-0.0020, 0.0102, -0.0611, ..., 0.0110, 0.0059, 0.0126], |
| [ 0.0326, -0.0040, -0.0305, ..., -0.0214, 0.0074, -0.0079], |
| ..., |
| [ 0.0168, -0.0271, 0.0049, ..., 0.0186, 0.0166, 0.0114], |
| [ 0.0006, -0.0189, -0.0212, ..., -0.0013, -0.0127, 0.0295], |
| [-0.0250, -0.0607, 0.0398, ..., -0.0388, -0.0233, -0.0004]], |
| device='cuda:0'), grad: tensor([[-1.2600e-04, 0.0000e+00, -5.7258e-06, ..., 2.7940e-08, |
| 8.4005e-07, 0.0000e+00], |
| [-2.2388e-04, 0.0000e+00, 2.0713e-05, ..., -2.2314e-06, |
| -1.1387e-03, 0.0000e+00], |
| [ 9.0837e-05, 0.0000e+00, 1.2839e-04, ..., 1.0990e-07, |
| 8.0287e-05, 0.0000e+00], |
| ..., |
| [ 7.8827e-06, 0.0000e+00, 1.3566e-04, ..., 2.3469e-07, |
| 1.9968e-04, 0.0000e+00], |
| [ 2.4009e-04, 0.0000e+00, -1.1176e-04, ..., 2.1607e-07, |
| 3.5644e-04, 0.0000e+00], |
| [ 1.2875e-04, 0.0000e+00, -2.6059e-04, ..., 4.6007e-07, |
| 1.1408e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 27, bias, value: tensor([-0.0038, 0.0279, 0.0062, 0.0042, 0.0051, -0.0086, 0.0203, -0.0245, |
| 0.0244, 0.0047], device='cuda:0'), grad: tensor([-5.7173e-04, -2.7485e-03, 6.3372e-04, -4.7188e-03, 1.0300e-03, |
| 4.5738e-03, 5.1320e-05, 8.1778e-04, 7.6199e-04, 1.6844e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 26---------------------------------------------------- |
| epoch 26, time 218.42, cls_loss 0.0370 cls_loss_mapping 0.0525 cls_loss_causal 0.8234 re_mapping 0.0278 re_causal 0.0727 /// teacc 98.29 lr 0.00010000 |
| Epoch 28, weight, value: tensor([[-0.0206, -0.0127, -0.0145, ..., -0.0603, 0.0286, 0.0219], |
| [-0.0021, 0.0102, -0.0619, ..., 0.0110, 0.0065, 0.0126], |
| [ 0.0332, -0.0039, -0.0310, ..., -0.0213, 0.0069, -0.0079], |
| ..., |
| [ 0.0168, -0.0272, 0.0049, ..., 0.0193, 0.0162, 0.0114], |
| [ 0.0002, -0.0190, -0.0212, ..., -0.0014, -0.0128, 0.0295], |
| [-0.0255, -0.0607, 0.0398, ..., -0.0389, -0.0230, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 1.9801e-04, 0.0000e+00, 1.0413e-04, ..., 4.6566e-08, |
| 6.3300e-05, 0.0000e+00], |
| [ 3.9062e-03, 0.0000e+00, 1.3137e-04, ..., -2.6841e-06, |
| 4.3106e-04, 0.0000e+00], |
| [-4.9362e-03, 0.0000e+00, -7.3195e-05, ..., 8.4937e-07, |
| -1.2693e-03, 0.0000e+00], |
| ..., |
| [-2.3804e-03, 0.0000e+00, -2.9869e-03, ..., 1.0431e-07, |
| -1.0738e-03, 0.0000e+00], |
| [ 1.0395e-03, 0.0000e+00, 5.9652e-04, ..., 4.0792e-07, |
| 3.7146e-04, 0.0000e+00], |
| [ 2.0847e-03, 0.0000e+00, 3.2883e-03, ..., 2.0489e-08, |
| 8.9359e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 28, bias, value: tensor([-0.0038, 0.0280, 0.0064, 0.0041, 0.0049, -0.0080, 0.0204, -0.0248, |
| 0.0242, 0.0047], device='cuda:0'), grad: tensor([ 0.0017, 0.0159, -0.0332, 0.0026, -0.0008, 0.0006, 0.0009, -0.0072, |
| 0.0081, 0.0114], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 27, time 217.22, cls_loss 0.0374 cls_loss_mapping 0.0512 cls_loss_causal 0.8379 re_mapping 0.0267 re_causal 0.0732 /// teacc 98.14 lr 0.00010000 |
| Epoch 29, weight, value: tensor([[-0.0209, -0.0128, -0.0149, ..., -0.0606, 0.0283, 0.0219], |
| [-0.0025, 0.0107, -0.0627, ..., 0.0110, 0.0073, 0.0126], |
| [ 0.0328, -0.0042, -0.0315, ..., -0.0215, 0.0066, -0.0079], |
| ..., |
| [ 0.0173, -0.0273, 0.0052, ..., 0.0194, 0.0156, 0.0114], |
| [-0.0002, -0.0196, -0.0214, ..., -0.0016, -0.0127, 0.0295], |
| [-0.0266, -0.0618, 0.0400, ..., -0.0389, -0.0226, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 1.1474e-04, 2.0489e-08, 1.8191e-04, ..., 2.9244e-07, |
| 1.1253e-04, 0.0000e+00], |
| [ 4.4346e-04, 6.4261e-07, 2.1696e-04, ..., 2.5276e-06, |
| 2.4587e-05, 0.0000e+00], |
| [ 2.0778e-04, 4.6194e-07, 8.2016e-04, ..., 1.9707e-06, |
| 9.1195e-05, 0.0000e+00], |
| ..., |
| [ 1.2541e-04, -1.9316e-06, 3.4761e-04, ..., -2.0005e-06, |
| 4.7833e-05, 0.0000e+00], |
| [ 4.6062e-04, 9.8720e-08, 6.0797e-04, ..., 1.9651e-06, |
| -3.4833e-04, 0.0000e+00], |
| [ 3.0637e-04, 2.3097e-07, 8.7786e-04, ..., 2.0489e-06, |
| 1.5116e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 29, bias, value: tensor([-0.0038, 0.0281, 0.0061, 0.0046, 0.0048, -0.0084, 0.0205, -0.0246, |
| 0.0241, 0.0046], device='cuda:0'), grad: tensor([ 5.0974e-04, 1.0605e-03, 1.8349e-03, -2.7809e-03, -2.4204e-03, |
| -1.8835e-03, 1.2693e-03, 7.3528e-04, 8.3804e-05, 1.5898e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 28---------------------------------------------------- |
| epoch 28, time 218.04, cls_loss 0.0366 cls_loss_mapping 0.0434 cls_loss_causal 0.8025 re_mapping 0.0265 re_causal 0.0670 /// teacc 98.47 lr 0.00010000 |
| Epoch 30, weight, value: tensor([[-0.0212, -0.0130, -0.0154, ..., -0.0621, 0.0280, 0.0219], |
| [-0.0031, 0.0109, -0.0637, ..., 0.0105, 0.0080, 0.0126], |
| [ 0.0331, -0.0036, -0.0323, ..., -0.0217, 0.0064, -0.0079], |
| ..., |
| [ 0.0176, -0.0284, 0.0047, ..., 0.0189, 0.0152, 0.0114], |
| [-0.0006, -0.0206, -0.0218, ..., -0.0004, -0.0124, 0.0295], |
| [-0.0269, -0.0667, 0.0405, ..., -0.0392, -0.0221, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 9.5308e-05, 0.0000e+00, 1.9133e-05, ..., 0.0000e+00, |
| 9.1851e-05, 0.0000e+00], |
| [ 1.1027e-04, 0.0000e+00, 1.6570e-05, ..., 0.0000e+00, |
| -6.4299e-06, 0.0000e+00], |
| [ 4.9210e-04, 0.0000e+00, 2.5392e-05, ..., 0.0000e+00, |
| 3.9411e-04, 0.0000e+00], |
| ..., |
| [-1.8203e-04, 0.0000e+00, -7.4729e-06, ..., 0.0000e+00, |
| 2.4661e-05, 0.0000e+00], |
| [ 2.7347e-04, 0.0000e+00, 1.6868e-04, ..., 0.0000e+00, |
| -6.8092e-04, 0.0000e+00], |
| [-1.5056e-04, 0.0000e+00, -3.8624e-04, ..., 0.0000e+00, |
| -5.8889e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 30, bias, value: tensor([-0.0039, 0.0277, 0.0064, 0.0043, 0.0051, -0.0079, 0.0200, -0.0249, |
| 0.0242, 0.0047], device='cuda:0'), grad: tensor([ 2.2161e-04, 1.6427e-04, 1.3714e-03, -5.7745e-04, 8.2445e-04, |
| -5.4550e-04, 4.0054e-05, -2.6464e-04, -6.6137e-04, -5.7173e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 29, time 217.56, cls_loss 0.0332 cls_loss_mapping 0.0423 cls_loss_causal 0.8212 re_mapping 0.0253 re_causal 0.0703 /// teacc 98.34 lr 0.00010000 |
| Epoch 31, weight, value: tensor([[-0.0215, -0.0131, -0.0149, ..., -0.0626, 0.0278, 0.0219], |
| [-0.0034, 0.0115, -0.0641, ..., 0.0104, 0.0086, 0.0126], |
| [ 0.0336, -0.0041, -0.0330, ..., -0.0219, 0.0061, -0.0079], |
| ..., |
| [ 0.0178, -0.0280, 0.0046, ..., 0.0190, 0.0150, 0.0114], |
| [-0.0012, -0.0212, -0.0222, ..., -0.0007, -0.0124, 0.0295], |
| [-0.0278, -0.0695, 0.0408, ..., -0.0395, -0.0219, -0.0004]], |
| device='cuda:0'), grad: tensor([[-7.0989e-05, 3.7253e-08, -8.7738e-05, ..., 3.7439e-07, |
| 3.6150e-05, 0.0000e+00], |
| [ 2.2411e-04, -1.8403e-06, 1.5962e-04, ..., 1.1474e-06, |
| 9.7752e-05, 0.0000e+00], |
| [-2.5868e-04, 4.3772e-07, 5.0992e-05, ..., 6.6496e-07, |
| 6.1274e-05, 0.0000e+00], |
| ..., |
| [-6.8069e-05, -5.2527e-07, -2.1726e-05, ..., -1.6779e-05, |
| 9.8199e-06, 0.0000e+00], |
| [ 2.5764e-05, 1.5832e-07, 5.1200e-05, ..., 2.1979e-07, |
| 3.4332e-05, 0.0000e+00], |
| [ 7.9811e-05, 3.2037e-07, 5.6028e-04, ..., 1.0386e-05, |
| 6.8806e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 31, bias, value: tensor([-0.0038, 0.0277, 0.0066, 0.0045, 0.0050, -0.0076, 0.0199, -0.0249, |
| 0.0237, 0.0048], device='cuda:0'), grad: tensor([-2.2163e-03, 8.8215e-04, 1.9491e-05, 1.6749e-04, -8.3923e-04, |
| 8.0490e-04, -4.4894e-04, 1.7673e-05, 3.0446e-04, 1.3094e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 30, time 217.41, cls_loss 0.0350 cls_loss_mapping 0.0488 cls_loss_causal 0.7968 re_mapping 0.0245 re_causal 0.0674 /// teacc 97.97 lr 0.00010000 |
| Epoch 32, weight, value: tensor([[-0.0212, -0.0132, -0.0150, ..., -0.0627, 0.0275, 0.0219], |
| [-0.0046, 0.0114, -0.0647, ..., 0.0103, 0.0087, 0.0126], |
| [ 0.0341, -0.0038, -0.0330, ..., -0.0221, 0.0058, -0.0079], |
| ..., |
| [ 0.0181, -0.0281, 0.0045, ..., 0.0193, 0.0153, 0.0114], |
| [-0.0012, -0.0216, -0.0222, ..., -0.0006, -0.0122, 0.0295], |
| [-0.0295, -0.0710, 0.0408, ..., -0.0396, -0.0222, -0.0004]], |
| device='cuda:0'), grad: tensor([[ 4.1604e-05, 2.1793e-07, 3.0303e-04, ..., 0.0000e+00, |
| 1.8805e-05, 0.0000e+00], |
| [ 6.8367e-05, 1.9558e-06, -6.0940e-04, ..., 0.0000e+00, |
| -3.4356e-04, 0.0000e+00], |
| [-2.9182e-04, 2.5127e-06, 1.1837e-04, ..., 0.0000e+00, |
| -9.7334e-05, 0.0000e+00], |
| ..., |
| [-1.2648e-04, -9.6485e-06, 2.4891e-04, ..., 0.0000e+00, |
| 1.1402e-04, 0.0000e+00], |
| [ 1.9145e-04, 6.1281e-07, 4.1699e-04, ..., 0.0000e+00, |
| 1.5914e-04, 0.0000e+00], |
| [-8.3590e-04, 4.8243e-07, -1.3245e-02, ..., 0.0000e+00, |
| -1.0490e-03, 0.0000e+00]], device='cuda:0') |
| Epoch 32, bias, value: tensor([-0.0040, 0.0270, 0.0071, 0.0045, 0.0047, -0.0075, 0.0204, -0.0248, |
| 0.0241, 0.0044], device='cuda:0'), grad: tensor([ 0.0004, -0.0014, -0.0006, 0.0012, 0.0153, 0.0024, 0.0020, 0.0004, |
| 0.0011, -0.0207], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 31, time 217.14, cls_loss 0.0289 cls_loss_mapping 0.0428 cls_loss_causal 0.8166 re_mapping 0.0234 re_causal 0.0661 /// teacc 97.91 lr 0.00010000 |
| Epoch 33, weight, value: tensor([[-0.0214, -0.0132, -0.0152, ..., -0.0628, 0.0274, 0.0219], |
| [-0.0048, 0.0114, -0.0644, ..., 0.0109, 0.0093, 0.0126], |
| [ 0.0341, -0.0038, -0.0331, ..., -0.0221, 0.0053, -0.0079], |
| ..., |
| [ 0.0187, -0.0280, 0.0044, ..., 0.0194, 0.0150, 0.0114], |
| [-0.0014, -0.0218, -0.0225, ..., -0.0006, -0.0120, 0.0295], |
| [-0.0298, -0.0713, 0.0406, ..., -0.0400, -0.0220, -0.0004]], |
| device='cuda:0'), grad: tensor([[-6.7186e-04, 0.0000e+00, 6.1020e-06, ..., 0.0000e+00, |
| -2.2709e-04, 0.0000e+00], |
| [-1.3947e-04, 0.0000e+00, 2.1249e-05, ..., 0.0000e+00, |
| -7.4196e-04, 0.0000e+00], |
| [ 6.3801e-04, 0.0000e+00, 1.8165e-05, ..., 0.0000e+00, |
| 3.7980e-04, 0.0000e+00], |
| ..., |
| [-5.5361e-04, 0.0000e+00, 2.7210e-05, ..., 0.0000e+00, |
| 7.0453e-05, 0.0000e+00], |
| [ 6.3300e-05, 0.0000e+00, -5.3346e-05, ..., 0.0000e+00, |
| 1.2684e-04, 0.0000e+00], |
| [ 1.3009e-05, 0.0000e+00, -1.7568e-05, ..., 0.0000e+00, |
| -8.2791e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 33, bias, value: tensor([-0.0037, 0.0273, 0.0070, 0.0044, 0.0052, -0.0077, 0.0198, -0.0251, |
| 0.0244, 0.0042], device='cuda:0'), grad: tensor([-4.0855e-03, -8.3303e-04, 1.0691e-03, 1.0443e-03, 2.8133e-04, |
| -1.4186e-04, 2.7294e-03, -1.8871e-04, 6.5625e-05, 5.9724e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 32, time 217.28, cls_loss 0.0318 cls_loss_mapping 0.0407 cls_loss_causal 0.7904 re_mapping 0.0228 re_causal 0.0645 /// teacc 98.26 lr 0.00010000 |
| Epoch 34, weight, value: tensor([[-0.0207, -0.0132, -0.0152, ..., -0.0629, 0.0278, 0.0213], |
| [-0.0056, 0.0116, -0.0641, ..., 0.0116, 0.0101, 0.0051], |
| [ 0.0342, -0.0039, -0.0335, ..., -0.0221, 0.0048, -0.0112], |
| ..., |
| [ 0.0189, -0.0280, 0.0043, ..., 0.0194, 0.0145, 0.0135], |
| [-0.0010, -0.0219, -0.0227, ..., -0.0006, -0.0121, 0.0260], |
| [-0.0304, -0.0714, 0.0409, ..., -0.0401, -0.0215, -0.0013]], |
| device='cuda:0'), grad: tensor([[ 1.5482e-05, 0.0000e+00, 3.5241e-06, ..., 0.0000e+00, |
| 1.1146e-05, 0.0000e+00], |
| [ 1.8328e-05, 0.0000e+00, 1.4216e-05, ..., 0.0000e+00, |
| -2.6524e-05, 0.0000e+00], |
| [-1.5581e-04, 0.0000e+00, 6.1542e-06, ..., 0.0000e+00, |
| -5.9605e-08, 0.0000e+00], |
| ..., |
| [ 3.0220e-05, 0.0000e+00, 1.3852e-04, ..., 0.0000e+00, |
| 1.9014e-04, 0.0000e+00], |
| [ 1.1468e-04, 0.0000e+00, 1.1104e-04, ..., 0.0000e+00, |
| 7.5102e-05, 0.0000e+00], |
| [ 5.3495e-05, 0.0000e+00, -2.1905e-05, ..., 0.0000e+00, |
| -2.3496e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 34, bias, value: tensor([-0.0036, 0.0275, 0.0067, 0.0043, 0.0046, -0.0077, 0.0200, -0.0251, |
| 0.0245, 0.0046], device='cuda:0'), grad: tensor([-2.5272e-05, 1.2353e-05, -2.8682e-04, 1.2231e-04, -2.1112e-04, |
| -3.1734e-04, -3.1590e-05, 4.6444e-04, 4.1318e-04, -1.4067e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 33, time 217.36, cls_loss 0.0300 cls_loss_mapping 0.0402 cls_loss_causal 0.7865 re_mapping 0.0223 re_causal 0.0598 /// teacc 98.47 lr 0.00010000 |
| Epoch 35, weight, value: tensor([[-0.0213, -0.0133, -0.0151, ..., -0.0633, 0.0277, 0.0212], |
| [-0.0064, 0.0115, -0.0647, ..., 0.0116, 0.0104, 0.0050], |
| [ 0.0345, -0.0037, -0.0340, ..., -0.0221, 0.0044, -0.0113], |
| ..., |
| [ 0.0197, -0.0281, 0.0043, ..., 0.0200, 0.0144, 0.0136], |
| [-0.0013, -0.0219, -0.0228, ..., -0.0007, -0.0119, 0.0259], |
| [-0.0314, -0.0715, 0.0409, ..., -0.0402, -0.0214, -0.0013]], |
| device='cuda:0'), grad: tensor([[ 5.9217e-05, 0.0000e+00, 8.9183e-06, ..., 3.8184e-07, |
| 4.2111e-05, 0.0000e+00], |
| [ 6.6757e-05, 0.0000e+00, 1.1727e-05, ..., 3.6694e-07, |
| -6.7115e-05, 0.0000e+00], |
| [ 6.3002e-05, 0.0000e+00, 1.0349e-05, ..., 9.0338e-07, |
| 3.3230e-05, 0.0000e+00], |
| ..., |
| [-3.2973e-04, 0.0000e+00, 1.2197e-05, ..., 1.5423e-06, |
| 8.5086e-06, 0.0000e+00], |
| [ 8.0228e-05, 0.0000e+00, 3.7044e-05, ..., 6.1430e-06, |
| 3.6687e-05, 0.0000e+00], |
| [ 1.8787e-04, 0.0000e+00, 3.0732e-04, ..., 4.8615e-06, |
| -6.1095e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 35, bias, value: tensor([-0.0034, 0.0271, 0.0068, 0.0039, 0.0048, -0.0075, 0.0202, -0.0246, |
| 0.0249, 0.0040], device='cuda:0'), grad: tensor([ 1.2267e-04, -1.3083e-05, 1.1319e-04, -8.4782e-04, -4.7731e-04, |
| 7.5340e-04, -6.8307e-05, -3.9458e-04, 1.8954e-04, 6.2323e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 34, time 217.43, cls_loss 0.0290 cls_loss_mapping 0.0377 cls_loss_causal 0.7935 re_mapping 0.0221 re_causal 0.0606 /// teacc 98.46 lr 0.00010000 |
| Epoch 36, weight, value: tensor([[-0.0215, -0.0133, -0.0156, ..., -0.0633, 0.0272, 0.0212], |
| [-0.0070, 0.0115, -0.0653, ..., 0.0117, 0.0107, 0.0012], |
| [ 0.0346, -0.0037, -0.0343, ..., -0.0222, 0.0038, -0.0143], |
| ..., |
| [ 0.0203, -0.0281, 0.0042, ..., 0.0201, 0.0145, 0.0167], |
| [-0.0013, -0.0220, -0.0233, ..., -0.0007, -0.0117, 0.0255], |
| [-0.0322, -0.0715, 0.0413, ..., -0.0402, -0.0213, -0.0013]], |
| device='cuda:0'), grad: tensor([[-9.2566e-05, 0.0000e+00, -6.7890e-05, ..., 0.0000e+00, |
| -1.5020e-04, 2.5146e-07], |
| [ 1.9088e-05, 0.0000e+00, 8.3447e-06, ..., 0.0000e+00, |
| -1.4699e-04, 5.5507e-07], |
| [-8.8394e-05, 0.0000e+00, 1.3910e-05, ..., 0.0000e+00, |
| 3.2872e-05, 3.9116e-08], |
| ..., |
| [-5.6922e-06, 0.0000e+00, 3.2634e-05, ..., 0.0000e+00, |
| 1.2732e-04, -3.9786e-06], |
| [ 1.0467e-04, 0.0000e+00, 2.1607e-05, ..., 0.0000e+00, |
| 1.4806e-04, 5.2154e-08], |
| [ 1.1663e-03, 0.0000e+00, 5.0694e-05, ..., 0.0000e+00, |
| 1.3676e-03, 2.2352e-06]], device='cuda:0') |
| Epoch 36, bias, value: tensor([-0.0035, 0.0269, 0.0066, 0.0038, 0.0050, -0.0077, 0.0205, -0.0244, |
| 0.0250, 0.0038], device='cuda:0'), grad: tensor([-7.6962e-04, -1.7488e-04, -9.6112e-06, -2.7161e-03, 1.9491e-05, |
| 3.1209e-04, 2.2793e-04, 2.0015e-04, 3.3307e-04, 2.5768e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 35---------------------------------------------------- |
| epoch 35, time 218.07, cls_loss 0.0257 cls_loss_mapping 0.0350 cls_loss_causal 0.7962 re_mapping 0.0212 re_causal 0.0601 /// teacc 98.51 lr 0.00010000 |
| Epoch 37, weight, value: tensor([[-0.0213, -0.0133, -0.0156, ..., -0.0634, 0.0269, 0.0212], |
| [-0.0077, 0.0115, -0.0658, ..., 0.0117, 0.0118, -0.0022], |
| [ 0.0351, -0.0037, -0.0346, ..., -0.0222, 0.0034, -0.0154], |
| ..., |
| [ 0.0205, -0.0281, 0.0044, ..., 0.0203, 0.0138, 0.0195], |
| [-0.0016, -0.0220, -0.0236, ..., -0.0007, -0.0118, 0.0251], |
| [-0.0328, -0.0715, 0.0414, ..., -0.0403, -0.0207, -0.0014]], |
| device='cuda:0'), grad: tensor([[ 1.1921e-04, 0.0000e+00, 2.6897e-05, ..., 0.0000e+00, |
| 4.2379e-05, 0.0000e+00], |
| [ 1.5955e-03, 0.0000e+00, 5.2512e-05, ..., 0.0000e+00, |
| 4.6206e-04, 0.0000e+00], |
| [-2.1667e-03, 0.0000e+00, 1.1832e-05, ..., 0.0000e+00, |
| -4.3035e-04, 0.0000e+00], |
| ..., |
| [ 2.1863e-04, 0.0000e+00, 5.2023e-04, ..., 0.0000e+00, |
| 3.4404e-04, 0.0000e+00], |
| [ 5.0497e-04, 0.0000e+00, 2.1124e-04, ..., 0.0000e+00, |
| 1.7059e-04, 0.0000e+00], |
| [-2.3887e-05, 0.0000e+00, -9.7942e-04, ..., 0.0000e+00, |
| -7.6103e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 37, bias, value: tensor([-0.0031, 0.0273, 0.0069, 0.0038, 0.0047, -0.0076, 0.0200, -0.0246, |
| 0.0248, 0.0040], device='cuda:0'), grad: tensor([ 0.0002, 0.0050, -0.0057, 0.0034, 0.0006, -0.0041, 0.0001, 0.0013, |
| 0.0011, -0.0020], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 36---------------------------------------------------- |
| epoch 36, time 218.13, cls_loss 0.0268 cls_loss_mapping 0.0352 cls_loss_causal 0.7471 re_mapping 0.0214 re_causal 0.0572 /// teacc 98.63 lr 0.00010000 |
| Epoch 38, weight, value: tensor([[-0.0217, -0.0133, -0.0153, ..., -0.0639, 0.0265, 0.0212], |
| [-0.0086, 0.0113, -0.0664, ..., 0.0117, 0.0120, -0.0027], |
| [ 0.0356, -0.0033, -0.0352, ..., -0.0223, 0.0031, -0.0160], |
| ..., |
| [ 0.0208, -0.0283, 0.0040, ..., 0.0208, 0.0135, 0.0202], |
| [-0.0017, -0.0222, -0.0241, ..., -0.0008, -0.0119, 0.0249], |
| [-0.0337, -0.0716, 0.0412, ..., -0.0407, -0.0203, -0.0014]], |
| device='cuda:0'), grad: tensor([[ 4.2975e-05, 0.0000e+00, 3.8356e-05, ..., 0.0000e+00, |
| 1.5348e-05, 0.0000e+00], |
| [ 1.4806e-04, 0.0000e+00, 4.3094e-05, ..., 0.0000e+00, |
| -3.4600e-05, 0.0000e+00], |
| [ 1.9860e-04, 0.0000e+00, 6.9320e-05, ..., 0.0000e+00, |
| 7.6115e-05, 0.0000e+00], |
| ..., |
| [-9.0742e-04, 0.0000e+00, -1.0812e-04, ..., 0.0000e+00, |
| -7.3731e-05, 0.0000e+00], |
| [ 1.5688e-04, 0.0000e+00, 2.8682e-04, ..., 0.0000e+00, |
| 1.5330e-04, 0.0000e+00], |
| [-4.9829e-05, 0.0000e+00, -3.7217e-04, ..., 0.0000e+00, |
| -2.8253e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 38, bias, value: tensor([-0.0029, 0.0267, 0.0069, 0.0039, 0.0051, -0.0077, 0.0204, -0.0243, |
| 0.0246, 0.0036], device='cuda:0'), grad: tensor([ 5.4747e-05, 2.1505e-04, 3.8195e-04, 6.9141e-04, 2.5535e-04, |
| -5.3883e-04, 2.3174e-04, -1.3437e-03, 6.3276e-04, -5.8079e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 37, time 217.49, cls_loss 0.0266 cls_loss_mapping 0.0349 cls_loss_causal 0.7600 re_mapping 0.0200 re_causal 0.0567 /// teacc 98.43 lr 0.00010000 |
| Epoch 39, weight, value: tensor([[-0.0225, -0.0134, -0.0157, ..., -0.0642, 0.0258, 0.0212], |
| [-0.0089, 0.0112, -0.0672, ..., 0.0116, 0.0124, -0.0028], |
| [ 0.0359, -0.0031, -0.0360, ..., -0.0225, 0.0028, -0.0160], |
| ..., |
| [ 0.0210, -0.0286, 0.0036, ..., 0.0205, 0.0130, 0.0203], |
| [-0.0019, -0.0223, -0.0243, ..., -0.0009, -0.0117, 0.0249], |
| [-0.0341, -0.0718, 0.0415, ..., -0.0407, -0.0194, -0.0014]], |
| device='cuda:0'), grad: tensor([[-2.5177e-04, 0.0000e+00, -1.1501e-03, ..., 7.4506e-08, |
| -1.1665e-04, 0.0000e+00], |
| [ 1.7837e-05, 0.0000e+00, 6.3360e-05, ..., 1.5087e-07, |
| -1.1736e-04, 0.0000e+00], |
| [-5.0992e-05, 0.0000e+00, 1.1599e-04, ..., 2.6450e-07, |
| 2.2128e-05, 0.0000e+00], |
| ..., |
| [ 7.1466e-05, 0.0000e+00, 2.2244e-04, ..., 7.7859e-07, |
| 1.9407e-04, 0.0000e+00], |
| [ 1.2863e-04, 0.0000e+00, 3.6979e-04, ..., 1.5087e-07, |
| 2.0814e-04, 0.0000e+00], |
| [ 7.9632e-05, 0.0000e+00, 9.6703e-04, ..., 2.8312e-07, |
| -3.3236e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 39, bias, value: tensor([-0.0038, 0.0265, 0.0064, 0.0037, 0.0051, -0.0071, 0.0204, -0.0245, |
| 0.0249, 0.0043], device='cuda:0'), grad: tensor([-2.2202e-03, -3.1859e-05, 1.6952e-04, -2.0468e-04, -1.1339e-03, |
| 4.4107e-04, 2.9254e-04, 6.3801e-04, 8.7500e-04, 1.1740e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 38, time 217.23, cls_loss 0.0263 cls_loss_mapping 0.0371 cls_loss_causal 0.7553 re_mapping 0.0206 re_causal 0.0557 /// teacc 98.47 lr 0.00010000 |
| Epoch 40, weight, value: tensor([[-0.0214, -0.0135, -0.0156, ..., -0.0643, 0.0260, 0.0212], |
| [-0.0097, 0.0114, -0.0677, ..., 0.0116, 0.0128, -0.0036], |
| [ 0.0359, -0.0032, -0.0363, ..., -0.0226, 0.0023, -0.0161], |
| ..., |
| [ 0.0220, -0.0287, 0.0035, ..., 0.0212, 0.0127, 0.0204], |
| [-0.0022, -0.0224, -0.0251, ..., -0.0010, -0.0117, 0.0249], |
| [-0.0352, -0.0721, 0.0420, ..., -0.0409, -0.0191, -0.0014]], |
| device='cuda:0'), grad: tensor([[ 7.6145e-06, 0.0000e+00, 3.9190e-05, ..., 0.0000e+00, |
| -6.0111e-05, 0.0000e+00], |
| [ 5.1069e-04, 0.0000e+00, 3.2663e-05, ..., 0.0000e+00, |
| 1.2982e-04, 0.0000e+00], |
| [ 5.7936e-04, 0.0000e+00, 1.4983e-05, ..., 0.0000e+00, |
| 4.1693e-05, 0.0000e+00], |
| ..., |
| [-1.4238e-03, 0.0000e+00, 2.0385e-04, ..., 0.0000e+00, |
| 1.4901e-07, 0.0000e+00], |
| [ 1.3757e-04, 0.0000e+00, 1.2898e-04, ..., 0.0000e+00, |
| 2.5320e-04, 0.0000e+00], |
| [ 2.9755e-04, 0.0000e+00, -1.0824e-03, ..., 0.0000e+00, |
| 3.3617e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 40, bias, value: tensor([-0.0032, 0.0264, 0.0060, 0.0038, 0.0051, -0.0071, 0.0203, -0.0242, |
| 0.0246, 0.0043], device='cuda:0'), grad: tensor([-0.0004, 0.0009, 0.0009, -0.0002, 0.0012, 0.0003, -0.0004, -0.0015, |
| 0.0009, -0.0017], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 39, time 217.27, cls_loss 0.0205 cls_loss_mapping 0.0267 cls_loss_causal 0.7271 re_mapping 0.0203 re_causal 0.0569 /// teacc 98.48 lr 0.00010000 |
| Epoch 41, weight, value: tensor([[-0.0212, -0.0136, -0.0148, ..., -0.0649, 0.0257, 0.0211], |
| [-0.0100, 0.0114, -0.0680, ..., 0.0118, 0.0136, -0.0063], |
| [ 0.0361, -0.0032, -0.0363, ..., -0.0230, 0.0018, -0.0191], |
| ..., |
| [ 0.0217, -0.0287, 0.0034, ..., 0.0215, 0.0125, 0.0233], |
| [-0.0023, -0.0225, -0.0255, ..., -0.0010, -0.0118, 0.0228], |
| [-0.0360, -0.0722, 0.0422, ..., -0.0411, -0.0187, -0.0014]], |
| device='cuda:0'), grad: tensor([[ 7.8857e-05, 0.0000e+00, 2.6524e-05, ..., 0.0000e+00, |
| 7.1645e-05, 7.6927e-07], |
| [ 4.2826e-05, 0.0000e+00, 3.3021e-05, ..., 0.0000e+00, |
| 4.4823e-05, 8.7917e-07], |
| [-2.6917e-04, 0.0000e+00, 2.1899e-04, ..., 0.0000e+00, |
| 6.7353e-05, 1.0021e-05], |
| ..., |
| [-6.6566e-04, 0.0000e+00, -1.1083e-06, ..., 0.0000e+00, |
| -4.2379e-05, -2.4244e-05], |
| [ 1.4257e-04, 0.0000e+00, 6.3479e-06, ..., 0.0000e+00, |
| 2.2042e-04, 6.6869e-07], |
| [ 1.9717e-04, 0.0000e+00, -5.4455e-04, ..., 0.0000e+00, |
| -2.4307e-04, 8.1956e-07]], device='cuda:0') |
| Epoch 41, bias, value: tensor([-0.0029, 0.0266, 0.0060, 0.0037, 0.0049, -0.0067, 0.0204, -0.0242, |
| 0.0243, 0.0041], device='cuda:0'), grad: tensor([ 1.5390e-04, 1.7715e-04, 1.4389e-04, 8.8358e-04, 2.3401e-04, |
| 1.8346e-04, -3.2663e-04, -9.5177e-04, -5.2661e-05, -4.4346e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 40, time 217.22, cls_loss 0.0278 cls_loss_mapping 0.0355 cls_loss_causal 0.7402 re_mapping 0.0194 re_causal 0.0535 /// teacc 98.50 lr 0.00010000 |
| Epoch 42, weight, value: tensor([[-0.0217, -0.0137, -0.0155, ..., -0.0652, 0.0252, 0.0212], |
| [-0.0104, 0.0124, -0.0687, ..., 0.0118, 0.0139, -0.0082], |
| [ 0.0365, -0.0038, -0.0374, ..., -0.0231, 0.0014, -0.0194], |
| ..., |
| [ 0.0220, -0.0299, 0.0028, ..., 0.0216, 0.0117, 0.0233], |
| [-0.0028, -0.0230, -0.0258, ..., -0.0010, -0.0123, 0.0218], |
| [-0.0360, -0.0725, 0.0420, ..., -0.0412, -0.0172, -0.0015]], |
| device='cuda:0'), grad: tensor([[ 1.2231e-04, 0.0000e+00, 1.9178e-05, ..., 2.5705e-07, |
| 3.7551e-05, 0.0000e+00], |
| [-2.3580e-04, 0.0000e+00, -1.3723e-03, ..., 5.2713e-07, |
| -1.8854e-03, 0.0000e+00], |
| [ 7.8058e-04, 0.0000e+00, 8.6844e-05, ..., 8.9034e-07, |
| 8.6546e-05, 0.0000e+00], |
| ..., |
| [ 2.9421e-04, 0.0000e+00, 1.8239e-04, ..., -4.6901e-06, |
| 1.4985e-04, 0.0000e+00], |
| [ 2.9325e-04, 0.0000e+00, 1.4365e-04, ..., 9.1270e-08, |
| 1.1625e-03, 0.0000e+00], |
| [ 4.1318e-04, 0.0000e+00, 1.2839e-04, ..., 3.1665e-07, |
| 3.6240e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 42, bias, value: tensor([-0.0038, 0.0265, 0.0060, 0.0041, 0.0053, -0.0069, 0.0206, -0.0245, |
| 0.0242, 0.0043], device='cuda:0'), grad: tensor([ 0.0002, -0.0046, 0.0013, -0.0137, 0.0022, 0.0096, 0.0004, 0.0008, |
| 0.0024, 0.0013], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 41, time 217.42, cls_loss 0.0263 cls_loss_mapping 0.0359 cls_loss_causal 0.7653 re_mapping 0.0191 re_causal 0.0547 /// teacc 98.61 lr 0.00010000 |
| Epoch 43, weight, value: tensor([[-0.0218, -0.0137, -0.0155, ..., -0.0653, 0.0249, 0.0212], |
| [-0.0102, 0.0125, -0.0683, ..., 0.0121, 0.0146, -0.0108], |
| [ 0.0366, -0.0037, -0.0374, ..., -0.0231, 0.0007, -0.0196], |
| ..., |
| [ 0.0223, -0.0300, 0.0030, ..., 0.0220, 0.0110, 0.0240], |
| [-0.0028, -0.0232, -0.0262, ..., -0.0011, -0.0122, 0.0206], |
| [-0.0373, -0.0726, 0.0423, ..., -0.0409, -0.0166, -0.0016]], |
| device='cuda:0'), grad: tensor([[ 1.1790e-04, 0.0000e+00, -1.4424e-04, ..., 7.0035e-07, |
| -5.1409e-05, 1.6764e-08], |
| [ 5.7489e-05, 0.0000e+00, 4.3094e-05, ..., 3.3993e-06, |
| -4.2230e-05, 1.1176e-08], |
| [-2.4452e-03, 0.0000e+00, -6.9332e-04, ..., 2.9132e-06, |
| 2.3738e-05, -4.3213e-07], |
| ..., |
| [ 1.6584e-03, 0.0000e+00, 1.9569e-03, ..., 1.1140e-04, |
| 2.1420e-07, 2.7008e-07], |
| [ 1.0443e-04, 0.0000e+00, 4.3392e-05, ..., 1.0524e-06, |
| 6.2585e-05, 5.9605e-08], |
| [ 1.3113e-04, 0.0000e+00, 5.0783e-04, ..., 5.1931e-06, |
| 3.3259e-05, 1.8626e-09]], device='cuda:0') |
| Epoch 43, bias, value: tensor([-0.0035, 0.0271, 0.0062, 0.0038, 0.0049, -0.0061, 0.0194, -0.0246, |
| 0.0241, 0.0045], device='cuda:0'), grad: tensor([-5.6553e-04, 4.1902e-05, -3.3340e-03, 4.6396e-04, -1.9264e-03, |
| 7.8506e-03, -7.3128e-03, 3.6144e-03, 2.8491e-04, 8.7690e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 42, time 217.23, cls_loss 0.0251 cls_loss_mapping 0.0325 cls_loss_causal 0.7533 re_mapping 0.0186 re_causal 0.0523 /// teacc 98.16 lr 0.00010000 |
| Epoch 44, weight, value: tensor([[-0.0216, -0.0139, -0.0156, ..., -0.0655, 0.0245, 0.0213], |
| [-0.0109, 0.0137, -0.0686, ..., 0.0123, 0.0155, -0.0119], |
| [ 0.0368, -0.0045, -0.0379, ..., -0.0232, 0.0003, -0.0195], |
| ..., |
| [ 0.0226, -0.0307, 0.0028, ..., 0.0224, 0.0106, 0.0241], |
| [-0.0035, -0.0238, -0.0263, ..., -0.0011, -0.0123, 0.0202], |
| [-0.0377, -0.0730, 0.0421, ..., -0.0409, -0.0160, -0.0016]], |
| device='cuda:0'), grad: tensor([[-1.4746e-04, 0.0000e+00, 1.7494e-05, ..., 1.8626e-09, |
| 1.7494e-05, 1.1362e-07], |
| [ 3.8475e-05, 0.0000e+00, 2.7761e-05, ..., -2.1979e-07, |
| -1.2405e-05, 3.2410e-07], |
| [ 3.8415e-05, 0.0000e+00, 1.3888e-05, ..., 1.8626e-08, |
| 1.3970e-05, 3.3528e-08], |
| ..., |
| [ 5.1558e-05, 0.0000e+00, 9.4235e-05, ..., 1.1735e-07, |
| 1.3721e-04, -5.2080e-06], |
| [-4.7445e-04, 0.0000e+00, -2.2697e-04, ..., 3.3528e-08, |
| -2.0921e-05, 9.4995e-08], |
| [-5.0604e-05, 0.0000e+00, -3.7098e-04, ..., 1.3039e-08, |
| -3.2878e-04, 4.0196e-06]], device='cuda:0') |
| Epoch 44, bias, value: tensor([-0.0039, 0.0273, 0.0063, 0.0042, 0.0051, -0.0066, 0.0193, -0.0245, |
| 0.0241, 0.0044], device='cuda:0'), grad: tensor([-0.0003, 0.0001, 0.0003, 0.0012, 0.0005, -0.0001, 0.0003, 0.0008, |
| -0.0011, -0.0016], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 43---------------------------------------------------- |
| epoch 43, time 218.05, cls_loss 0.0216 cls_loss_mapping 0.0296 cls_loss_causal 0.7488 re_mapping 0.0182 re_causal 0.0537 /// teacc 98.70 lr 0.00010000 |
| Epoch 45, weight, value: tensor([[-2.1662e-02, -1.3942e-02, -1.5540e-02, ..., -6.5847e-02, |
| 2.4035e-02, 2.0561e-02], |
| [-1.1763e-02, 1.3926e-02, -6.9201e-02, ..., 1.3015e-02, |
| 1.5959e-02, -1.9042e-02], |
| [ 3.7467e-02, -4.6319e-03, -3.7659e-02, ..., -2.3430e-02, |
| 3.5998e-05, -1.6238e-02], |
| ..., |
| [ 2.2948e-02, -3.1033e-02, 2.3940e-03, ..., 2.2459e-02, |
| 1.0317e-02, 2.4452e-02], |
| [-3.4721e-03, -2.3909e-02, -2.6404e-02, ..., -1.1909e-03, |
| -1.2683e-02, 1.4079e-02], |
| [-3.8343e-02, -7.3186e-02, 4.2162e-02, ..., -4.1136e-02, |
| -1.5362e-02, -1.8112e-03]], device='cuda:0'), grad: tensor([[ 1.6704e-05, 0.0000e+00, 7.9453e-05, ..., 1.8626e-08, |
| 9.8169e-05, -5.2266e-06], |
| [ 5.4359e-05, 0.0000e+00, 2.2650e-06, ..., -2.9579e-06, |
| -9.3818e-05, 4.8522e-07], |
| [ 7.5674e-04, 0.0000e+00, 3.5539e-06, ..., 6.1467e-08, |
| 2.9892e-05, 1.1362e-06], |
| ..., |
| [ 1.8448e-02, 0.0000e+00, -6.4559e-06, ..., -4.6752e-07, |
| 1.9878e-05, 1.4808e-06], |
| [ 1.7715e-04, 0.0000e+00, 7.3731e-05, ..., 1.0990e-06, |
| 1.0115e-04, 4.2468e-07], |
| [ 1.3864e-04, 0.0000e+00, 4.4018e-05, ..., 1.0347e-06, |
| 2.5824e-05, 1.1446e-06]], device='cuda:0') |
| Epoch 45, bias, value: tensor([-0.0041, 0.0270, 0.0068, 0.0040, 0.0051, -0.0067, 0.0195, -0.0247, |
| 0.0241, 0.0047], device='cuda:0'), grad: tensor([ 3.6454e-04, -7.8022e-05, 7.7581e-04, -2.2430e-02, 1.1027e-04, |
| 8.7309e-04, -1.9178e-03, 2.1240e-02, 7.0715e-04, 3.5405e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 44, time 217.20, cls_loss 0.0268 cls_loss_mapping 0.0348 cls_loss_causal 0.7402 re_mapping 0.0183 re_causal 0.0492 /// teacc 98.68 lr 0.00010000 |
| Epoch 46, weight, value: tensor([[-0.0223, -0.0140, -0.0157, ..., -0.0668, 0.0235, 0.0201], |
| [-0.0123, 0.0147, -0.0699, ..., 0.0135, 0.0164, -0.0214], |
| [ 0.0382, -0.0053, -0.0381, ..., -0.0237, -0.0004, -0.0166], |
| ..., |
| [ 0.0234, -0.0312, 0.0023, ..., 0.0235, 0.0099, 0.0263], |
| [-0.0044, -0.0240, -0.0268, ..., -0.0015, -0.0125, 0.0133], |
| [-0.0390, -0.0734, 0.0420, ..., -0.0419, -0.0142, -0.0015]], |
| device='cuda:0'), grad: tensor([[ 1.0669e-05, 0.0000e+00, 4.6045e-06, ..., 2.3190e-07, |
| 1.3091e-05, 0.0000e+00], |
| [ 1.4126e-04, 0.0000e+00, 4.9621e-06, ..., 7.6089e-07, |
| 2.8700e-05, 0.0000e+00], |
| [ 3.1638e-04, 0.0000e+00, 1.7406e-06, ..., 1.5926e-07, |
| 3.5882e-05, 0.0000e+00], |
| ..., |
| [-6.7902e-04, 0.0000e+00, 1.1735e-07, ..., -2.6859e-06, |
| -8.9183e-06, 0.0000e+00], |
| [ 2.8312e-05, 0.0000e+00, 1.7777e-05, ..., 8.1025e-08, |
| 2.1660e-04, 0.0000e+00], |
| [ 6.2466e-05, 0.0000e+00, -4.7497e-06, ..., 2.1793e-07, |
| -2.8229e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 46, bias, value: tensor([-0.0042, 0.0267, 0.0071, 0.0033, 0.0050, -0.0061, 0.0196, -0.0244, |
| 0.0241, 0.0045], device='cuda:0'), grad: tensor([-6.1035e-04, 3.5381e-04, 7.6532e-04, 4.0317e-04, 6.7353e-05, |
| 3.9721e-04, 1.4015e-05, -9.4509e-04, -1.1098e-04, -3.3498e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 45, time 217.33, cls_loss 0.0208 cls_loss_mapping 0.0290 cls_loss_causal 0.7351 re_mapping 0.0180 re_causal 0.0513 /// teacc 98.69 lr 0.00010000 |
| Epoch 47, weight, value: tensor([[-0.0228, -0.0141, -0.0160, ..., -0.0670, 0.0233, 0.0196], |
| [-0.0126, 0.0150, -0.0703, ..., 0.0137, 0.0169, -0.0235], |
| [ 0.0384, -0.0053, -0.0385, ..., -0.0238, -0.0008, -0.0166], |
| ..., |
| [ 0.0239, -0.0319, 0.0024, ..., 0.0238, 0.0096, 0.0267], |
| [-0.0047, -0.0242, -0.0272, ..., -0.0015, -0.0126, 0.0130], |
| [-0.0400, -0.0736, 0.0419, ..., -0.0418, -0.0144, -0.0011]], |
| device='cuda:0'), grad: tensor([[ 8.3447e-06, 1.8626e-08, 1.8269e-05, ..., 1.1269e-07, |
| 1.8924e-05, 1.1893e-06], |
| [ 1.6347e-05, 8.1956e-08, 1.2800e-05, ..., -6.3181e-06, |
| -5.4389e-05, 8.4285e-07], |
| [-1.5616e-05, -6.2399e-07, 4.8965e-05, ..., 1.2694e-06, |
| 4.0561e-05, -6.8657e-06], |
| ..., |
| [-2.3615e-04, 4.6566e-07, 1.6987e-05, ..., 1.6680e-06, |
| -8.1837e-05, -3.3863e-06], |
| [ 3.1978e-05, 1.6764e-08, -4.4346e-05, ..., 5.5321e-07, |
| -6.6936e-05, 4.3400e-06], |
| [ 1.6749e-04, 2.7940e-09, 6.3753e-04, ..., 3.1479e-07, |
| 7.5817e-05, 2.1495e-06]], device='cuda:0') |
| Epoch 47, bias, value: tensor([-0.0041, 0.0264, 0.0074, 0.0033, 0.0052, -0.0056, 0.0195, -0.0239, |
| 0.0235, 0.0040], device='cuda:0'), grad: tensor([-3.1424e-04, 1.0319e-06, 1.4210e-04, 1.0484e-04, -1.3399e-03, |
| 1.2827e-04, 2.6202e-04, -4.3726e-04, -2.3150e-04, 1.6851e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 46, time 217.28, cls_loss 0.0208 cls_loss_mapping 0.0304 cls_loss_causal 0.7148 re_mapping 0.0173 re_causal 0.0485 /// teacc 98.64 lr 0.00010000 |
| Epoch 48, weight, value: tensor([[-0.0234, -0.0142, -0.0162, ..., -0.0672, 0.0230, 0.0196], |
| [-0.0131, 0.0151, -0.0706, ..., 0.0152, 0.0175, -0.0253], |
| [ 0.0389, -0.0050, -0.0388, ..., -0.0239, -0.0016, -0.0169], |
| ..., |
| [ 0.0238, -0.0329, 0.0023, ..., 0.0241, 0.0090, 0.0277], |
| [-0.0052, -0.0244, -0.0277, ..., -0.0017, -0.0124, 0.0122], |
| [-0.0410, -0.0738, 0.0418, ..., -0.0422, -0.0139, -0.0014]], |
| device='cuda:0'), grad: tensor([[ 1.7062e-05, 0.0000e+00, 2.8804e-05, ..., 2.7940e-09, |
| 3.5763e-05, 0.0000e+00], |
| [ 8.6799e-06, 0.0000e+00, 1.0848e-05, ..., 3.2596e-08, |
| -3.9190e-05, 0.0000e+00], |
| [ 4.5896e-05, 0.0000e+00, 3.2723e-05, ..., 1.3970e-08, |
| 2.1017e-04, 0.0000e+00], |
| ..., |
| [ 1.2845e-05, 0.0000e+00, 8.6963e-05, ..., -1.5739e-07, |
| 3.2812e-05, 0.0000e+00], |
| [ 7.5638e-05, 0.0000e+00, -5.3406e-05, ..., 8.3819e-09, |
| -3.9005e-04, 0.0000e+00], |
| [ 1.3161e-04, 0.0000e+00, -6.5625e-05, ..., 6.7987e-08, |
| -5.2527e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 48, bias, value: tensor([-0.0047, 0.0262, 0.0074, 0.0039, 0.0061, -0.0060, 0.0197, -0.0240, |
| 0.0235, 0.0036], device='cuda:0'), grad: tensor([-2.6345e-04, 1.5981e-06, 9.1696e-04, -2.0218e-03, 6.6423e-04, |
| -1.0996e-03, 3.2692e-03, 3.5763e-04, -2.2240e-03, 4.0007e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 47, time 217.62, cls_loss 0.0205 cls_loss_mapping 0.0240 cls_loss_causal 0.7104 re_mapping 0.0168 re_causal 0.0484 /// teacc 98.68 lr 0.00010000 |
| Epoch 49, weight, value: tensor([[-0.0235, -0.0142, -0.0164, ..., -0.0674, 0.0224, 0.0165], |
| [-0.0133, 0.0151, -0.0709, ..., 0.0155, 0.0183, -0.0294], |
| [ 0.0385, -0.0049, -0.0396, ..., -0.0241, -0.0021, -0.0166], |
| ..., |
| [ 0.0241, -0.0333, 0.0022, ..., 0.0244, 0.0084, 0.0283], |
| [-0.0050, -0.0247, -0.0281, ..., -0.0018, -0.0123, 0.0110], |
| [-0.0410, -0.0740, 0.0420, ..., -0.0424, -0.0130, 0.0020]], |
| device='cuda:0'), grad: tensor([[ 3.2723e-05, 0.0000e+00, 6.3896e-05, ..., 0.0000e+00, |
| 2.8729e-05, 5.4110e-07], |
| [ 1.4234e-04, 0.0000e+00, 8.8066e-06, ..., 0.0000e+00, |
| -3.2149e-06, 8.9593e-07], |
| [ 4.6778e-04, 0.0000e+00, 5.3532e-06, ..., 0.0000e+00, |
| 1.2696e-04, 1.5814e-06], |
| ..., |
| [ 7.8559e-05, 0.0000e+00, 9.2626e-05, ..., 0.0000e+00, |
| 1.0234e-04, 1.0051e-05], |
| [ 1.8215e-04, 0.0000e+00, 1.2577e-05, ..., 0.0000e+00, |
| 4.3094e-05, 3.9581e-07], |
| [ 9.0778e-05, 0.0000e+00, -2.1911e-04, ..., 0.0000e+00, |
| -1.9205e-04, -1.6764e-05]], device='cuda:0') |
| Epoch 49, bias, value: tensor([-0.0053, 0.0265, 0.0061, 0.0034, 0.0063, -0.0054, 0.0199, -0.0241, |
| 0.0236, 0.0044], device='cuda:0'), grad: tensor([ 0.0003, 0.0001, 0.0006, -0.0014, 0.0006, -0.0008, 0.0002, 0.0004, |
| 0.0003, -0.0004], device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 48, time 217.58, cls_loss 0.0171 cls_loss_mapping 0.0232 cls_loss_causal 0.7122 re_mapping 0.0174 re_causal 0.0492 /// teacc 98.60 lr 0.00010000 |
| Epoch 50, weight, value: tensor([[-0.0229, -0.0144, -0.0163, ..., -0.0677, 0.0216, 0.0158], |
| [-0.0135, 0.0156, -0.0712, ..., 0.0171, 0.0185, -0.0293], |
| [ 0.0386, -0.0052, -0.0398, ..., -0.0242, -0.0027, -0.0153], |
| ..., |
| [ 0.0244, -0.0335, 0.0021, ..., 0.0249, 0.0083, 0.0310], |
| [-0.0054, -0.0249, -0.0284, ..., -0.0019, -0.0120, 0.0078], |
| [-0.0418, -0.0750, 0.0423, ..., -0.0418, -0.0124, 0.0005]], |
| device='cuda:0'), grad: tensor([[ 5.0142e-06, 0.0000e+00, 6.1020e-06, ..., 2.1420e-08, |
| -1.4156e-05, 9.3132e-10], |
| [-6.5379e-07, 0.0000e+00, 4.9174e-06, ..., -2.3823e-06, |
| -2.0295e-05, 9.3132e-10], |
| [ 6.5267e-06, 0.0000e+00, 1.5005e-05, ..., 1.9372e-07, |
| 1.5408e-05, -1.2107e-08], |
| ..., |
| [ 1.2331e-05, 0.0000e+00, 2.1517e-05, ..., 1.3039e-06, |
| 1.7717e-05, 2.7940e-09], |
| [-9.4116e-05, 0.0000e+00, -2.7895e-05, ..., 1.8533e-07, |
| 8.9705e-05, 2.7940e-09], |
| [-2.3156e-05, 0.0000e+00, -5.6237e-05, ..., 2.6729e-07, |
| -1.2684e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 50, bias, value: tensor([-0.0048, 0.0265, 0.0059, 0.0037, 0.0057, -0.0055, 0.0199, -0.0240, |
| 0.0235, 0.0046], device='cuda:0'), grad: tensor([-4.3488e-04, -5.6811e-06, 1.0598e-04, 9.5308e-05, 4.9978e-05, |
| 4.1246e-04, 2.6131e-04, 1.4341e-04, -4.8876e-04, -1.3745e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 49---------------------------------------------------- |
| epoch 49, time 218.07, cls_loss 0.0172 cls_loss_mapping 0.0231 cls_loss_causal 0.7000 re_mapping 0.0167 re_causal 0.0471 /// teacc 98.72 lr 0.00010000 |
| Epoch 51, weight, value: tensor([[-0.0231, -0.0144, -0.0166, ..., -0.0680, 0.0211, 0.0147], |
| [-0.0139, 0.0156, -0.0716, ..., 0.0181, 0.0188, -0.0318], |
| [ 0.0396, -0.0049, -0.0398, ..., -0.0246, -0.0024, -0.0150], |
| ..., |
| [ 0.0245, -0.0342, 0.0018, ..., 0.0246, 0.0076, 0.0297], |
| [-0.0057, -0.0250, -0.0284, ..., -0.0021, -0.0119, 0.0064], |
| [-0.0426, -0.0752, 0.0422, ..., -0.0421, -0.0116, 0.0011]], |
| device='cuda:0'), grad: tensor([[-8.6594e-04, 0.0000e+00, 1.1921e-05, ..., 0.0000e+00, |
| 5.4687e-06, 0.0000e+00], |
| [ 4.1068e-05, 0.0000e+00, 4.1574e-05, ..., 0.0000e+00, |
| -7.4089e-05, 0.0000e+00], |
| [ 1.1700e-04, 0.0000e+00, 6.5446e-05, ..., 0.0000e+00, |
| 1.3456e-05, -2.7940e-09], |
| ..., |
| [-2.2620e-05, 0.0000e+00, 3.8862e-05, ..., 0.0000e+00, |
| 4.2856e-05, 0.0000e+00], |
| [ 7.3814e-04, 0.0000e+00, -8.7082e-05, ..., 0.0000e+00, |
| 2.8610e-05, 9.3132e-10], |
| [-2.6718e-05, 0.0000e+00, 5.5805e-06, ..., 0.0000e+00, |
| -1.3721e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 51, bias, value: tensor([-0.0046, 0.0264, 0.0070, 0.0030, 0.0056, -0.0055, 0.0200, -0.0245, |
| 0.0235, 0.0046], device='cuda:0'), grad: tensor([-5.1727e-03, 7.4387e-05, 6.0892e-04, 3.2425e-03, 3.1328e-04, |
| -3.0842e-03, 1.1027e-04, 3.2401e-04, 3.7804e-03, -1.9515e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 50, time 217.49, cls_loss 0.0194 cls_loss_mapping 0.0260 cls_loss_causal 0.6824 re_mapping 0.0163 re_causal 0.0452 /// teacc 98.56 lr 0.00010000 |
| Epoch 52, weight, value: tensor([[-0.0237, -0.0151, -0.0168, ..., -0.0683, 0.0208, 0.0146], |
| [-0.0139, 0.0150, -0.0722, ..., 0.0181, 0.0196, -0.0351], |
| [ 0.0401, -0.0038, -0.0403, ..., -0.0247, -0.0028, -0.0152], |
| ..., |
| [ 0.0245, -0.0348, 0.0019, ..., 0.0255, 0.0076, 0.0301], |
| [-0.0063, -0.0263, -0.0286, ..., -0.0021, -0.0122, 0.0049], |
| [-0.0436, -0.0768, 0.0415, ..., -0.0425, -0.0117, 0.0009]], |
| device='cuda:0'), grad: tensor([[-6.1452e-05, 0.0000e+00, -4.2200e-05, ..., 0.0000e+00, |
| 7.9721e-06, 1.6764e-08], |
| [ 4.0352e-05, 0.0000e+00, 2.9981e-05, ..., 0.0000e+00, |
| 8.1509e-06, 3.9116e-08], |
| [ 2.7388e-05, 0.0000e+00, 1.7679e-04, ..., 0.0000e+00, |
| 1.1191e-05, -3.2131e-07], |
| ..., |
| [-3.9756e-05, 0.0000e+00, 3.2693e-05, ..., 0.0000e+00, |
| 5.7667e-05, 5.7742e-08], |
| [ 1.0622e-04, 0.0000e+00, 1.6844e-04, ..., 0.0000e+00, |
| -1.8692e-04, 5.2154e-08], |
| [ 3.5346e-05, 0.0000e+00, 1.6868e-05, ..., 0.0000e+00, |
| -8.7202e-05, 3.7253e-09]], device='cuda:0') |
| Epoch 52, bias, value: tensor([-0.0044, 0.0265, 0.0073, 0.0034, 0.0061, -0.0054, 0.0201, -0.0242, |
| 0.0230, 0.0035], device='cuda:0'), grad: tensor([-2.1553e-04, 1.7560e-04, 1.6892e-04, 1.2817e-03, -1.4615e-04, |
| -3.2043e-03, 1.0386e-03, 1.3685e-04, 6.8235e-04, 8.4579e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 51, time 217.38, cls_loss 0.0181 cls_loss_mapping 0.0232 cls_loss_causal 0.6852 re_mapping 0.0160 re_causal 0.0452 /// teacc 98.68 lr 0.00010000 |
| Epoch 53, weight, value: tensor([[-0.0240, -0.0152, -0.0172, ..., -0.0684, 0.0205, 0.0145], |
| [-0.0140, 0.0152, -0.0721, ..., 0.0189, 0.0205, -0.0359], |
| [ 0.0398, -0.0039, -0.0409, ..., -0.0248, -0.0036, -0.0153], |
| ..., |
| [ 0.0249, -0.0349, 0.0021, ..., 0.0253, 0.0072, 0.0312], |
| [-0.0060, -0.0263, -0.0290, ..., -0.0022, -0.0128, 0.0040], |
| [-0.0438, -0.0769, 0.0416, ..., -0.0423, -0.0112, 0.0007]], |
| device='cuda:0'), grad: tensor([[ 1.5512e-05, 1.1791e-06, 9.5218e-06, ..., 0.0000e+00, |
| 1.8209e-05, 6.9477e-07], |
| [ 2.7180e-05, 8.4750e-08, 3.7942e-06, ..., 0.0000e+00, |
| -6.8605e-05, 2.7940e-07], |
| [-1.0514e-04, -1.0476e-05, 1.2526e-06, ..., 0.0000e+00, |
| 2.2516e-05, -1.9297e-06], |
| ..., |
| [-8.1122e-05, 8.3633e-07, 4.9248e-06, ..., 0.0000e+00, |
| 1.2346e-05, 4.5076e-07], |
| [ 1.5162e-06, 5.1688e-07, 5.5701e-05, ..., 0.0000e+00, |
| 8.7023e-05, 4.5002e-06], |
| [ 4.0025e-05, 1.7881e-07, 4.8578e-06, ..., 0.0000e+00, |
| 6.1989e-06, 8.6799e-07]], device='cuda:0') |
| Epoch 53, bias, value: tensor([-0.0037, 0.0273, 0.0067, 0.0031, 0.0058, -0.0056, 0.0206, -0.0238, |
| 0.0227, 0.0034], device='cuda:0'), grad: tensor([ 5.4300e-05, -5.0664e-05, -1.0252e-04, 1.6451e-04, 4.5300e-05, |
| -7.2670e-03, 6.9962e-03, -1.2708e-04, 1.9574e-04, 8.8096e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 52, time 217.56, cls_loss 0.0145 cls_loss_mapping 0.0191 cls_loss_causal 0.7120 re_mapping 0.0158 re_causal 0.0456 /// teacc 98.68 lr 0.00010000 |
| Epoch 54, weight, value: tensor([[-0.0241, -0.0153, -0.0176, ..., -0.0685, 0.0203, 0.0144], |
| [-0.0138, 0.0154, -0.0725, ..., 0.0190, 0.0211, -0.0371], |
| [ 0.0397, -0.0038, -0.0413, ..., -0.0249, -0.0041, -0.0159], |
| ..., |
| [ 0.0248, -0.0359, 0.0021, ..., 0.0253, 0.0066, 0.0319], |
| [-0.0059, -0.0269, -0.0288, ..., -0.0023, -0.0124, 0.0032], |
| [-0.0443, -0.0789, 0.0415, ..., -0.0426, -0.0106, 0.0006]], |
| device='cuda:0'), grad: tensor([[-8.9034e-07, 9.3132e-10, 3.8929e-06, ..., 8.7265e-07, |
| 1.2085e-05, 6.5193e-09], |
| [ 4.7952e-05, -6.5193e-08, 2.8968e-05, ..., 1.3120e-05, |
| -9.6142e-05, 3.0734e-08], |
| [-5.5321e-06, 8.3819e-09, 3.2540e-06, ..., 5.8860e-07, |
| 1.3672e-05, 3.3528e-08], |
| ..., |
| [-8.9288e-05, 4.6566e-09, -2.4661e-05, ..., -3.7491e-05, |
| 3.8385e-05, 7.0781e-08], |
| [ 3.9227e-06, 3.6322e-08, -9.1419e-06, ..., 7.3947e-07, |
| 1.2167e-05, 5.5879e-09], |
| [-1.1146e-04, 9.3132e-10, -1.2898e-04, ..., 1.4231e-05, |
| -9.1851e-05, 6.1467e-08]], device='cuda:0') |
| Epoch 54, bias, value: tensor([-0.0036, 0.0274, 0.0066, 0.0028, 0.0057, -0.0053, 0.0204, -0.0241, |
| 0.0234, 0.0033], device='cuda:0'), grad: tensor([-2.7001e-05, -1.7390e-05, 1.7717e-05, 8.5473e-05, 3.6478e-04, |
| 5.4330e-05, 8.6904e-05, -7.1585e-05, -1.2189e-05, -4.8113e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 53, time 217.60, cls_loss 0.0164 cls_loss_mapping 0.0233 cls_loss_causal 0.6930 re_mapping 0.0159 re_causal 0.0448 /// teacc 98.62 lr 0.00010000 |
| Epoch 55, weight, value: tensor([[-0.0245, -0.0155, -0.0180, ..., -0.0689, 0.0198, 0.0144], |
| [-0.0141, 0.0152, -0.0729, ..., 0.0195, 0.0218, -0.0375], |
| [ 0.0400, -0.0033, -0.0416, ..., -0.0252, -0.0046, -0.0159], |
| ..., |
| [ 0.0251, -0.0366, 0.0018, ..., 0.0248, 0.0064, 0.0318], |
| [-0.0060, -0.0278, -0.0297, ..., -0.0027, -0.0125, 0.0026], |
| [-0.0445, -0.0798, 0.0418, ..., -0.0417, -0.0100, 0.0015]], |
| device='cuda:0'), grad: tensor([[ 5.4855e-07, 1.8626e-09, 2.0880e-06, ..., 9.3132e-10, |
| 7.0147e-06, 1.8626e-09], |
| [ 1.0625e-05, -3.6322e-08, 4.1723e-05, ..., 2.7008e-08, |
| 1.2495e-05, 2.7940e-09], |
| [-1.1854e-05, 2.3283e-08, 1.8954e-05, ..., 3.7253e-09, |
| 5.8934e-06, -2.9802e-08], |
| ..., |
| [-3.2354e-06, -7.6368e-08, 4.4376e-05, ..., -7.4506e-08, |
| 1.7107e-05, 1.0245e-08], |
| [-4.6849e-05, 1.3039e-08, 7.5586e-06, ..., 1.8626e-09, |
| 2.4036e-05, 6.5193e-09], |
| [ 9.4324e-06, 3.0734e-08, 2.6035e-04, ..., 2.2352e-08, |
| 8.3864e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 55, bias, value: tensor([-0.0039, 0.0272, 0.0065, 0.0021, 0.0061, -0.0050, 0.0206, -0.0239, |
| 0.0228, 0.0038], device='cuda:0'), grad: tensor([-3.8713e-05, 1.0645e-04, 2.7493e-05, 6.2644e-05, -8.4352e-04, |
| 2.0945e-04, -2.6345e-04, 1.0085e-04, -3.8370e-07, 6.3944e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 54, time 217.37, cls_loss 0.0150 cls_loss_mapping 0.0199 cls_loss_causal 0.7068 re_mapping 0.0156 re_causal 0.0453 /// teacc 98.60 lr 0.00010000 |
| Epoch 56, weight, value: tensor([[-0.0247, -0.0156, -0.0177, ..., -0.0693, 0.0197, 0.0145], |
| [-0.0148, 0.0154, -0.0735, ..., 0.0200, 0.0215, -0.0382], |
| [ 0.0410, -0.0032, -0.0419, ..., -0.0252, -0.0052, -0.0155], |
| ..., |
| [ 0.0249, -0.0371, 0.0017, ..., 0.0249, 0.0060, 0.0317], |
| [-0.0065, -0.0281, -0.0298, ..., -0.0029, -0.0117, 0.0022], |
| [-0.0451, -0.0802, 0.0417, ..., -0.0417, -0.0097, 0.0015]], |
| device='cuda:0'), grad: tensor([[ 1.1310e-05, 0.0000e+00, 1.1489e-05, ..., 2.4214e-07, |
| 8.5384e-06, 2.0489e-07], |
| [ 1.0908e-05, 0.0000e+00, 9.8133e-04, ..., 1.5553e-07, |
| -1.4871e-05, 1.3039e-07], |
| [-6.3062e-05, 0.0000e+00, 1.6928e-05, ..., 8.8476e-08, |
| 1.0617e-05, 7.4506e-08], |
| ..., |
| [ 7.4267e-05, 0.0000e+00, 1.1963e-04, ..., -8.7079e-07, |
| 2.7373e-05, -7.2457e-07], |
| [-7.1704e-05, 0.0000e+00, -1.0040e-06, ..., 1.3970e-08, |
| -5.5641e-05, 1.3970e-08], |
| [-2.6345e-04, 0.0000e+00, -1.4257e-04, ..., 1.2480e-07, |
| -1.0115e-04, 8.6613e-08]], device='cuda:0') |
| Epoch 56, bias, value: tensor([-0.0034, 0.0261, 0.0076, 0.0022, 0.0060, -0.0051, 0.0208, -0.0239, |
| 0.0226, 0.0035], device='cuda:0'), grad: tensor([-6.1512e-05, 1.6642e-03, 1.2144e-05, 6.1369e-04, -2.0256e-03, |
| 1.1718e-04, 2.0981e-04, 3.4213e-04, -2.2316e-04, -6.4898e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 55---------------------------------------------------- |
| epoch 55, time 218.09, cls_loss 0.0158 cls_loss_mapping 0.0216 cls_loss_causal 0.6807 re_mapping 0.0153 re_causal 0.0438 /// teacc 98.75 lr 0.00010000 |
| Epoch 57, weight, value: tensor([[-0.0248, -0.0158, -0.0183, ..., -0.0694, 0.0195, 0.0146], |
| [-0.0153, 0.0157, -0.0749, ..., 0.0201, 0.0222, -0.0391], |
| [ 0.0409, -0.0034, -0.0421, ..., -0.0253, -0.0060, -0.0154], |
| ..., |
| [ 0.0251, -0.0371, 0.0017, ..., 0.0247, 0.0056, 0.0319], |
| [-0.0071, -0.0287, -0.0297, ..., -0.0029, -0.0114, 0.0019], |
| [-0.0456, -0.0806, 0.0416, ..., -0.0414, -0.0094, 0.0013]], |
| device='cuda:0'), grad: tensor([[ 4.9710e-05, 0.0000e+00, 4.1053e-06, ..., 0.0000e+00, |
| 8.2478e-06, 2.0489e-08], |
| [ 6.7830e-05, 0.0000e+00, 8.4657e-07, ..., 0.0000e+00, |
| -6.4932e-06, 3.7253e-09], |
| [ 1.3816e-04, 0.0000e+00, 2.0117e-06, ..., 0.0000e+00, |
| 6.0797e-05, -1.6019e-07], |
| ..., |
| [ 1.2720e-04, 0.0000e+00, 1.6131e-06, ..., 0.0000e+00, |
| 2.3380e-05, 2.4214e-08], |
| [ 3.1528e-03, 0.0000e+00, 1.2845e-05, ..., 0.0000e+00, |
| 5.5122e-04, 4.5635e-08], |
| [ 4.6790e-05, 0.0000e+00, -7.7952e-07, ..., 0.0000e+00, |
| -5.1558e-06, 1.8626e-09]], device='cuda:0') |
| Epoch 57, bias, value: tensor([-0.0044, 0.0262, 0.0070, 0.0028, 0.0058, -0.0049, 0.0213, -0.0238, |
| 0.0228, 0.0034], device='cuda:0'), grad: tensor([-4.8339e-05, 9.0957e-05, 3.6979e-04, -6.9618e-03, 2.9474e-05, |
| 6.2561e-04, -7.4267e-05, 2.2662e-04, 5.6686e-03, 7.0810e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 56, time 217.63, cls_loss 0.0186 cls_loss_mapping 0.0258 cls_loss_causal 0.7043 re_mapping 0.0158 re_causal 0.0433 /// teacc 98.57 lr 0.00010000 |
| Epoch 58, weight, value: tensor([[-0.0250, -0.0158, -0.0180, ..., -0.0695, 0.0192, 0.0145], |
| [-0.0156, 0.0161, -0.0754, ..., 0.0215, 0.0232, -0.0394], |
| [ 0.0417, -0.0034, -0.0439, ..., -0.0253, -0.0064, -0.0157], |
| ..., |
| [ 0.0254, -0.0379, 0.0012, ..., 0.0248, 0.0045, 0.0322], |
| [-0.0078, -0.0289, -0.0301, ..., -0.0030, -0.0117, 0.0004], |
| [-0.0466, -0.0810, 0.0413, ..., -0.0424, -0.0096, 0.0033]], |
| device='cuda:0'), grad: tensor([[ 1.6138e-05, 0.0000e+00, 3.1888e-05, ..., 4.6566e-09, |
| 4.7117e-05, 0.0000e+00], |
| [-9.5248e-05, 0.0000e+00, 3.1412e-05, ..., 4.0978e-08, |
| -2.5892e-04, 0.0000e+00], |
| [ 2.8133e-04, 0.0000e+00, 4.5395e-04, ..., 8.3819e-09, |
| 2.1851e-04, 0.0000e+00], |
| ..., |
| [-4.9233e-05, 0.0000e+00, 5.6535e-05, ..., -1.4808e-07, |
| 1.3351e-04, 0.0000e+00], |
| [-4.2677e-04, 0.0000e+00, -5.2357e-04, ..., 3.7253e-09, |
| -8.2636e-04, 0.0000e+00], |
| [ 1.4293e-04, 0.0000e+00, 2.3687e-04, ..., 4.3772e-08, |
| 3.1137e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 58, bias, value: tensor([-0.0039, 0.0267, 0.0069, 0.0029, 0.0070, -0.0045, 0.0206, -0.0240, |
| 0.0224, 0.0025], device='cuda:0'), grad: tensor([ 1.6534e-04, -4.2868e-04, 1.7815e-03, 7.1716e-04, -1.0710e-03, |
| 5.3740e-04, 1.2405e-05, 2.7108e-04, -3.0842e-03, 1.1015e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 57---------------------------------------------------- |
| epoch 57, time 218.30, cls_loss 0.0125 cls_loss_mapping 0.0167 cls_loss_causal 0.6787 re_mapping 0.0154 re_causal 0.0436 /// teacc 98.78 lr 0.00010000 |
| Epoch 59, weight, value: tensor([[-0.0252, -0.0158, -0.0180, ..., -0.0697, 0.0189, 0.0145], |
| [-0.0163, 0.0161, -0.0760, ..., 0.0215, 0.0231, -0.0400], |
| [ 0.0415, -0.0034, -0.0442, ..., -0.0254, -0.0066, -0.0156], |
| ..., |
| [ 0.0263, -0.0379, 0.0009, ..., 0.0247, 0.0043, 0.0323], |
| [-0.0084, -0.0290, -0.0303, ..., -0.0030, -0.0117, -0.0010], |
| [-0.0472, -0.0810, 0.0410, ..., -0.0421, -0.0095, 0.0040]], |
| device='cuda:0'), grad: tensor([[ 2.0528e-04, 0.0000e+00, 8.5640e-04, ..., 1.2200e-07, |
| 8.0228e-05, 2.2352e-08], |
| [ 3.8892e-05, 0.0000e+00, 4.8578e-06, ..., 7.1153e-07, |
| -3.3855e-05, 2.9802e-08], |
| [-1.9681e-04, 0.0000e+00, 1.1034e-05, ..., 1.6112e-07, |
| 3.5137e-05, -4.0792e-07], |
| ..., |
| [ 2.4706e-05, 0.0000e+00, 3.2902e-05, ..., -3.7421e-06, |
| 1.9372e-05, 4.4703e-08], |
| [ 6.1870e-05, 0.0000e+00, 1.7241e-05, ..., 3.6322e-08, |
| 3.2216e-05, 1.0710e-07], |
| [ 1.3340e-04, 0.0000e+00, -2.8968e-05, ..., 1.3616e-06, |
| 3.5495e-05, 5.6811e-08]], device='cuda:0') |
| Epoch 59, bias, value: tensor([-0.0037, 0.0260, 0.0064, 0.0036, 0.0073, -0.0051, 0.0210, -0.0234, |
| 0.0222, 0.0023], device='cuda:0'), grad: tensor([ 6.5384e-03, 3.2693e-05, -2.4128e-04, -1.9169e-03, 6.7711e-05, |
| -5.8861e-03, 7.1907e-04, 1.6999e-04, 1.9586e-04, 3.2806e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 58, time 217.54, cls_loss 0.0172 cls_loss_mapping 0.0211 cls_loss_causal 0.6828 re_mapping 0.0150 re_causal 0.0416 /// teacc 98.75 lr 0.00010000 |
| Epoch 60, weight, value: tensor([[-0.0257, -0.0160, -0.0187, ..., -0.0699, 0.0184, 0.0121], |
| [-0.0173, 0.0160, -0.0761, ..., 0.0232, 0.0238, -0.0407], |
| [ 0.0416, -0.0029, -0.0445, ..., -0.0256, -0.0071, -0.0150], |
| ..., |
| [ 0.0271, -0.0385, 0.0006, ..., 0.0239, 0.0037, 0.0321], |
| [-0.0088, -0.0295, -0.0307, ..., -0.0032, -0.0118, -0.0034], |
| [-0.0476, -0.0819, 0.0410, ..., -0.0422, -0.0087, 0.0062]], |
| device='cuda:0'), grad: tensor([[ 1.1347e-05, 0.0000e+00, 2.9624e-05, ..., 5.4296e-07, |
| 1.6227e-05, 2.1514e-07], |
| [ 2.5064e-05, 0.0000e+00, 3.1590e-05, ..., 2.9374e-06, |
| -8.3625e-05, 4.2841e-08], |
| [ 2.3520e-04, 0.0000e+00, 1.5926e-04, ..., 1.3541e-06, |
| 6.3181e-05, -2.0284e-06], |
| ..., |
| [-7.4530e-04, 0.0000e+00, -2.9373e-04, ..., -1.1764e-05, |
| -1.0717e-04, 5.2806e-07], |
| [ 1.2094e-04, 0.0000e+00, 5.5695e-04, ..., 2.0768e-07, |
| 2.5153e-04, 4.9639e-07], |
| [ 3.8266e-04, 0.0000e+00, 3.5048e-04, ..., 1.9874e-06, |
| 1.6320e-04, 5.0291e-08]], device='cuda:0') |
| Epoch 60, bias, value: tensor([-0.0046, 0.0262, 0.0061, 0.0037, 0.0078, -0.0049, 0.0213, -0.0230, |
| 0.0215, 0.0025], device='cuda:0'), grad: tensor([-7.2300e-05, -8.2433e-05, 6.9761e-04, 3.4237e-03, 1.0049e-04, |
| -6.0387e-03, 7.3135e-05, -1.3237e-03, 1.8244e-03, 1.3971e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 59, time 217.55, cls_loss 0.0166 cls_loss_mapping 0.0204 cls_loss_causal 0.6826 re_mapping 0.0141 re_causal 0.0387 /// teacc 98.65 lr 0.00010000 |
| Epoch 61, weight, value: tensor([[-0.0254, -0.0160, -0.0193, ..., -0.0705, 0.0180, 0.0111], |
| [-0.0169, 0.0160, -0.0763, ..., 0.0229, 0.0251, -0.0412], |
| [ 0.0423, -0.0028, -0.0438, ..., -0.0259, -0.0079, -0.0139], |
| ..., |
| [ 0.0263, -0.0387, 0.0005, ..., 0.0245, 0.0034, 0.0320], |
| [-0.0092, -0.0297, -0.0309, ..., -0.0033, -0.0117, -0.0040], |
| [-0.0483, -0.0820, 0.0406, ..., -0.0423, -0.0077, 0.0070]], |
| device='cuda:0'), grad: tensor([[-3.3736e-05, 0.0000e+00, 2.2724e-06, ..., 1.7416e-07, |
| 3.0883e-06, 0.0000e+00], |
| [ 1.7837e-05, 0.0000e+00, 2.3052e-05, ..., 3.1590e-06, |
| 2.2829e-05, 0.0000e+00], |
| [-1.8287e-04, 0.0000e+00, 2.1011e-06, ..., 1.8347e-07, |
| 6.0201e-06, 0.0000e+00], |
| ..., |
| [ 1.4937e-04, 0.0000e+00, 9.5516e-06, ..., 1.8999e-07, |
| 4.2841e-06, 0.0000e+00], |
| [ 1.1183e-05, 0.0000e+00, 1.4611e-05, ..., 2.2724e-07, |
| 8.9109e-06, 0.0000e+00], |
| [ 6.5155e-06, 0.0000e+00, 2.1946e-04, ..., 3.0637e-05, |
| 2.4462e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 61, bias, value: tensor([-0.0037, 0.0270, 0.0065, 0.0033, 0.0075, -0.0048, 0.0207, -0.0236, |
| 0.0217, 0.0021], device='cuda:0'), grad: tensor([-1.5318e-04, 1.0294e-04, -2.0874e-04, 4.4703e-05, -2.3991e-05, |
| -3.2872e-05, -7.9489e-04, 2.3258e-04, 7.4685e-05, 7.5865e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 60, time 217.42, cls_loss 0.0137 cls_loss_mapping 0.0218 cls_loss_causal 0.6739 re_mapping 0.0142 re_causal 0.0422 /// teacc 98.61 lr 0.00010000 |
| Epoch 62, weight, value: tensor([[-0.0256, -0.0160, -0.0195, ..., -0.0706, 0.0177, 0.0111], |
| [-0.0171, 0.0160, -0.0770, ..., 0.0231, 0.0256, -0.0414], |
| [ 0.0417, -0.0027, -0.0439, ..., -0.0260, -0.0082, -0.0138], |
| ..., |
| [ 0.0272, -0.0389, 0.0004, ..., 0.0248, 0.0031, 0.0320], |
| [-0.0095, -0.0297, -0.0307, ..., -0.0034, -0.0116, -0.0042], |
| [-0.0486, -0.0820, 0.0405, ..., -0.0426, -0.0075, 0.0070]], |
| device='cuda:0'), grad: tensor([[ 2.0191e-05, 0.0000e+00, 1.8001e-05, ..., 1.0617e-05, |
| 9.7305e-06, 0.0000e+00], |
| [ 2.0862e-04, 0.0000e+00, 2.8238e-05, ..., -3.7265e-04, |
| -2.3079e-04, 0.0000e+00], |
| [ 2.2903e-05, 0.0000e+00, 6.5118e-06, ..., 1.6198e-05, |
| 4.0196e-06, 0.0000e+00], |
| ..., |
| [-1.2789e-03, 0.0000e+00, 8.1062e-05, ..., -3.9506e-04, |
| 1.4150e-04, 0.0000e+00], |
| [ 1.8105e-05, 0.0000e+00, 9.6202e-05, ..., 1.5512e-05, |
| 1.6421e-05, 0.0000e+00], |
| [ 7.0238e-04, 0.0000e+00, -1.8911e-03, ..., 4.2415e-04, |
| -2.1935e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 62, bias, value: tensor([-0.0035, 0.0276, 0.0059, 0.0031, 0.0076, -0.0045, 0.0199, -0.0230, |
| 0.0217, 0.0017], device='cuda:0'), grad: tensor([ 8.7202e-05, -2.6779e-03, 1.0395e-04, 2.7990e-04, 4.8065e-03, |
| -1.0052e-03, 8.5497e-04, -1.5268e-03, 2.6345e-04, -1.1806e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 61, time 217.58, cls_loss 0.0129 cls_loss_mapping 0.0180 cls_loss_causal 0.6543 re_mapping 0.0144 re_causal 0.0404 /// teacc 98.73 lr 0.00010000 |
| Epoch 63, weight, value: tensor([[-0.0264, -0.0161, -0.0200, ..., -0.0708, 0.0174, 0.0110], |
| [-0.0186, 0.0160, -0.0776, ..., 0.0244, 0.0262, -0.0416], |
| [ 0.0426, -0.0027, -0.0440, ..., -0.0261, -0.0085, -0.0137], |
| ..., |
| [ 0.0280, -0.0389, 0.0004, ..., 0.0248, 0.0028, 0.0320], |
| [-0.0099, -0.0298, -0.0309, ..., -0.0035, -0.0125, -0.0043], |
| [-0.0492, -0.0821, 0.0408, ..., -0.0434, -0.0069, 0.0070]], |
| device='cuda:0'), grad: tensor([[ 1.1988e-05, 0.0000e+00, 6.2250e-06, ..., 4.4219e-06, |
| 1.3508e-05, 9.5461e-07], |
| [ 2.5082e-04, 0.0000e+00, 1.0999e-06, ..., 1.0228e-04, |
| -3.6192e-04, 3.7812e-07], |
| [ 1.3149e-04, 0.0000e+00, 2.9393e-06, ..., 2.8592e-06, |
| 5.7220e-05, 2.0675e-07], |
| ..., |
| [-2.6727e-04, 0.0000e+00, 9.2462e-06, ..., -4.9973e-04, |
| -3.0804e-04, 6.6962e-07], |
| [ 8.1360e-06, 0.0000e+00, 1.0498e-05, ..., 5.7146e-06, |
| 2.4289e-05, 1.7351e-06], |
| [ 1.4722e-04, 0.0000e+00, 1.9461e-05, ..., 1.4126e-04, |
| 1.3447e-04, 3.5390e-08]], device='cuda:0') |
| Epoch 63, bias, value: tensor([-0.0045, 0.0273, 0.0065, 0.0025, 0.0070, -0.0043, 0.0207, -0.0226, |
| 0.0212, 0.0024], device='cuda:0'), grad: tensor([ 5.1260e-05, -3.9196e-04, 2.0123e-04, -3.1924e-04, 1.1377e-03, |
| -4.2605e-04, 3.1304e-04, -1.0786e-03, 1.2651e-05, 5.0116e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 62, time 217.42, cls_loss 0.0098 cls_loss_mapping 0.0155 cls_loss_causal 0.6317 re_mapping 0.0139 re_causal 0.0408 /// teacc 98.67 lr 0.00010000 |
| Epoch 64, weight, value: tensor([[-0.0267, -0.0161, -0.0201, ..., -0.0709, 0.0170, 0.0106], |
| [-0.0182, 0.0160, -0.0781, ..., 0.0245, 0.0267, -0.0451], |
| [ 0.0432, -0.0027, -0.0442, ..., -0.0263, -0.0086, -0.0112], |
| ..., |
| [ 0.0278, -0.0389, 0.0005, ..., 0.0252, 0.0024, 0.0331], |
| [-0.0105, -0.0298, -0.0310, ..., -0.0036, -0.0123, -0.0077], |
| [-0.0498, -0.0821, 0.0407, ..., -0.0436, -0.0066, 0.0070]], |
| device='cuda:0'), grad: tensor([[ 6.0536e-07, 0.0000e+00, -3.2447e-06, ..., 2.3562e-07, |
| 2.0694e-06, 0.0000e+00], |
| [ 2.2396e-05, 0.0000e+00, 1.0796e-05, ..., 4.9584e-06, |
| 9.0897e-07, 0.0000e+00], |
| [-2.8551e-05, 0.0000e+00, 3.1125e-06, ..., 6.1560e-07, |
| 3.4608e-06, 0.0000e+00], |
| ..., |
| [-3.1620e-05, 0.0000e+00, 8.2999e-06, ..., -9.4622e-06, |
| 3.0138e-06, 0.0000e+00], |
| [ 5.6699e-06, 0.0000e+00, 6.8732e-06, ..., 1.3225e-07, |
| -5.4669e-07, 0.0000e+00], |
| [ 1.3776e-05, 0.0000e+00, 6.4913e-07, ..., 2.0433e-06, |
| -9.7081e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 64, bias, value: tensor([-0.0043, 0.0275, 0.0069, 0.0023, 0.0069, -0.0045, 0.0206, -0.0225, |
| 0.0213, 0.0020], device='cuda:0'), grad: tensor([-6.9439e-05, 5.7817e-05, -2.4870e-05, 1.9029e-05, -3.3706e-05, |
| 2.5228e-05, 1.1884e-05, -2.7746e-05, 1.3225e-05, 2.8580e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 63, time 217.33, cls_loss 0.0105 cls_loss_mapping 0.0174 cls_loss_causal 0.6285 re_mapping 0.0139 re_causal 0.0398 /// teacc 98.61 lr 0.00010000 |
| Epoch 65, weight, value: tensor([[-2.7072e-02, -1.6177e-02, -2.0683e-02, ..., -7.1142e-02, |
| 1.6065e-02, 1.0269e-02], |
| [-1.8219e-02, 1.6318e-02, -7.8408e-02, ..., 2.4330e-02, |
| 2.6914e-02, -4.5978e-02], |
| [ 4.3092e-02, -3.2435e-03, -4.4140e-02, ..., -2.6427e-02, |
| -9.2436e-03, -1.0578e-02], |
| ..., |
| [ 2.8469e-02, -3.7164e-02, 9.7241e-05, ..., 2.5689e-02, |
| 2.2677e-03, 3.3124e-02], |
| [-1.1044e-02, -2.8373e-02, -3.1328e-02, ..., -3.7202e-03, |
| -1.2169e-02, -8.0873e-03], |
| [-5.0665e-02, -8.2488e-02, 4.0945e-02, ..., -4.3980e-02, |
| -5.8393e-03, 7.2448e-03]], device='cuda:0'), grad: tensor([[-4.5806e-05, 0.0000e+00, -6.1654e-06, ..., 0.0000e+00, |
| -8.8692e-05, 2.5891e-07], |
| [-2.2078e-04, 0.0000e+00, 3.5428e-06, ..., -7.4506e-09, |
| -1.3590e-04, 8.4750e-08], |
| [ 1.2517e-04, 0.0000e+00, 9.6038e-06, ..., 9.3132e-10, |
| 8.9407e-05, -2.4959e-06], |
| ..., |
| [-1.9395e-04, 0.0000e+00, 1.1474e-05, ..., 2.7940e-09, |
| 2.5004e-05, 2.8219e-07], |
| [ 3.0190e-05, 0.0000e+00, -9.6262e-05, ..., 9.3132e-10, |
| 2.9400e-05, 9.4995e-07], |
| [ 9.4056e-05, 0.0000e+00, 3.0756e-05, ..., 9.3132e-10, |
| 1.1668e-05, 2.9430e-07]], device='cuda:0') |
| Epoch 65, bias, value: tensor([-0.0056, 0.0274, 0.0066, 0.0022, 0.0069, -0.0040, 0.0208, -0.0223, |
| 0.0213, 0.0025], device='cuda:0'), grad: tensor([-6.9189e-04, -4.8566e-04, 6.0892e-04, 4.1437e-04, -1.0639e-05, |
| 3.4142e-04, 4.0531e-05, -2.5105e-04, -3.0065e-04, 3.3569e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 64, time 217.47, cls_loss 0.0115 cls_loss_mapping 0.0163 cls_loss_causal 0.6585 re_mapping 0.0138 re_causal 0.0398 /// teacc 98.60 lr 0.00010000 |
| Epoch 66, weight, value: tensor([[-0.0272, -0.0163, -0.0210, ..., -0.0715, 0.0157, 0.0102], |
| [-0.0181, 0.0168, -0.0790, ..., 0.0249, 0.0274, -0.0474], |
| [ 0.0430, -0.0037, -0.0443, ..., -0.0267, -0.0095, -0.0101], |
| ..., |
| [ 0.0285, -0.0373, -0.0003, ..., 0.0255, 0.0015, 0.0334], |
| [-0.0114, -0.0279, -0.0316, ..., -0.0040, -0.0121, -0.0090], |
| [-0.0507, -0.0830, 0.0411, ..., -0.0438, -0.0049, 0.0073]], |
| device='cuda:0'), grad: tensor([[ 3.2187e-05, 0.0000e+00, 7.4133e-07, ..., 4.4703e-08, |
| 6.4336e-06, 3.7253e-09], |
| [ 1.0520e-04, 0.0000e+00, -1.7323e-07, ..., 3.7067e-07, |
| -4.6074e-05, 1.8626e-09], |
| [-1.4615e-04, 0.0000e+00, 3.5223e-06, ..., 6.5193e-08, |
| 1.0267e-05, 0.0000e+00], |
| ..., |
| [-1.4150e-04, 0.0000e+00, 3.3677e-06, ..., 3.0547e-07, |
| -3.5204e-07, 1.8626e-09], |
| [ 1.1258e-05, 0.0000e+00, -1.8906e-06, ..., 8.1956e-08, |
| 2.2426e-05, 1.6764e-08], |
| [ 4.5061e-05, 0.0000e+00, 9.7603e-06, ..., 2.1700e-06, |
| 2.8387e-06, 7.4506e-09]], device='cuda:0') |
| Epoch 66, bias, value: tensor([-0.0053, 0.0279, 0.0062, 0.0022, 0.0073, -0.0040, 0.0208, -0.0230, |
| 0.0211, 0.0029], device='cuda:0'), grad: tensor([ 4.7386e-05, 1.1557e-04, -1.2052e-04, 1.2934e-04, 3.9935e-06, |
| -3.7272e-06, 4.7654e-05, -3.1686e-04, 5.4836e-05, 4.2111e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 65, time 217.73, cls_loss 0.0131 cls_loss_mapping 0.0205 cls_loss_causal 0.6611 re_mapping 0.0136 re_causal 0.0398 /// teacc 98.61 lr 0.00010000 |
| Epoch 67, weight, value: tensor([[-0.0276, -0.0163, -0.0225, ..., -0.0717, 0.0154, 0.0086], |
| [-0.0182, 0.0173, -0.0797, ..., 0.0261, 0.0281, -0.0498], |
| [ 0.0430, -0.0041, -0.0445, ..., -0.0268, -0.0098, -0.0102], |
| ..., |
| [ 0.0291, -0.0373, -0.0002, ..., 0.0245, 0.0011, 0.0325], |
| [-0.0117, -0.0280, -0.0316, ..., -0.0040, -0.0123, -0.0114], |
| [-0.0516, -0.0832, 0.0411, ..., -0.0444, -0.0043, 0.0091]], |
| device='cuda:0'), grad: tensor([[ 2.9281e-06, 0.0000e+00, -4.0010e-06, ..., 2.9802e-08, |
| 9.4622e-07, 0.0000e+00], |
| [ 5.2080e-06, 0.0000e+00, 8.5607e-06, ..., 2.2911e-07, |
| -5.0589e-06, 0.0000e+00], |
| [ 1.6773e-04, 0.0000e+00, 1.8571e-06, ..., 1.4901e-08, |
| 4.1246e-05, 0.0000e+00], |
| ..., |
| [ 4.6015e-04, 0.0000e+00, 4.7415e-05, ..., -8.4005e-07, |
| 1.1665e-04, 0.0000e+00], |
| [ 2.2560e-05, 0.0000e+00, 4.4592e-06, ..., 9.3132e-09, |
| 1.9707e-06, 0.0000e+00], |
| [-4.7588e-04, 0.0000e+00, 9.3818e-05, ..., 4.3772e-07, |
| -1.2022e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 67, bias, value: tensor([-0.0062, 0.0282, 0.0058, 0.0026, 0.0072, -0.0043, 0.0209, -0.0223, |
| 0.0210, 0.0028], device='cuda:0'), grad: tensor([-4.2653e-04, 3.5793e-05, 2.2519e-04, -1.6630e-04, -3.9911e-04, |
| -6.3926e-06, 4.6432e-05, 1.2360e-03, 3.9428e-05, -5.8317e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 66, time 217.57, cls_loss 0.0108 cls_loss_mapping 0.0163 cls_loss_causal 0.6815 re_mapping 0.0131 re_causal 0.0391 /// teacc 98.44 lr 0.00010000 |
| Epoch 68, weight, value: tensor([[-0.0278, -0.0164, -0.0225, ..., -0.0718, 0.0154, 0.0085], |
| [-0.0188, 0.0172, -0.0803, ..., 0.0262, 0.0280, -0.0518], |
| [ 0.0430, -0.0038, -0.0447, ..., -0.0269, -0.0103, -0.0097], |
| ..., |
| [ 0.0290, -0.0376, -0.0004, ..., 0.0245, 0.0009, 0.0319], |
| [-0.0116, -0.0282, -0.0319, ..., -0.0040, -0.0119, -0.0129], |
| [-0.0521, -0.0835, 0.0407, ..., -0.0446, -0.0044, 0.0092]], |
| device='cuda:0'), grad: tensor([[-4.2558e-05, 0.0000e+00, -1.0401e-04, ..., 1.7323e-06, |
| 1.0937e-05, 0.0000e+00], |
| [ 2.0385e-05, 0.0000e+00, 2.5779e-06, ..., 7.1898e-07, |
| 7.6070e-06, 0.0000e+00], |
| [-4.4405e-06, 0.0000e+00, 2.7329e-05, ..., 2.0117e-07, |
| 1.4424e-05, 0.0000e+00], |
| ..., |
| [ 1.9610e-05, 0.0000e+00, 6.6310e-06, ..., 4.5672e-06, |
| 4.9099e-06, 0.0000e+00], |
| [ 4.4078e-05, 0.0000e+00, 2.2978e-05, ..., 5.1223e-07, |
| 3.1263e-05, 0.0000e+00], |
| [ 1.4924e-05, 0.0000e+00, 8.5980e-06, ..., 1.3504e-06, |
| -5.0217e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 68, bias, value: tensor([-0.0052, 0.0277, 0.0054, 0.0029, 0.0079, -0.0038, 0.0199, -0.0224, |
| 0.0219, 0.0017], device='cuda:0'), grad: tensor([-2.2864e-04, 4.4703e-05, 6.5088e-05, -1.7500e-04, 6.8188e-04, |
| 8.3596e-06, -5.8270e-04, 7.8678e-05, 1.7023e-04, -6.3062e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 67, time 217.62, cls_loss 0.0130 cls_loss_mapping 0.0167 cls_loss_causal 0.6240 re_mapping 0.0135 re_causal 0.0360 /// teacc 98.68 lr 0.00010000 |
| Epoch 69, weight, value: tensor([[-0.0273, -0.0171, -0.0232, ..., -0.0719, 0.0148, 0.0083], |
| [-0.0195, 0.0175, -0.0807, ..., 0.0266, 0.0282, -0.0539], |
| [ 0.0436, -0.0041, -0.0447, ..., -0.0273, -0.0102, -0.0084], |
| ..., |
| [ 0.0291, -0.0375, -0.0004, ..., 0.0244, 0.0005, 0.0320], |
| [-0.0122, -0.0285, -0.0323, ..., -0.0042, -0.0117, -0.0147], |
| [-0.0524, -0.0865, 0.0411, ..., -0.0446, -0.0034, 0.0093]], |
| device='cuda:0'), grad: tensor([[ 8.6054e-07, 0.0000e+00, 5.7556e-07, ..., 0.0000e+00, |
| 4.5784e-06, 1.4901e-08], |
| [ 8.1062e-06, 0.0000e+00, 7.7114e-07, ..., 0.0000e+00, |
| 2.4028e-06, 3.7253e-09], |
| [-4.3273e-05, 0.0000e+00, 6.4075e-07, ..., 0.0000e+00, |
| -8.1807e-06, -1.8999e-07], |
| ..., |
| [-7.8604e-06, 0.0000e+00, 1.1940e-06, ..., 0.0000e+00, |
| 1.1660e-06, 2.6077e-08], |
| [ 7.7963e-05, 0.0000e+00, -3.0361e-07, ..., 0.0000e+00, |
| -4.7296e-05, 6.7055e-08], |
| [ 7.4506e-06, 0.0000e+00, 2.8443e-06, ..., 0.0000e+00, |
| -2.8815e-06, 1.1176e-08]], device='cuda:0') |
| Epoch 69, bias, value: tensor([-0.0056, 0.0271, 0.0061, 0.0027, 0.0071, -0.0043, 0.0207, -0.0226, |
| 0.0216, 0.0029], device='cuda:0'), grad: tensor([ 4.1723e-06, 3.2455e-05, -3.0112e-04, -6.1929e-05, 3.6150e-05, |
| 2.8145e-06, 5.7995e-05, 2.3823e-06, 2.0814e-04, 1.8746e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 68, time 217.61, cls_loss 0.0105 cls_loss_mapping 0.0142 cls_loss_causal 0.6285 re_mapping 0.0132 re_causal 0.0378 /// teacc 98.70 lr 0.00010000 |
| Epoch 70, weight, value: tensor([[-0.0276, -0.0177, -0.0233, ..., -0.0720, 0.0147, 0.0083], |
| [-0.0199, 0.0174, -0.0809, ..., 0.0267, 0.0285, -0.0540], |
| [ 0.0452, -0.0029, -0.0452, ..., -0.0276, -0.0104, -0.0083], |
| ..., |
| [ 0.0280, -0.0391, -0.0008, ..., 0.0243, 0.0003, 0.0320], |
| [-0.0127, -0.0294, -0.0326, ..., -0.0043, -0.0117, -0.0150], |
| [-0.0531, -0.0898, 0.0410, ..., -0.0447, -0.0032, 0.0093]], |
| device='cuda:0'), grad: tensor([[-2.1040e-05, 1.4901e-08, 2.5287e-05, ..., 5.8301e-07, |
| 1.3039e-06, 0.0000e+00], |
| [-3.9265e-06, 1.8626e-09, 3.8929e-06, ..., 2.6822e-07, |
| -2.9355e-05, 0.0000e+00], |
| [ 2.5421e-05, -1.9185e-07, 1.0908e-05, ..., 2.4028e-07, |
| 6.5304e-06, 0.0000e+00], |
| ..., |
| [-3.5763e-05, 5.7742e-08, 3.8091e-06, ..., 4.4703e-07, |
| 1.3202e-05, 0.0000e+00], |
| [ 4.1649e-06, 3.5390e-08, 2.1327e-06, ..., 6.3889e-07, |
| 1.5311e-06, 0.0000e+00], |
| [ 9.0748e-06, 1.8626e-09, 2.4605e-04, ..., 9.0152e-07, |
| -4.3884e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 70, bias, value: tensor([-0.0059, 0.0271, 0.0073, 0.0026, 0.0070, -0.0035, 0.0207, -0.0235, |
| 0.0211, 0.0028], device='cuda:0'), grad: tensor([-5.0545e-05, -5.3406e-05, 7.9989e-05, 2.6679e-04, -6.3086e-04, |
| -1.8489e-04, 1.1188e-04, -2.3812e-05, -5.2899e-06, 4.9019e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 69, time 217.57, cls_loss 0.0124 cls_loss_mapping 0.0166 cls_loss_causal 0.6367 re_mapping 0.0129 re_causal 0.0360 /// teacc 98.55 lr 0.00010000 |
| Epoch 71, weight, value: tensor([[-0.0278, -0.0180, -0.0226, ..., -0.0724, 0.0145, 0.0083], |
| [-0.0206, 0.0190, -0.0813, ..., 0.0269, 0.0288, -0.0541], |
| [ 0.0452, -0.0032, -0.0453, ..., -0.0278, -0.0111, -0.0083], |
| ..., |
| [ 0.0292, -0.0418, -0.0010, ..., 0.0238, 0.0003, 0.0320], |
| [-0.0134, -0.0293, -0.0328, ..., -0.0043, -0.0119, -0.0150], |
| [-0.0540, -0.0909, 0.0407, ..., -0.0438, -0.0030, 0.0093]], |
| device='cuda:0'), grad: tensor([[ 9.9465e-06, 1.1735e-07, 5.9120e-06, ..., 0.0000e+00, |
| 3.9116e-06, 0.0000e+00], |
| [ 5.7316e-04, 3.5204e-07, 1.4305e-05, ..., 0.0000e+00, |
| 1.9228e-04, 0.0000e+00], |
| [ 1.3344e-05, -3.5763e-06, 9.8124e-06, ..., 0.0000e+00, |
| 4.1306e-05, 0.0000e+00], |
| ..., |
| [-6.3467e-04, 2.5537e-06, -1.3620e-05, ..., 0.0000e+00, |
| -2.1434e-04, 0.0000e+00], |
| [ 4.4584e-05, 2.5891e-07, 3.2842e-05, ..., 0.0000e+00, |
| 8.7693e-06, 0.0000e+00], |
| [ 1.7524e-05, 7.6368e-08, 7.8306e-06, ..., 0.0000e+00, |
| -2.5794e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 71, bias, value: tensor([-0.0054, 0.0273, 0.0069, 0.0026, 0.0068, -0.0034, 0.0209, -0.0228, |
| 0.0209, 0.0021], device='cuda:0'), grad: tensor([ 2.7493e-05, 1.0395e-03, -3.6693e-04, 1.1736e-04, 5.2404e-04, |
| -1.9684e-03, 1.3933e-03, -1.0023e-03, 2.4009e-04, -2.1681e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 70, time 217.65, cls_loss 0.0096 cls_loss_mapping 0.0137 cls_loss_causal 0.6347 re_mapping 0.0125 re_causal 0.0354 /// teacc 98.60 lr 0.00010000 |
| Epoch 72, weight, value: tensor([[-2.8076e-02, -1.8209e-02, -2.2606e-02, ..., -7.2785e-02, |
| 1.3884e-02, 8.2790e-03], |
| [-2.1000e-02, 1.8953e-02, -8.1715e-02, ..., 2.7560e-02, |
| 2.9278e-02, -5.4273e-02], |
| [ 4.5328e-02, -2.9603e-03, -4.5887e-02, ..., -2.8830e-02, |
| -1.1658e-02, -7.9245e-03], |
| ..., |
| [ 2.9706e-02, -4.2213e-02, -5.4599e-05, ..., 2.3662e-02, |
| -8.8294e-05, 3.1973e-02], |
| [-1.3631e-02, -2.9744e-02, -3.3184e-02, ..., -4.5730e-03, |
| -1.1871e-02, -1.5075e-02], |
| [-5.4218e-02, -9.1713e-02, 4.1074e-02, ..., -4.3564e-02, |
| -2.3413e-03, 9.3247e-03]], device='cuda:0'), grad: tensor([[-5.3257e-05, 0.0000e+00, 4.0978e-05, ..., 1.2480e-07, |
| 6.3553e-06, 0.0000e+00], |
| [ 5.7034e-06, 0.0000e+00, 9.4920e-06, ..., -1.6868e-05, |
| -2.9728e-05, 0.0000e+00], |
| [ 8.7470e-06, 0.0000e+00, 1.0051e-05, ..., 1.1176e-06, |
| 5.6326e-06, -9.3132e-09], |
| ..., |
| [ 5.7891e-06, 0.0000e+00, 5.3674e-05, ..., 3.6340e-06, |
| 5.5254e-05, 0.0000e+00], |
| [ 2.6226e-05, 0.0000e+00, 6.0536e-06, ..., 2.4773e-07, |
| -4.4554e-06, 1.8626e-09], |
| [ 6.3255e-06, 0.0000e+00, 6.3837e-05, ..., 9.2015e-07, |
| -4.6134e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 72, bias, value: tensor([-0.0056, 0.0270, 0.0067, 0.0025, 0.0060, -0.0033, 0.0209, -0.0222, |
| 0.0209, 0.0024], device='cuda:0'), grad: tensor([-3.5524e-04, 6.8918e-08, 1.2147e-04, 3.8385e-04, -1.1522e-04, |
| -6.2704e-04, -1.8954e-05, 2.8110e-04, 2.0301e-04, 1.2720e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 71, time 217.80, cls_loss 0.0099 cls_loss_mapping 0.0126 cls_loss_causal 0.6549 re_mapping 0.0128 re_causal 0.0372 /// teacc 98.77 lr 0.00010000 |
| Epoch 73, weight, value: tensor([[-0.0276, -0.0184, -0.0229, ..., -0.0733, 0.0139, 0.0083], |
| [-0.0213, 0.0189, -0.0816, ..., 0.0293, 0.0302, -0.0545], |
| [ 0.0446, -0.0027, -0.0462, ..., -0.0308, -0.0128, -0.0083], |
| ..., |
| [ 0.0303, -0.0423, -0.0002, ..., 0.0235, -0.0001, 0.0324], |
| [-0.0142, -0.0300, -0.0337, ..., -0.0053, -0.0121, -0.0152], |
| [-0.0547, -0.0924, 0.0411, ..., -0.0437, -0.0021, 0.0093]], |
| device='cuda:0'), grad: tensor([[ 1.6242e-06, 0.0000e+00, -1.7297e-04, ..., 5.3272e-07, |
| 1.2666e-06, 0.0000e+00], |
| [ 5.1230e-05, 0.0000e+00, 5.2564e-06, ..., 4.6402e-05, |
| -1.1206e-05, 0.0000e+00], |
| [-5.2661e-05, 0.0000e+00, 4.1425e-06, ..., 6.0163e-07, |
| 8.9779e-07, 0.0000e+00], |
| ..., |
| [-5.5909e-05, 0.0000e+00, 1.0967e-05, ..., -5.7161e-05, |
| 9.8050e-06, 0.0000e+00], |
| [ 1.1679e-06, 0.0000e+00, 1.2025e-05, ..., 3.9674e-07, |
| 2.4810e-06, 0.0000e+00], |
| [ 4.8950e-06, 0.0000e+00, 1.2124e-04, ..., 6.9030e-06, |
| -1.0848e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 73, bias, value: tensor([-0.0045, 0.0273, 0.0058, 0.0027, 0.0059, -0.0034, 0.0206, -0.0217, |
| 0.0205, 0.0024], device='cuda:0'), grad: tensor([-7.5912e-04, 5.9128e-04, -6.1035e-05, 9.1970e-05, -3.3379e-04, |
| 3.0175e-05, 6.6280e-04, -6.4182e-04, 5.1618e-05, 3.6740e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 72, time 217.57, cls_loss 0.0129 cls_loss_mapping 0.0171 cls_loss_causal 0.6663 re_mapping 0.0127 re_causal 0.0349 /// teacc 98.62 lr 0.00010000 |
| Epoch 74, weight, value: tensor([[-0.0277, -0.0190, -0.0246, ..., -0.0738, 0.0134, 0.0078], |
| [-0.0223, 0.0185, -0.0821, ..., 0.0293, 0.0309, -0.0553], |
| [ 0.0448, -0.0023, -0.0465, ..., -0.0308, -0.0133, -0.0077], |
| ..., |
| [ 0.0310, -0.0423, -0.0001, ..., 0.0239, -0.0003, 0.0325], |
| [-0.0142, -0.0309, -0.0344, ..., -0.0055, -0.0121, -0.0163], |
| [-0.0555, -0.0943, 0.0415, ..., -0.0436, -0.0013, 0.0097]], |
| device='cuda:0'), grad: tensor([[ 6.0387e-06, 3.1665e-08, 1.0455e-04, ..., 1.9558e-07, |
| 1.3933e-06, 4.6566e-08], |
| [-2.6073e-03, -4.9174e-05, 8.5458e-06, ..., 2.5313e-06, |
| 2.6897e-06, 5.5879e-09], |
| [ 2.2202e-03, 4.1455e-05, 1.2144e-05, ..., 2.7753e-07, |
| 4.5113e-06, -4.3958e-07], |
| ..., |
| [ 2.5654e-04, 5.5470e-06, 1.0699e-05, ..., -1.8671e-05, |
| 2.0545e-06, 2.1048e-07], |
| [ 1.6987e-05, 1.9744e-07, 1.4409e-05, ..., 8.3819e-08, |
| -6.2399e-06, 4.0978e-08], |
| [ 5.2214e-05, 4.0978e-08, 1.0767e-03, ..., 1.3493e-05, |
| -2.0359e-06, 7.4506e-09]], device='cuda:0') |
| Epoch 74, bias, value: tensor([-0.0050, 0.0271, 0.0057, 0.0025, 0.0058, -0.0030, 0.0199, -0.0212, |
| 0.0201, 0.0029], device='cuda:0'), grad: tensor([ 1.8048e-04, -4.6577e-03, 4.0016e-03, 8.1718e-05, 3.7625e-06, |
| -1.9464e-03, 2.7511e-06, 4.8470e-04, 3.1322e-05, 1.8177e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 73, time 217.83, cls_loss 0.0099 cls_loss_mapping 0.0152 cls_loss_causal 0.6255 re_mapping 0.0126 re_causal 0.0362 /// teacc 98.67 lr 0.00010000 |
| Epoch 75, weight, value: tensor([[-0.0281, -0.0192, -0.0255, ..., -0.0741, 0.0130, 0.0078], |
| [-0.0216, 0.0191, -0.0827, ..., 0.0298, 0.0315, -0.0558], |
| [ 0.0450, -0.0026, -0.0463, ..., -0.0315, -0.0137, -0.0075], |
| ..., |
| [ 0.0308, -0.0425, -0.0002, ..., 0.0239, -0.0010, 0.0326], |
| [-0.0145, -0.0312, -0.0355, ..., -0.0056, -0.0123, -0.0165], |
| [-0.0560, -0.0947, 0.0413, ..., -0.0436, -0.0015, 0.0097]], |
| device='cuda:0'), grad: tensor([[ 4.6119e-06, 0.0000e+00, 6.6981e-06, ..., 6.7055e-08, |
| 3.1859e-05, 2.0489e-07], |
| [ 5.7220e-06, 0.0000e+00, 2.3358e-06, ..., 1.9558e-07, |
| 3.3903e-04, 2.4214e-08], |
| [ 1.8284e-05, 0.0000e+00, 1.1176e-08, ..., 2.5518e-07, |
| 2.1055e-05, -6.8732e-07], |
| ..., |
| [-6.1058e-06, 0.0000e+00, -2.6077e-07, ..., 8.4005e-07, |
| 5.3197e-05, 9.8720e-08], |
| [ 1.2532e-05, 0.0000e+00, 1.9684e-05, ..., 1.2852e-07, |
| -3.5381e-04, 1.0990e-07], |
| [ 1.6481e-05, 0.0000e+00, 4.2245e-06, ..., 3.0231e-06, |
| -7.0453e-05, 6.5193e-08]], device='cuda:0') |
| Epoch 75, bias, value: tensor([-0.0054, 0.0278, 0.0057, 0.0021, 0.0069, -0.0031, 0.0206, -0.0215, |
| 0.0197, 0.0025], device='cuda:0'), grad: tensor([ 1.7381e-04, 8.4591e-04, 6.5923e-05, -6.0111e-05, 8.6784e-05, |
| -4.0359e-03, 3.4924e-03, 1.2183e-04, -4.9400e-04, -1.9872e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 74, time 217.65, cls_loss 0.0095 cls_loss_mapping 0.0135 cls_loss_causal 0.6296 re_mapping 0.0118 re_causal 0.0349 /// teacc 98.65 lr 0.00010000 |
| Epoch 76, weight, value: tensor([[-2.8399e-02, -1.9514e-02, -2.5600e-02, ..., -7.4149e-02, |
| 1.2587e-02, 7.4560e-03], |
| [-2.1964e-02, 2.1534e-02, -8.3086e-02, ..., 2.9884e-02, |
| 3.1886e-02, -5.7684e-02], |
| [ 4.5360e-02, -4.0404e-03, -4.5648e-02, ..., -3.1736e-02, |
| -1.4179e-02, -6.7006e-03], |
| ..., |
| [ 3.1116e-02, -4.3402e-02, 9.1840e-05, ..., 2.4230e-02, |
| -9.5036e-04, 3.2503e-02], |
| [-1.5044e-02, -3.1580e-02, -3.5740e-02, ..., -5.6757e-03, |
| -1.3008e-02, -1.7315e-02], |
| [-5.6663e-02, -9.5329e-02, 4.0913e-02, ..., -4.3756e-02, |
| -9.6210e-04, 9.7934e-03]], device='cuda:0'), grad: tensor([[ 7.1004e-06, 0.0000e+00, 8.8848e-07, ..., 3.9116e-07, |
| 1.0788e-05, 1.7881e-07], |
| [ 3.9116e-06, 0.0000e+00, 1.3895e-06, ..., -3.4243e-05, |
| -2.4652e-04, 9.8720e-08], |
| [ 2.9707e-04, 0.0000e+00, 1.1399e-06, ..., 5.3085e-07, |
| 4.3124e-05, -2.5798e-06], |
| ..., |
| [ 2.3693e-05, 0.0000e+00, 2.4587e-06, ..., 3.8706e-06, |
| 3.9846e-05, 4.7684e-07], |
| [-1.6236e-04, 0.0000e+00, 9.8720e-07, ..., 2.9802e-07, |
| -2.1088e-04, 7.6741e-07], |
| [ 8.5160e-06, 0.0000e+00, 4.1574e-06, ..., 4.7944e-06, |
| -3.5197e-05, 7.8231e-08]], device='cuda:0') |
| Epoch 76, bias, value: tensor([-0.0049, 0.0282, 0.0057, 0.0022, 0.0063, -0.0026, 0.0197, -0.0212, |
| 0.0190, 0.0024], device='cuda:0'), grad: tensor([-1.1748e-04, -4.1556e-04, 4.7827e-04, -2.1830e-06, 9.2149e-05, |
| -3.1948e-05, 2.3785e-03, 1.1194e-04, -2.4166e-03, -7.4744e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 75, time 217.53, cls_loss 0.0075 cls_loss_mapping 0.0109 cls_loss_causal 0.6068 re_mapping 0.0120 re_causal 0.0356 /// teacc 98.64 lr 0.00010000 |
| Epoch 77, weight, value: tensor([[-2.8559e-02, -1.9567e-02, -2.5859e-02, ..., -7.4169e-02, |
| 1.2412e-02, 7.1361e-03], |
| [-2.1615e-02, 2.1567e-02, -8.3421e-02, ..., 3.0023e-02, |
| 3.2002e-02, -5.9072e-02], |
| [ 4.5470e-02, -3.6383e-03, -4.6237e-02, ..., -3.1782e-02, |
| -1.4606e-02, -6.3415e-03], |
| ..., |
| [ 3.1047e-02, -4.3923e-02, 1.2460e-05, ..., 2.4196e-02, |
| -1.2116e-03, 3.2991e-02], |
| [-1.5367e-02, -3.1813e-02, -3.5822e-02, ..., -5.6927e-03, |
| -1.2372e-02, -1.8245e-02], |
| [-5.7009e-02, -9.5381e-02, 4.0660e-02, ..., -4.3789e-02, |
| -5.2460e-04, 1.0220e-02]], device='cuda:0'), grad: tensor([[ 2.3916e-06, 0.0000e+00, -1.6969e-06, ..., 0.0000e+00, |
| 1.8533e-06, 1.4156e-07], |
| [ 1.2130e-05, 0.0000e+00, 4.8429e-07, ..., 0.0000e+00, |
| -6.6012e-06, 2.4214e-08], |
| [ 6.9700e-06, 0.0000e+00, 1.1772e-06, ..., 0.0000e+00, |
| 2.9169e-06, 3.1665e-08], |
| ..., |
| [-1.7917e-04, 0.0000e+00, 1.1697e-06, ..., 0.0000e+00, |
| 5.1945e-05, 5.5879e-09], |
| [ 2.0468e-04, 0.0000e+00, 3.3528e-05, ..., 0.0000e+00, |
| 4.0650e-05, 2.4773e-07], |
| [ 2.4766e-05, 0.0000e+00, -1.6280e-06, ..., 0.0000e+00, |
| -6.6042e-05, 2.0489e-08]], device='cuda:0') |
| Epoch 77, bias, value: tensor([-0.0048, 0.0285, 0.0054, 0.0022, 0.0076, -0.0024, 0.0174, -0.0213, |
| 0.0196, 0.0022], device='cuda:0'), grad: tensor([-9.1791e-06, 1.7196e-05, 4.1872e-05, -8.0395e-04, 9.6023e-05, |
| 3.4380e-04, -1.3918e-05, -2.5916e-04, 5.5361e-04, 3.5286e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 76, time 217.60, cls_loss 0.0090 cls_loss_mapping 0.0118 cls_loss_causal 0.6183 re_mapping 0.0120 re_causal 0.0344 /// teacc 98.77 lr 0.00010000 |
| Epoch 78, weight, value: tensor([[-0.0289, -0.0196, -0.0262, ..., -0.0742, 0.0118, 0.0068], |
| [-0.0222, 0.0217, -0.0851, ..., 0.0300, 0.0325, -0.0612], |
| [ 0.0466, -0.0037, -0.0468, ..., -0.0318, -0.0142, -0.0054], |
| ..., |
| [ 0.0313, -0.0438, 0.0001, ..., 0.0243, -0.0014, 0.0335], |
| [-0.0163, -0.0320, -0.0355, ..., -0.0057, -0.0129, -0.0190], |
| [-0.0577, -0.0956, 0.0403, ..., -0.0438, -0.0003, 0.0103]], |
| device='cuda:0'), grad: tensor([[ 2.5295e-06, 7.6368e-08, 1.7136e-06, ..., 7.4506e-09, |
| 4.4405e-06, 2.4773e-07], |
| [-3.1924e-04, -1.0008e-04, 1.5330e-06, ..., -4.2096e-07, |
| -6.0129e-04, 4.8429e-08], |
| [ 3.0065e-04, 9.2387e-05, 1.8813e-06, ..., 4.0978e-08, |
| 5.3644e-04, 2.7195e-07], |
| ..., |
| [ 1.4678e-05, 4.0904e-06, 2.4028e-06, ..., 1.7695e-07, |
| 3.1292e-05, 1.3039e-08], |
| [ 3.8333e-06, 2.7008e-07, -2.9337e-06, ..., 3.7253e-08, |
| 2.9244e-06, 1.6261e-06], |
| [ 1.4510e-06, 7.8231e-08, 1.1213e-05, ..., 2.2352e-08, |
| 6.5528e-06, 3.5390e-08]], device='cuda:0') |
| Epoch 78, bias, value: tensor([-0.0048, 0.0283, 0.0063, 0.0019, 0.0073, -0.0020, 0.0177, -0.0210, |
| 0.0189, 0.0017], device='cuda:0'), grad: tensor([ 4.7758e-06, -2.2602e-03, 2.0752e-03, -1.2435e-05, 7.8321e-05, |
| 7.6950e-05, -4.9859e-05, 1.1832e-04, -9.5487e-05, 6.1035e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 77, time 217.53, cls_loss 0.0098 cls_loss_mapping 0.0129 cls_loss_causal 0.6230 re_mapping 0.0113 re_causal 0.0339 /// teacc 98.74 lr 0.00010000 |
| Epoch 79, weight, value: tensor([[-0.0286, -0.0196, -0.0263, ..., -0.0747, 0.0114, 0.0067], |
| [-0.0218, 0.0221, -0.0858, ..., 0.0331, 0.0339, -0.0616], |
| [ 0.0466, -0.0041, -0.0468, ..., -0.0320, -0.0141, -0.0048], |
| ..., |
| [ 0.0310, -0.0438, -0.0003, ..., 0.0228, -0.0034, 0.0334], |
| [-0.0166, -0.0321, -0.0360, ..., -0.0059, -0.0131, -0.0194], |
| [-0.0586, -0.0956, 0.0403, ..., -0.0447, 0.0010, 0.0103]], |
| device='cuda:0'), grad: tensor([[ 9.2328e-05, 0.0000e+00, 1.8060e-05, ..., 1.2480e-07, |
| 1.3292e-04, 3.7253e-09], |
| [ 3.3855e-05, 0.0000e+00, 1.5460e-06, ..., 1.7025e-06, |
| 8.7991e-06, 0.0000e+00], |
| [-4.5121e-05, 0.0000e+00, 6.0461e-06, ..., 8.3633e-07, |
| 6.4932e-06, 0.0000e+00], |
| ..., |
| [-7.1228e-06, 0.0000e+00, -4.9695e-06, ..., 1.5073e-05, |
| 4.0442e-05, 0.0000e+00], |
| [ 2.5570e-05, 0.0000e+00, -4.3809e-06, ..., 6.4634e-07, |
| -3.2812e-05, 1.4901e-08], |
| [-8.4221e-05, 0.0000e+00, -1.4782e-04, ..., 2.6580e-06, |
| -2.7251e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 79, bias, value: tensor([-0.0045, 0.0297, 0.0063, 0.0016, 0.0076, -0.0024, 0.0181, -0.0226, |
| 0.0188, 0.0018], device='cuda:0'), grad: tensor([ 4.7183e-04, 6.1333e-05, -6.2108e-05, -1.4615e-04, 6.9666e-04, |
| 6.6876e-05, 9.8869e-06, 3.1382e-05, -5.8174e-05, -1.0710e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 78---------------------------------------------------- |
| epoch 78, time 218.26, cls_loss 0.0097 cls_loss_mapping 0.0133 cls_loss_causal 0.6282 re_mapping 0.0121 re_causal 0.0346 /// teacc 98.79 lr 0.00010000 |
| Epoch 80, weight, value: tensor([[-0.0289, -0.0198, -0.0275, ..., -0.0754, 0.0105, 0.0049], |
| [-0.0217, 0.0221, -0.0861, ..., 0.0336, 0.0338, -0.0620], |
| [ 0.0463, -0.0040, -0.0467, ..., -0.0322, -0.0146, -0.0046], |
| ..., |
| [ 0.0314, -0.0439, -0.0003, ..., 0.0226, -0.0035, 0.0333], |
| [-0.0166, -0.0324, -0.0359, ..., -0.0063, -0.0126, -0.0197], |
| [-0.0595, -0.0957, 0.0409, ..., -0.0447, 0.0018, 0.0122]], |
| device='cuda:0'), grad: tensor([[ 7.0930e-05, 1.9558e-07, -5.0217e-06, ..., 1.3813e-05, |
| 3.9786e-06, 3.7253e-09], |
| [ 1.4639e-04, -1.7136e-06, 3.3509e-06, ..., 1.9018e-06, |
| -8.3148e-05, 2.4214e-08], |
| [-7.3957e-04, -7.9274e-06, 1.2778e-06, ..., 2.4438e-06, |
| 4.7535e-05, 2.7940e-08], |
| ..., |
| [ 7.0858e-04, 6.4857e-06, 1.8418e-05, ..., 3.2043e-04, |
| 6.5744e-05, -1.3970e-07], |
| [ 7.5758e-05, 1.1027e-06, 2.8443e-06, ..., 1.7229e-06, |
| -7.4692e-07, 5.5879e-09], |
| [-5.4598e-04, 1.9372e-07, -3.1471e-05, ..., -3.9339e-04, |
| -8.2970e-05, 1.4901e-08]], device='cuda:0') |
| Epoch 80, bias, value: tensor([-0.0055, 0.0298, 0.0057, 0.0021, 0.0071, -0.0028, 0.0181, -0.0223, |
| 0.0192, 0.0028], device='cuda:0'), grad: tensor([ 9.1434e-05, -9.4593e-05, -1.1358e-03, 4.4203e-04, 1.4544e-04, |
| 1.0180e-04, 4.5955e-05, 1.7052e-03, 1.6081e-04, -1.4639e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 79, time 217.32, cls_loss 0.0091 cls_loss_mapping 0.0132 cls_loss_causal 0.6258 re_mapping 0.0123 re_causal 0.0346 /// teacc 98.70 lr 0.00010000 |
| Epoch 81, weight, value: tensor([[-0.0290, -0.0202, -0.0279, ..., -0.0756, 0.0100, 0.0050], |
| [-0.0223, 0.0217, -0.0868, ..., 0.0338, 0.0339, -0.0622], |
| [ 0.0470, -0.0040, -0.0466, ..., -0.0323, -0.0145, -0.0045], |
| ..., |
| [ 0.0315, -0.0419, -0.0005, ..., 0.0224, -0.0039, 0.0332], |
| [-0.0165, -0.0328, -0.0365, ..., -0.0064, -0.0124, -0.0198], |
| [-0.0605, -0.0962, 0.0407, ..., -0.0441, 0.0022, 0.0122]], |
| device='cuda:0'), grad: tensor([[ 2.7776e-05, 8.9407e-08, -2.6405e-05, ..., 0.0000e+00, |
| 6.1616e-06, 0.0000e+00], |
| [ 1.4842e-04, 3.5390e-08, 6.4820e-06, ..., 0.0000e+00, |
| 6.1333e-05, 0.0000e+00], |
| [-1.3828e-04, 1.0803e-07, 4.7684e-06, ..., 0.0000e+00, |
| 1.9372e-05, 0.0000e+00], |
| ..., |
| [-1.0990e-07, 1.4342e-07, 5.3495e-06, ..., 0.0000e+00, |
| 4.9882e-06, 0.0000e+00], |
| [-5.6219e-04, 1.0617e-07, 1.2136e-04, ..., 0.0000e+00, |
| -2.1923e-04, 0.0000e+00], |
| [-4.3064e-05, 3.5949e-07, 6.5923e-05, ..., 0.0000e+00, |
| -1.8394e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 81, bias, value: tensor([-0.0056, 0.0293, 0.0064, 0.0021, 0.0071, -0.0025, 0.0179, -0.0224, |
| 0.0191, 0.0028], device='cuda:0'), grad: tensor([-1.6165e-04, 3.9744e-04, -1.3697e-04, 1.2474e-03, 2.3350e-05, |
| -1.8339e-03, 8.6641e-04, 6.3539e-05, -4.9400e-04, 2.9072e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 80---------------------------------------------------- |
| epoch 80, time 218.07, cls_loss 0.0083 cls_loss_mapping 0.0122 cls_loss_causal 0.5981 re_mapping 0.0119 re_causal 0.0344 /// teacc 98.88 lr 0.00010000 |
| Epoch 82, weight, value: tensor([[-0.0295, -0.0203, -0.0289, ..., -0.0757, 0.0097, 0.0042], |
| [-0.0221, 0.0221, -0.0875, ..., 0.0342, 0.0343, -0.0623], |
| [ 0.0472, -0.0043, -0.0468, ..., -0.0335, -0.0148, -0.0045], |
| ..., |
| [ 0.0320, -0.0418, -0.0002, ..., 0.0227, -0.0041, 0.0332], |
| [-0.0169, -0.0341, -0.0370, ..., -0.0065, -0.0128, -0.0199], |
| [-0.0610, -0.0966, 0.0413, ..., -0.0442, 0.0028, 0.0130]], |
| device='cuda:0'), grad: tensor([[ 5.7220e-06, 1.7695e-07, -3.6713e-06, ..., 2.3805e-06, |
| 1.2200e-06, 3.7253e-09], |
| [ 4.9382e-05, 9.1456e-07, 2.1867e-06, ..., 1.8761e-05, |
| -5.6028e-06, 3.7253e-09], |
| [ 1.0651e-04, 4.0755e-06, 1.1295e-05, ..., 5.3078e-05, |
| 1.7025e-06, -6.5193e-08], |
| ..., |
| [ 1.6308e-04, 2.2724e-05, 6.4746e-06, ..., 3.0184e-04, |
| 2.1812e-06, 3.7253e-09], |
| [ 3.4869e-05, 1.2256e-06, 7.9334e-05, ..., 1.4484e-05, |
| 3.1199e-06, 2.0489e-08], |
| [ 1.7846e-04, 1.9930e-07, -4.9400e-03, ..., 3.4031e-06, |
| -1.0654e-06, 9.3132e-09]], device='cuda:0') |
| Epoch 82, bias, value: tensor([-0.0060, 0.0293, 0.0065, 0.0020, 0.0067, -0.0030, 0.0182, -0.0220, |
| 0.0185, 0.0036], device='cuda:0'), grad: tensor([-1.7142e-04, 7.7486e-05, 1.7238e-04, -8.3065e-04, 6.1455e-03, |
| -1.7178e-04, 4.5568e-05, -2.8342e-05, 2.2936e-04, -5.4665e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 81, time 217.48, cls_loss 0.0085 cls_loss_mapping 0.0113 cls_loss_causal 0.6304 re_mapping 0.0117 re_causal 0.0343 /// teacc 98.83 lr 0.00010000 |
| Epoch 83, weight, value: tensor([[-0.0297, -0.0208, -0.0294, ..., -0.0767, 0.0092, 0.0033], |
| [-0.0226, 0.0220, -0.0880, ..., 0.0343, 0.0345, -0.0626], |
| [ 0.0475, -0.0028, -0.0478, ..., -0.0341, -0.0153, -0.0042], |
| ..., |
| [ 0.0323, -0.0433, -0.0006, ..., 0.0221, -0.0041, 0.0332], |
| [-0.0169, -0.0308, -0.0373, ..., -0.0069, -0.0139, -0.0200], |
| [-0.0618, -0.0982, 0.0415, ..., -0.0443, 0.0032, 0.0138]], |
| device='cuda:0'), grad: tensor([[ 1.3448e-06, 3.7253e-08, -6.2063e-06, ..., 3.7253e-09, |
| -1.6429e-06, 7.8231e-08], |
| [ 4.7348e-06, 5.5879e-09, 1.9372e-07, ..., 1.8626e-09, |
| 7.8604e-07, 3.1665e-08], |
| [-5.2378e-06, -8.8662e-07, 2.8312e-07, ..., 0.0000e+00, |
| 4.0755e-06, -9.8720e-07], |
| ..., |
| [-9.8944e-06, 8.0094e-08, 4.0792e-07, ..., 3.7253e-09, |
| 8.3074e-07, 1.9744e-07], |
| [ 1.0118e-05, 4.9919e-07, 2.5332e-07, ..., 5.5879e-09, |
| -7.5065e-06, 5.2713e-07], |
| [ 1.1526e-05, 1.1176e-08, 3.8259e-06, ..., 1.1176e-08, |
| 4.9993e-06, 1.1176e-08]], device='cuda:0') |
| Epoch 83, bias, value: tensor([-0.0063, 0.0290, 0.0062, 0.0021, 0.0065, -0.0021, 0.0192, -0.0217, |
| 0.0179, 0.0033], device='cuda:0'), grad: tensor([-9.0420e-05, 1.2666e-05, 2.2035e-06, -5.9414e-04, 3.4347e-06, |
| 5.9795e-04, 2.0131e-05, -1.6093e-05, -5.9605e-06, 6.9916e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 82, time 217.31, cls_loss 0.0095 cls_loss_mapping 0.0135 cls_loss_causal 0.6320 re_mapping 0.0115 re_causal 0.0318 /// teacc 98.85 lr 0.00010000 |
| Epoch 84, weight, value: tensor([[-0.0307, -0.0219, -0.0295, ..., -0.0770, 0.0084, 0.0033], |
| [-0.0229, 0.0219, -0.0883, ..., 0.0352, 0.0349, -0.0627], |
| [ 0.0477, -0.0026, -0.0476, ..., -0.0344, -0.0156, -0.0041], |
| ..., |
| [ 0.0325, -0.0433, -0.0002, ..., 0.0227, -0.0041, 0.0332], |
| [-0.0162, -0.0286, -0.0376, ..., -0.0074, -0.0139, -0.0201], |
| [-0.0632, -0.0995, 0.0413, ..., -0.0446, 0.0031, 0.0140]], |
| device='cuda:0'), grad: tensor([[ 7.9051e-06, 0.0000e+00, 5.6066e-07, ..., 1.1027e-06, |
| 1.5385e-06, 1.8626e-09], |
| [ 3.9554e-04, 0.0000e+00, 7.1339e-07, ..., 9.8348e-05, |
| -3.7253e-07, 1.8626e-09], |
| [-3.3236e-04, 0.0000e+00, 2.4214e-06, ..., -1.3435e-04, |
| 7.2829e-07, -3.3528e-08], |
| ..., |
| [-4.0102e-04, 0.0000e+00, -4.8429e-06, ..., 8.5086e-06, |
| 1.6112e-06, 3.7253e-09], |
| [ 1.6546e-04, 0.0000e+00, 5.8860e-06, ..., 2.3358e-06, |
| 3.5595e-06, 1.8626e-09], |
| [ 1.1906e-05, 0.0000e+00, 1.7118e-06, ..., 6.2957e-07, |
| -3.7514e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 84, bias, value: tensor([-0.0060, 0.0289, 0.0062, 0.0016, 0.0061, -0.0018, 0.0198, -0.0207, |
| 0.0182, 0.0022], device='cuda:0'), grad: tensor([ 6.0052e-06, 7.1144e-04, -5.6505e-04, 2.3878e-04, 3.9935e-05, |
| -3.9458e-05, 2.0295e-05, -7.6151e-04, 3.2234e-04, 2.7657e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 83, time 217.22, cls_loss 0.0081 cls_loss_mapping 0.0107 cls_loss_causal 0.6117 re_mapping 0.0115 re_causal 0.0324 /// teacc 98.82 lr 0.00010000 |
| Epoch 85, weight, value: tensor([[-3.1685e-02, -2.2533e-02, -2.9514e-02, ..., -7.7276e-02, |
| 7.8431e-03, 8.4042e-04], |
| [-2.2637e-02, 2.1747e-02, -8.8562e-02, ..., 3.5310e-02, |
| 3.5274e-02, -6.3473e-02], |
| [ 4.9551e-02, -4.4498e-04, -4.7656e-02, ..., -3.4386e-02, |
| -1.5733e-02, -3.9914e-03], |
| ..., |
| [ 3.0931e-02, -4.5407e-02, -4.3666e-05, ..., 2.3128e-02, |
| -4.6640e-03, 3.3222e-02], |
| [-1.6874e-02, -2.8735e-02, -3.7578e-02, ..., -7.5971e-03, |
| -1.3717e-02, -2.0330e-02], |
| [-6.3224e-02, -1.0058e-01, 4.1021e-02, ..., -4.4998e-02, |
| 3.1767e-03, 1.6374e-02]], device='cuda:0'), grad: tensor([[-2.7150e-05, 3.7253e-09, 1.3225e-07, ..., 1.3039e-08, |
| 6.4224e-06, 3.7253e-09], |
| [ 5.2787e-06, 3.7253e-09, 1.2293e-07, ..., -2.7940e-08, |
| 5.6601e-04, 1.8626e-09], |
| [-2.3559e-05, -7.0781e-08, 2.4214e-08, ..., 3.5390e-08, |
| 1.1571e-05, -3.9116e-08], |
| ..., |
| [ 7.3686e-06, 1.8626e-08, 2.0117e-07, ..., 3.3528e-08, |
| 3.8669e-06, 7.4506e-09], |
| [ 2.4468e-05, 2.0489e-08, 3.0156e-06, ..., 4.2841e-08, |
| -7.7391e-04, 1.3039e-08], |
| [ 4.1910e-06, 0.0000e+00, 1.2629e-06, ..., 3.5390e-08, |
| 6.4783e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 85, bias, value: tensor([-0.0068, 0.0291, 0.0075, 0.0017, 0.0062, -0.0014, 0.0201, -0.0219, |
| 0.0181, 0.0021], device='cuda:0'), grad: tensor([-7.9930e-05, 1.6823e-03, 1.3895e-05, 4.8667e-05, 2.1145e-05, |
| 8.5354e-05, 3.5524e-04, 2.8700e-05, -2.1915e-03, 3.5554e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 84, time 217.28, cls_loss 0.0073 cls_loss_mapping 0.0105 cls_loss_causal 0.6257 re_mapping 0.0116 re_causal 0.0334 /// teacc 98.88 lr 0.00010000 |
| Epoch 86, weight, value: tensor([[-3.1825e-02, -2.3469e-02, -2.9618e-02, ..., -7.7369e-02, |
| 7.4945e-03, 5.2248e-04], |
| [-2.2788e-02, 2.1964e-02, -8.8911e-02, ..., 3.5498e-02, |
| 3.5503e-02, -6.3982e-02], |
| [ 4.9456e-02, 4.7805e-05, -4.8111e-02, ..., -3.4768e-02, |
| -1.6050e-02, -3.4699e-03], |
| ..., |
| [ 3.1461e-02, -4.5332e-02, -9.5766e-05, ..., 2.3146e-02, |
| -4.7482e-03, 3.3152e-02], |
| [-1.6918e-02, -2.9669e-02, -3.7855e-02, ..., -7.7859e-03, |
| -1.3847e-02, -2.0583e-02], |
| [-6.3803e-02, -1.0353e-01, 4.1037e-02, ..., -4.5113e-02, |
| 3.5997e-03, 1.6679e-02]], device='cuda:0'), grad: tensor([[ 1.1260e-06, 1.7695e-08, 1.9503e-04, ..., 6.5193e-09, |
| 3.2596e-07, 3.6322e-08], |
| [ 1.2983e-06, 6.2399e-08, 6.6198e-06, ..., -2.1420e-08, |
| -4.6790e-06, 1.2107e-08], |
| [-1.0736e-05, -7.5437e-07, 4.1395e-05, ..., 3.2596e-08, |
| 1.0859e-06, -6.5099e-07], |
| ..., |
| [-2.7418e-06, 2.0489e-08, 1.0639e-05, ..., 4.2841e-08, |
| 2.0191e-06, 2.0489e-08], |
| [ 5.7966e-06, 3.8277e-07, 8.2552e-05, ..., 5.3085e-08, |
| 2.1383e-06, 3.4086e-07], |
| [ 1.7062e-06, 5.5879e-09, -1.0139e-04, ..., 1.1269e-07, |
| -1.9521e-06, 4.0978e-08]], device='cuda:0') |
| Epoch 86, bias, value: tensor([-0.0065, 0.0290, 0.0072, 0.0018, 0.0061, -0.0014, 0.0200, -0.0214, |
| 0.0178, 0.0020], device='cuda:0'), grad: tensor([ 3.5858e-04, 4.9174e-06, 6.3777e-05, 4.7207e-05, -8.9359e-04, |
| 2.5129e-04, 1.4913e-04, 2.2173e-05, 1.9062e-04, -1.9395e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 85, time 217.40, cls_loss 0.0081 cls_loss_mapping 0.0098 cls_loss_causal 0.5964 re_mapping 0.0113 re_causal 0.0326 /// teacc 98.49 lr 0.00010000 |
| Epoch 87, weight, value: tensor([[-3.1343e-02, -2.4470e-02, -2.9745e-02, ..., -7.7509e-02, |
| 7.6419e-03, 5.0189e-04], |
| [-2.2495e-02, 2.0924e-02, -8.9341e-02, ..., 3.6480e-02, |
| 3.6347e-02, -6.4996e-02], |
| [ 4.9868e-02, 1.4075e-03, -4.8820e-02, ..., -3.4958e-02, |
| -1.5985e-02, -3.3829e-03], |
| ..., |
| [ 3.1334e-02, -4.5443e-02, 5.2919e-05, ..., 2.3273e-02, |
| -5.9362e-03, 3.3070e-02], |
| [-1.6709e-02, -3.0405e-02, -3.7853e-02, ..., -7.9055e-03, |
| -1.3858e-02, -2.2335e-02], |
| [-6.4422e-02, -1.0611e-01, 4.0836e-02, ..., -4.5450e-02, |
| 3.9890e-03, 1.6682e-02]], device='cuda:0'), grad: tensor([[ 4.0978e-07, 9.3132e-10, -1.0170e-05, ..., 0.0000e+00, |
| 5.4296e-07, 7.4506e-09], |
| [ 1.5544e-06, 9.3132e-10, 2.5164e-06, ..., 0.0000e+00, |
| -3.7178e-06, 6.5193e-09], |
| [-1.3791e-05, -1.3970e-08, 5.5507e-07, ..., 0.0000e+00, |
| 8.4005e-07, -9.8720e-08], |
| ..., |
| [-8.5030e-07, 9.3132e-09, 1.6736e-06, ..., 0.0000e+00, |
| 5.3823e-05, 4.9360e-08], |
| [ 7.9200e-06, 9.3132e-10, 6.5506e-05, ..., 0.0000e+00, |
| 9.7379e-06, 1.0245e-08], |
| [ 1.4817e-06, 0.0000e+00, 1.8880e-05, ..., 0.0000e+00, |
| -1.5497e-04, 1.8626e-09]], device='cuda:0') |
| Epoch 87, bias, value: tensor([-0.0061, 0.0293, 0.0075, 0.0013, 0.0064, -0.0016, 0.0196, -0.0218, |
| 0.0183, 0.0017], device='cuda:0'), grad: tensor([-1.3685e-04, 2.7679e-06, -4.1686e-06, 8.9228e-05, 3.0661e-04, |
| -2.4045e-04, 2.9087e-05, 2.0981e-04, 2.5105e-04, -5.0783e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 86, time 217.68, cls_loss 0.0067 cls_loss_mapping 0.0099 cls_loss_causal 0.5950 re_mapping 0.0114 re_causal 0.0325 /// teacc 98.70 lr 0.00010000 |
| Epoch 88, weight, value: tensor([[-0.0312, -0.0246, -0.0295, ..., -0.0782, 0.0075, 0.0002], |
| [-0.0228, 0.0209, -0.0900, ..., 0.0364, 0.0365, -0.0667], |
| [ 0.0502, 0.0013, -0.0498, ..., -0.0350, -0.0165, -0.0029], |
| ..., |
| [ 0.0315, -0.0452, 0.0002, ..., 0.0235, -0.0060, 0.0331], |
| [-0.0170, -0.0305, -0.0380, ..., -0.0079, -0.0142, -0.0247], |
| [-0.0646, -0.1065, 0.0410, ..., -0.0454, 0.0050, 0.0170]], |
| device='cuda:0'), grad: tensor([[ 4.7907e-06, 0.0000e+00, -5.7649e-07, ..., 0.0000e+00, |
| 1.0449e-06, 6.6496e-07], |
| [ 1.1303e-05, 0.0000e+00, 2.1413e-05, ..., 4.6566e-09, |
| -1.8990e-06, 2.6338e-06], |
| [-3.0100e-05, 0.0000e+00, 4.1813e-05, ..., 1.8626e-09, |
| -6.7204e-06, 4.8764e-06], |
| ..., |
| [ 1.7524e-05, 0.0000e+00, 3.8296e-06, ..., 3.2596e-08, |
| 4.2953e-06, 1.5181e-07], |
| [ 8.2701e-06, 0.0000e+00, -2.0396e-06, ..., 9.3132e-10, |
| -4.3735e-06, 3.8464e-07], |
| [ 2.4214e-06, 0.0000e+00, 6.9737e-05, ..., 8.4843e-07, |
| -4.8727e-06, 1.0524e-07]], device='cuda:0') |
| Epoch 88, bias, value: tensor([-0.0055, 0.0290, 0.0079, 0.0007, 0.0065, -0.0022, 0.0192, -0.0216, |
| 0.0183, 0.0020], device='cuda:0'), grad: tensor([ 1.0412e-06, 7.9870e-05, 1.3269e-05, 2.9147e-05, 1.2331e-03, |
| 7.5459e-05, -1.6127e-03, 5.8949e-05, 8.2776e-06, 1.1331e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 87, time 217.48, cls_loss 0.0085 cls_loss_mapping 0.0108 cls_loss_causal 0.6280 re_mapping 0.0110 re_causal 0.0322 /// teacc 98.77 lr 0.00010000 |
| Epoch 89, weight, value: tensor([[-0.0314, -0.0251, -0.0293, ..., -0.0785, 0.0071, 0.0002], |
| [-0.0230, 0.0212, -0.0907, ..., 0.0371, 0.0365, -0.0686], |
| [ 0.0491, 0.0004, -0.0504, ..., -0.0356, -0.0174, -0.0023], |
| ..., |
| [ 0.0328, -0.0442, 0.0004, ..., 0.0232, -0.0067, 0.0329], |
| [-0.0175, -0.0311, -0.0381, ..., -0.0081, -0.0145, -0.0269], |
| [-0.0650, -0.1072, 0.0408, ..., -0.0456, 0.0066, 0.0170]], |
| device='cuda:0'), grad: tensor([[ 1.4510e-06, 6.4261e-08, 5.1036e-07, ..., 0.0000e+00, |
| -3.2574e-05, 4.5635e-08], |
| [ 3.4392e-05, 2.6412e-06, 1.1539e-06, ..., 0.0000e+00, |
| -6.3963e-06, 2.1420e-08], |
| [-2.8417e-05, -4.3958e-06, 6.0070e-07, ..., 0.0000e+00, |
| 5.9754e-06, -2.8312e-07], |
| ..., |
| [ 2.1935e-05, 1.0682e-06, 4.6752e-07, ..., 0.0000e+00, |
| 4.7982e-06, 2.9802e-08], |
| [ 1.4707e-05, 6.3330e-08, -4.3139e-06, ..., 0.0000e+00, |
| 1.6503e-06, 3.7253e-08], |
| [ 4.8801e-06, 4.8429e-08, 3.6657e-05, ..., 0.0000e+00, |
| 1.2271e-05, 9.3132e-09]], device='cuda:0') |
| Epoch 89, bias, value: tensor([-0.0057, 0.0289, 0.0068, 0.0005, 0.0063, -0.0010, 0.0189, -0.0212, |
| 0.0179, 0.0025], device='cuda:0'), grad: tensor([-6.7234e-04, 4.2140e-05, -4.7266e-05, -4.3631e-05, -1.5020e-04, |
| 5.5313e-05, 5.9032e-04, 4.2647e-05, 1.5251e-05, 1.6809e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 88, time 217.44, cls_loss 0.0074 cls_loss_mapping 0.0116 cls_loss_causal 0.5915 re_mapping 0.0114 re_causal 0.0316 /// teacc 98.73 lr 0.00010000 |
| Epoch 90, weight, value: tensor([[-3.1579e-02, -2.6152e-02, -2.9042e-02, ..., -7.8652e-02, |
| 6.8017e-03, 1.4344e-04], |
| [-2.3562e-02, 2.0799e-02, -9.0943e-02, ..., 3.7086e-02, |
| 3.6621e-02, -7.0911e-02], |
| [ 4.7967e-02, -5.5387e-04, -5.0581e-02, ..., -3.5739e-02, |
| -1.8544e-02, -1.5847e-03], |
| ..., |
| [ 3.4704e-02, -4.2498e-02, 4.0799e-05, ..., 2.3258e-02, |
| -6.1374e-03, 3.2744e-02], |
| [-1.8080e-02, -3.3078e-02, -3.8440e-02, ..., -8.2613e-03, |
| -1.4512e-02, -2.7978e-02], |
| [-6.5961e-02, -1.0942e-01, 4.0714e-02, ..., -4.5746e-02, |
| 7.1932e-03, 1.6938e-02]], device='cuda:0'), grad: tensor([[ 9.8124e-06, 1.8347e-07, 2.8126e-07, ..., 6.7707e-07, |
| 3.8259e-06, 7.4506e-09], |
| [ 4.4203e-04, 1.8571e-06, 2.7195e-07, ..., 7.1898e-06, |
| 8.2016e-05, 1.8626e-09], |
| [-1.6975e-04, -7.3671e-05, 8.1956e-08, ..., 3.5483e-07, |
| 2.9877e-05, -5.5879e-09], |
| ..., |
| [-2.3403e-03, 7.0214e-05, 1.0729e-06, ..., -3.7074e-05, |
| -1.5855e-04, 4.6566e-09], |
| [ 8.0824e-05, 6.4634e-07, 8.0653e-07, ..., 4.1258e-07, |
| 2.5138e-05, 2.0489e-08], |
| [ 1.4281e-04, 3.3528e-08, 9.7230e-07, ..., 2.3738e-05, |
| 2.8461e-05, 3.7253e-09]], device='cuda:0') |
| Epoch 90, bias, value: tensor([-0.0057, 0.0286, 0.0057, 0.0009, 0.0060, -0.0014, 0.0198, -0.0198, |
| 0.0174, 0.0025], device='cuda:0'), grad: tensor([-7.1563e-06, 1.0080e-03, -9.0539e-05, 2.0084e-03, 1.7658e-05, |
| 2.0057e-05, -1.9744e-06, -3.4561e-03, 1.9920e-04, 3.0518e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 89, time 217.67, cls_loss 0.0073 cls_loss_mapping 0.0093 cls_loss_causal 0.5837 re_mapping 0.0107 re_causal 0.0308 /// teacc 98.78 lr 0.00010000 |
| Epoch 91, weight, value: tensor([[-3.1746e-02, -2.7109e-02, -2.9686e-02, ..., -7.8891e-02, |
| 6.5501e-03, -1.6570e-03], |
| [-2.3629e-02, 2.0780e-02, -9.1575e-02, ..., 3.6932e-02, |
| 3.7027e-02, -7.1353e-02], |
| [ 4.8177e-02, 2.5558e-05, -5.0932e-02, ..., -3.5760e-02, |
| -1.8929e-02, -1.5281e-03], |
| ..., |
| [ 3.4894e-02, -4.3069e-02, 1.3605e-04, ..., 2.3350e-02, |
| -6.3779e-03, 3.2673e-02], |
| [-1.8058e-02, -3.6473e-02, -3.9193e-02, ..., -8.4988e-03, |
| -1.4524e-02, -2.8189e-02], |
| [-6.6684e-02, -1.1092e-01, 4.1127e-02, ..., -4.6283e-02, |
| 7.5552e-03, 1.8712e-02]], device='cuda:0'), grad: tensor([[ 1.2424e-06, 2.1420e-08, 2.6003e-06, ..., 0.0000e+00, |
| -2.1964e-05, 2.1607e-07], |
| [ 1.0744e-05, 6.5193e-09, 2.0824e-06, ..., 0.0000e+00, |
| 6.2399e-07, 2.3283e-08], |
| [-2.1346e-06, -2.1234e-07, 1.6196e-06, ..., 0.0000e+00, |
| 4.5374e-06, 3.0361e-07], |
| ..., |
| [-1.4091e-04, 6.5193e-08, 1.2927e-06, ..., 0.0000e+00, |
| -6.7055e-05, 4.4703e-08], |
| [ 5.2713e-06, 8.1025e-08, 1.5963e-06, ..., 0.0000e+00, |
| 5.4426e-06, 3.8184e-08], |
| [ 1.2803e-04, 2.7940e-09, 1.3679e-05, ..., 0.0000e+00, |
| 7.4387e-05, -7.7114e-07]], device='cuda:0') |
| Epoch 91, bias, value: tensor([-0.0062, 0.0285, 0.0056, 0.0005, 0.0059, -0.0013, 0.0196, -0.0199, |
| 0.0176, 0.0032], device='cuda:0'), grad: tensor([-2.4843e-04, 2.7999e-05, 1.0908e-05, 1.1347e-05, -4.0501e-05, |
| 1.3694e-05, 1.4402e-05, -4.5466e-04, 3.1620e-05, 6.3324e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 90, time 217.48, cls_loss 0.0071 cls_loss_mapping 0.0088 cls_loss_causal 0.6120 re_mapping 0.0111 re_causal 0.0317 /// teacc 98.75 lr 0.00010000 |
| Epoch 92, weight, value: tensor([[-0.0322, -0.0274, -0.0297, ..., -0.0794, 0.0062, -0.0017], |
| [-0.0244, 0.0209, -0.0918, ..., 0.0371, 0.0372, -0.0720], |
| [ 0.0483, 0.0001, -0.0511, ..., -0.0358, -0.0193, -0.0015], |
| ..., |
| [ 0.0347, -0.0432, 0.0004, ..., 0.0243, -0.0064, 0.0330], |
| [-0.0170, -0.0365, -0.0396, ..., -0.0088, -0.0141, -0.0284], |
| [-0.0680, -0.1112, 0.0409, ..., -0.0468, 0.0070, 0.0187]], |
| device='cuda:0'), grad: tensor([[ 1.1742e-05, 0.0000e+00, 1.6717e-06, ..., 0.0000e+00, |
| 2.2665e-05, 4.1723e-07], |
| [ 3.9458e-05, 0.0000e+00, 2.5239e-06, ..., 0.0000e+00, |
| 9.7901e-06, 1.8626e-09], |
| [-3.8058e-05, 0.0000e+00, 9.5833e-07, ..., 0.0000e+00, |
| 3.7700e-06, 2.7940e-09], |
| ..., |
| [ 2.9221e-05, 0.0000e+00, 2.1160e-06, ..., 0.0000e+00, |
| 5.3123e-06, 9.3132e-10], |
| [ 1.7121e-05, 0.0000e+00, 6.3255e-06, ..., 0.0000e+00, |
| -2.2724e-06, 1.1176e-08], |
| [ 3.3647e-05, 0.0000e+00, 4.2655e-07, ..., 0.0000e+00, |
| -2.1141e-06, 1.1176e-08]], device='cuda:0') |
| Epoch 92, bias, value: tensor([-0.0058, 0.0281, 0.0056, 0.0013, 0.0061, -0.0017, 0.0193, -0.0199, |
| 0.0184, 0.0025], device='cuda:0'), grad: tensor([ 2.1122e-06, 1.1557e-04, 6.3404e-06, 1.5147e-05, -7.5400e-05, |
| 2.9594e-05, -2.5630e-04, 6.0707e-05, 4.5806e-05, 5.6356e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 91, time 217.71, cls_loss 0.0082 cls_loss_mapping 0.0113 cls_loss_causal 0.5898 re_mapping 0.0106 re_causal 0.0300 /// teacc 98.74 lr 0.00010000 |
| Epoch 93, weight, value: tensor([[-0.0327, -0.0276, -0.0300, ..., -0.0796, 0.0058, -0.0017], |
| [-0.0256, 0.0210, -0.0923, ..., 0.0383, 0.0380, -0.0738], |
| [ 0.0484, 0.0001, -0.0511, ..., -0.0359, -0.0198, -0.0015], |
| ..., |
| [ 0.0346, -0.0431, 0.0007, ..., 0.0239, -0.0064, 0.0339], |
| [-0.0152, -0.0369, -0.0400, ..., -0.0089, -0.0150, -0.0288], |
| [-0.0696, -0.1117, 0.0409, ..., -0.0471, 0.0072, 0.0187]], |
| device='cuda:0'), grad: tensor([[ 1.6345e-06, 0.0000e+00, -2.1327e-07, ..., 0.0000e+00, |
| 9.4995e-07, 1.0245e-08], |
| [ 9.8720e-06, 0.0000e+00, 2.3022e-06, ..., 0.0000e+00, |
| 5.2676e-06, 8.3819e-08], |
| [ 1.1444e-05, 0.0000e+00, 3.8836e-07, ..., 0.0000e+00, |
| 2.6487e-06, 1.3970e-08], |
| ..., |
| [-7.4387e-05, 0.0000e+00, -2.9467e-06, ..., 0.0000e+00, |
| -1.2696e-05, 6.3330e-08], |
| [ 7.1935e-06, 0.0000e+00, 2.0880e-06, ..., 0.0000e+00, |
| 2.8312e-05, 6.5193e-09], |
| [ 4.0323e-05, 0.0000e+00, -8.0705e-05, ..., 0.0000e+00, |
| -2.6989e-04, 1.1269e-07]], device='cuda:0') |
| Epoch 93, bias, value: tensor([-0.0055, 0.0282, 0.0057, 0.0009, 0.0062, -0.0014, 0.0195, -0.0204, |
| 0.0192, 0.0019], device='cuda:0'), grad: tensor([-3.8457e-04, 3.9786e-05, 2.4736e-05, -1.7537e-06, 8.3256e-04, |
| 2.4509e-04, 1.2743e-04, -1.9467e-04, 9.7811e-05, -7.8630e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 92, time 217.78, cls_loss 0.0079 cls_loss_mapping 0.0109 cls_loss_causal 0.6241 re_mapping 0.0104 re_causal 0.0300 /// teacc 98.67 lr 0.00010000 |
| Epoch 94, weight, value: tensor([[-0.0335, -0.0284, -0.0298, ..., -0.0800, 0.0075, -0.0018], |
| [-0.0254, 0.0211, -0.0932, ..., 0.0390, 0.0385, -0.0767], |
| [ 0.0486, 0.0006, -0.0513, ..., -0.0360, -0.0200, -0.0013], |
| ..., |
| [ 0.0347, -0.0436, 0.0005, ..., 0.0237, -0.0067, 0.0343], |
| [-0.0154, -0.0376, -0.0412, ..., -0.0090, -0.0151, -0.0310], |
| [-0.0708, -0.1126, 0.0407, ..., -0.0479, 0.0077, 0.0189]], |
| device='cuda:0'), grad: tensor([[ 3.9227e-06, 0.0000e+00, 1.1586e-05, ..., 1.3132e-07, |
| 4.2245e-06, 0.0000e+00], |
| [ 2.0355e-05, 0.0000e+00, 6.8992e-06, ..., 2.9057e-07, |
| 2.5891e-07, 0.0000e+00], |
| [-2.3365e-05, 0.0000e+00, -2.0117e-06, ..., 6.9849e-08, |
| 6.9384e-07, 0.0000e+00], |
| ..., |
| [-2.1830e-05, 0.0000e+00, 7.8380e-06, ..., 2.4494e-07, |
| 6.2808e-06, 0.0000e+00], |
| [ 5.6356e-05, 0.0000e+00, 1.0610e-05, ..., 5.4203e-07, |
| 1.2092e-05, 0.0000e+00], |
| [-8.9686e-07, 0.0000e+00, 3.0994e-05, ..., 8.9407e-08, |
| -2.6330e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 94, bias, value: tensor([-0.0044, 0.0283, 0.0060, 0.0006, 0.0068, -0.0011, 0.0183, -0.0203, |
| 0.0190, 0.0012], device='cuda:0'), grad: tensor([ 6.1356e-06, 5.4240e-05, -3.8385e-05, 3.0786e-05, 2.3693e-05, |
| -1.9336e-04, 2.2739e-05, -1.4365e-05, 1.3149e-04, -2.3007e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 93, time 217.53, cls_loss 0.0064 cls_loss_mapping 0.0085 cls_loss_causal 0.5821 re_mapping 0.0112 re_causal 0.0301 /// teacc 98.77 lr 0.00010000 |
| Epoch 95, weight, value: tensor([[-0.0337, -0.0289, -0.0299, ..., -0.0801, 0.0071, -0.0018], |
| [-0.0257, 0.0209, -0.0937, ..., 0.0391, 0.0383, -0.0771], |
| [ 0.0490, -0.0002, -0.0512, ..., -0.0360, -0.0202, -0.0012], |
| ..., |
| [ 0.0351, -0.0427, 0.0004, ..., 0.0236, -0.0069, 0.0343], |
| [-0.0163, -0.0364, -0.0413, ..., -0.0091, -0.0152, -0.0313], |
| [-0.0713, -0.1136, 0.0402, ..., -0.0481, 0.0087, 0.0189]], |
| device='cuda:0'), grad: tensor([[ 2.8685e-06, 0.0000e+00, -4.3027e-07, ..., 0.0000e+00, |
| 4.1053e-06, 0.0000e+00], |
| [ 5.6252e-06, 0.0000e+00, 1.0505e-06, ..., 0.0000e+00, |
| 3.9935e-06, 0.0000e+00], |
| [-2.7657e-04, 0.0000e+00, 1.4780e-06, ..., 0.0000e+00, |
| -5.3763e-05, 0.0000e+00], |
| ..., |
| [ 5.6103e-06, 0.0000e+00, 3.8296e-06, ..., 0.0000e+00, |
| 2.5593e-06, 0.0000e+00], |
| [ 4.9353e-05, 0.0000e+00, 5.8208e-07, ..., 0.0000e+00, |
| -5.4762e-06, 0.0000e+00], |
| [ 1.3048e-06, 0.0000e+00, 9.8441e-07, ..., 0.0000e+00, |
| -5.4687e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 95, bias, value: tensor([-0.0041, 0.0277, 0.0060, 0.0004, 0.0066, -0.0012, 0.0192, -0.0197, |
| 0.0181, 0.0013], device='cuda:0'), grad: tensor([-1.1310e-05, 2.5257e-05, -3.2830e-04, 2.7680e-04, 8.7693e-06, |
| 1.5780e-05, -4.0382e-05, 5.8264e-06, 2.8074e-05, 1.9282e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 94, time 217.70, cls_loss 0.0068 cls_loss_mapping 0.0090 cls_loss_causal 0.5953 re_mapping 0.0108 re_causal 0.0296 /// teacc 98.76 lr 0.00010000 |
| Epoch 96, weight, value: tensor([[-0.0332, -0.0295, -0.0301, ..., -0.0801, 0.0071, -0.0027], |
| [-0.0264, 0.0209, -0.0953, ..., 0.0391, 0.0382, -0.0777], |
| [ 0.0489, -0.0002, -0.0499, ..., -0.0361, -0.0207, -0.0007], |
| ..., |
| [ 0.0356, -0.0426, 0.0003, ..., 0.0236, -0.0074, 0.0342], |
| [-0.0162, -0.0369, -0.0418, ..., -0.0092, -0.0156, -0.0314], |
| [-0.0722, -0.1155, 0.0398, ..., -0.0483, 0.0097, 0.0198]], |
| device='cuda:0'), grad: tensor([[ 1.5683e-06, 0.0000e+00, 9.8534e-07, ..., 0.0000e+00, |
| 3.6806e-06, 2.4214e-08], |
| [-9.3639e-05, 0.0000e+00, 3.6042e-07, ..., 0.0000e+00, |
| -2.0456e-04, 1.8626e-09], |
| [-2.2817e-06, 0.0000e+00, 1.1735e-07, ..., 0.0000e+00, |
| 6.6198e-06, -9.6858e-08], |
| ..., |
| [-1.4678e-05, 0.0000e+00, 2.6356e-07, ..., 0.0000e+00, |
| 1.8373e-05, 2.0489e-08], |
| [ 7.4446e-05, 0.0000e+00, 1.0848e-05, ..., 0.0000e+00, |
| 1.9288e-04, 1.8626e-08], |
| [ 3.4552e-06, 0.0000e+00, 1.7444e-06, ..., 0.0000e+00, |
| -5.3734e-05, 6.5193e-09]], device='cuda:0') |
| Epoch 96, bias, value: tensor([-3.8285e-03, 2.7058e-02, 5.7196e-03, 4.5091e-05, 6.8383e-03, |
| -1.6129e-03, 1.9928e-02, -1.9388e-02, 1.8173e-02, 1.3574e-03], |
| device='cuda:0'), grad: tensor([ 1.0476e-05, -4.3154e-04, 7.2643e-06, 4.7475e-05, 3.8713e-05, |
| -1.9848e-04, 1.8823e-04, 2.1607e-05, 4.4274e-04, -1.2684e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 95, time 217.70, cls_loss 0.0067 cls_loss_mapping 0.0095 cls_loss_causal 0.5823 re_mapping 0.0103 re_causal 0.0293 /// teacc 98.85 lr 0.00010000 |
| Epoch 97, weight, value: tensor([[-0.0331, -0.0296, -0.0303, ..., -0.0802, 0.0070, -0.0033], |
| [-0.0267, 0.0209, -0.0961, ..., 0.0392, 0.0389, -0.0779], |
| [ 0.0491, -0.0002, -0.0502, ..., -0.0361, -0.0209, -0.0006], |
| ..., |
| [ 0.0356, -0.0426, 0.0006, ..., 0.0236, -0.0087, 0.0342], |
| [-0.0166, -0.0374, -0.0426, ..., -0.0093, -0.0163, -0.0314], |
| [-0.0717, -0.1157, 0.0396, ..., -0.0484, 0.0107, 0.0204]], |
| device='cuda:0'), grad: tensor([[ 5.3905e-06, 9.3132e-09, -1.1951e-05, ..., 0.0000e+00, |
| 1.0353e-04, 0.0000e+00], |
| [ 3.4533e-06, 2.1420e-08, 5.6438e-07, ..., 0.0000e+00, |
| -4.2629e-04, 0.0000e+00], |
| [-3.0383e-05, -5.0571e-07, 2.6450e-07, ..., 0.0000e+00, |
| 3.0875e-05, 0.0000e+00], |
| ..., |
| [ 6.2659e-06, 3.0547e-07, 1.0934e-06, ..., 0.0000e+00, |
| 1.9431e-05, 0.0000e+00], |
| [ 1.3940e-05, 1.0151e-07, 1.5991e-06, ..., 0.0000e+00, |
| 1.8382e-04, 0.0000e+00], |
| [ 1.6302e-05, 9.3132e-10, 1.0524e-06, ..., 0.0000e+00, |
| 1.2964e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 97, bias, value: tensor([-0.0044, 0.0272, 0.0058, -0.0002, 0.0070, -0.0013, 0.0195, -0.0194, |
| 0.0173, 0.0024], device='cuda:0'), grad: tensor([ 1.3947e-04, -8.3208e-04, 2.7224e-05, 4.8429e-07, 5.4121e-05, |
| 5.3167e-05, 1.0347e-04, 5.0157e-05, 3.6764e-04, 3.7253e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 96, time 218.00, cls_loss 0.0065 cls_loss_mapping 0.0105 cls_loss_causal 0.6096 re_mapping 0.0104 re_causal 0.0302 /// teacc 98.85 lr 0.00010000 |
| Epoch 98, weight, value: tensor([[-0.0333, -0.0296, -0.0303, ..., -0.0803, 0.0063, -0.0033], |
| [-0.0270, 0.0209, -0.0973, ..., 0.0392, 0.0411, -0.0783], |
| [ 0.0490, -0.0002, -0.0507, ..., -0.0361, -0.0218, -0.0003], |
| ..., |
| [ 0.0359, -0.0426, 0.0009, ..., 0.0240, -0.0096, 0.0342], |
| [-0.0169, -0.0373, -0.0425, ..., -0.0093, -0.0166, -0.0316], |
| [-0.0730, -0.1157, 0.0383, ..., -0.0485, 0.0098, 0.0204]], |
| device='cuda:0'), grad: tensor([[ 9.4157e-07, 0.0000e+00, 2.0452e-06, ..., 0.0000e+00, |
| 1.8045e-05, 5.8934e-06], |
| [ 1.5616e-05, 0.0000e+00, 8.4005e-07, ..., 0.0000e+00, |
| -5.6811e-06, 3.9116e-08], |
| [-2.0695e-03, 0.0000e+00, 5.0385e-07, ..., 0.0000e+00, |
| 2.2054e-06, 4.9360e-08], |
| ..., |
| [ 3.4332e-05, 0.0000e+00, 3.7346e-07, ..., 0.0000e+00, |
| 4.5411e-06, 1.5832e-08], |
| [ 2.0199e-03, 0.0000e+00, 1.4147e-06, ..., 0.0000e+00, |
| 5.1633e-06, 2.9150e-07], |
| [ 4.5076e-06, 0.0000e+00, 7.2829e-06, ..., 0.0000e+00, |
| 5.1744e-06, 2.1141e-07]], device='cuda:0') |
| Epoch 98, bias, value: tensor([-0.0042, 0.0278, 0.0052, -0.0019, 0.0079, 0.0004, 0.0197, -0.0191, |
| 0.0174, 0.0009], device='cuda:0'), grad: tensor([ 7.0572e-05, 6.8724e-05, -1.0391e-02, -1.7971e-05, 8.3745e-06, |
| 2.9594e-05, -9.5189e-05, 1.7679e-04, 1.0132e-02, 2.8670e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 97, time 219.00, cls_loss 0.0060 cls_loss_mapping 0.0092 cls_loss_causal 0.6126 re_mapping 0.0106 re_causal 0.0308 /// teacc 98.75 lr 0.00010000 |
| Epoch 99, weight, value: tensor([[-0.0339, -0.0298, -0.0312, ..., -0.0804, 0.0058, -0.0038], |
| [-0.0272, 0.0208, -0.0988, ..., 0.0392, 0.0420, -0.0788], |
| [ 0.0487, -0.0002, -0.0513, ..., -0.0361, -0.0223, 0.0007], |
| ..., |
| [ 0.0366, -0.0425, 0.0013, ..., 0.0241, -0.0114, 0.0342], |
| [-0.0174, -0.0375, -0.0420, ..., -0.0093, -0.0159, -0.0318], |
| [-0.0740, -0.1160, 0.0385, ..., -0.0486, 0.0104, 0.0209]], |
| device='cuda:0'), grad: tensor([[ 1.1455e-06, 0.0000e+00, 3.3118e-06, ..., 0.0000e+00, |
| 2.7604e-06, 0.0000e+00], |
| [ 1.1334e-06, 0.0000e+00, 2.2903e-05, ..., 0.0000e+00, |
| 2.7925e-05, 0.0000e+00], |
| [-2.0444e-05, 0.0000e+00, 2.8647e-06, ..., 0.0000e+00, |
| 3.4068e-06, 0.0000e+00], |
| ..., |
| [ 1.0906e-06, 0.0000e+00, 1.6332e-05, ..., 0.0000e+00, |
| 1.2338e-05, 0.0000e+00], |
| [ 9.9763e-06, 0.0000e+00, 2.6286e-05, ..., 0.0000e+00, |
| 3.4183e-05, 0.0000e+00], |
| [-9.5218e-06, 0.0000e+00, -1.2856e-03, ..., 0.0000e+00, |
| -1.5278e-03, 0.0000e+00]], device='cuda:0') |
| Epoch 99, bias, value: tensor([-0.0046, 0.0282, 0.0053, -0.0019, 0.0082, 0.0003, 0.0200, -0.0192, |
| 0.0171, 0.0006], device='cuda:0'), grad: tensor([ 1.3858e-05, 1.1659e-04, -2.6584e-05, 2.7001e-05, 5.9814e-03, |
| 2.6464e-05, 5.9083e-06, 6.7651e-05, 1.5199e-04, -6.3629e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 98, time 220.62, cls_loss 0.0061 cls_loss_mapping 0.0087 cls_loss_causal 0.5846 re_mapping 0.0107 re_causal 0.0294 /// teacc 98.87 lr 0.00010000 |
| Epoch 100, weight, value: tensor([[-3.3953e-02, -3.0393e-02, -3.1401e-02, ..., -8.0508e-02, |
| 5.6541e-03, -3.9072e-03], |
| [-2.8621e-02, 2.0676e-02, -9.9945e-02, ..., 3.9273e-02, |
| 4.1239e-02, -7.9373e-02], |
| [ 4.9674e-02, 1.0520e-04, -5.1749e-02, ..., -3.6169e-02, |
| -2.1992e-02, 3.4256e-04], |
| ..., |
| [ 3.6689e-02, -4.3312e-02, 6.7844e-04, ..., 2.4578e-02, |
| -1.0763e-02, 3.4235e-02], |
| [-1.7645e-02, -3.4443e-02, -4.2634e-02, ..., -9.3818e-03, |
| -1.5728e-02, -3.1914e-02], |
| [-7.5739e-02, -1.1693e-01, 3.9002e-02, ..., -4.8760e-02, |
| 1.0871e-02, 2.0944e-02]], device='cuda:0'), grad: tensor([[ 4.7684e-06, 0.0000e+00, -7.3668e-07, ..., 0.0000e+00, |
| 3.2447e-06, 0.0000e+00], |
| [ 2.9672e-06, 0.0000e+00, 8.9407e-08, ..., 0.0000e+00, |
| -2.3264e-06, 9.3132e-10], |
| [ 3.5405e-05, 0.0000e+00, 2.9709e-07, ..., 0.0000e+00, |
| 2.1495e-06, 9.3132e-10], |
| ..., |
| [ 4.1038e-05, 0.0000e+00, 2.0862e-07, ..., 0.0000e+00, |
| 2.0117e-06, -3.7253e-09], |
| [ 3.2037e-05, 0.0000e+00, -1.1316e-06, ..., 0.0000e+00, |
| -1.0459e-06, 0.0000e+00], |
| [ 1.9386e-05, 0.0000e+00, 1.7621e-06, ..., 0.0000e+00, |
| -1.4435e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 100, bias, value: tensor([-0.0045, 0.0270, 0.0061, -0.0019, 0.0081, 0.0003, 0.0199, -0.0191, |
| 0.0172, 0.0007], device='cuda:0'), grad: tensor([ 9.5665e-06, 1.8738e-06, 6.9201e-05, -2.7299e-04, 1.2249e-05, |
| 2.7418e-05, -2.0176e-05, 7.9393e-05, 5.6714e-05, 3.6359e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 99, time 219.76, cls_loss 0.0061 cls_loss_mapping 0.0098 cls_loss_causal 0.5776 re_mapping 0.0097 re_causal 0.0281 /// teacc 98.79 lr 0.00010000 |
| Epoch 101, weight, value: tensor([[-0.0344, -0.0308, -0.0316, ..., -0.0808, 0.0052, -0.0039], |
| [-0.0280, 0.0221, -0.1008, ..., 0.0393, 0.0423, -0.0805], |
| [ 0.0494, -0.0005, -0.0518, ..., -0.0363, -0.0229, 0.0003], |
| ..., |
| [ 0.0369, -0.0435, 0.0007, ..., 0.0246, -0.0113, 0.0342], |
| [-0.0179, -0.0344, -0.0427, ..., -0.0095, -0.0154, -0.0321], |
| [-0.0764, -0.1177, 0.0383, ..., -0.0488, 0.0093, 0.0210]], |
| device='cuda:0'), grad: tensor([[ 8.3353e-07, 0.0000e+00, 6.8918e-08, ..., 9.3132e-10, |
| 2.0228e-06, 0.0000e+00], |
| [ 9.1866e-06, 0.0000e+00, 1.9092e-07, ..., 4.6566e-09, |
| -4.9263e-05, 0.0000e+00], |
| [ 7.2643e-06, 0.0000e+00, 6.3330e-08, ..., 9.3132e-10, |
| 1.4089e-05, 0.0000e+00], |
| ..., |
| [-1.6123e-05, 0.0000e+00, -3.3062e-07, ..., 6.5193e-09, |
| 6.4969e-06, 0.0000e+00], |
| [-7.9162e-07, 0.0000e+00, 6.7055e-08, ..., 0.0000e+00, |
| -1.4335e-05, 0.0000e+00], |
| [ 5.4426e-06, 0.0000e+00, 1.1902e-06, ..., 1.8626e-09, |
| 2.3276e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 101, bias, value: tensor([-0.0043, 0.0280, 0.0055, -0.0019, 0.0094, 0.0004, 0.0195, -0.0191, |
| 0.0174, -0.0006], device='cuda:0'), grad: tensor([ 8.6520e-07, -1.1861e-04, 4.8667e-05, -3.8683e-05, 2.2829e-05, |
| 4.7565e-05, 1.4812e-05, -1.3679e-05, -3.6657e-05, 7.2837e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 100, time 220.23, cls_loss 0.0049 cls_loss_mapping 0.0077 cls_loss_causal 0.5981 re_mapping 0.0095 re_causal 0.0302 /// teacc 98.85 lr 0.00010000 |
| Epoch 102, weight, value: tensor([[-3.6489e-02, -3.0842e-02, -3.2005e-02, ..., -8.1052e-02, |
| 4.7590e-03, -5.4737e-03], |
| [-2.8206e-02, 2.2080e-02, -1.0224e-01, ..., 3.9480e-02, |
| 4.2716e-02, -8.1424e-02], |
| [ 4.9432e-02, -5.1822e-04, -5.2023e-02, ..., -3.6279e-02, |
| -2.3179e-02, 7.6837e-05], |
| ..., |
| [ 3.7007e-02, -4.3526e-02, 6.6154e-04, ..., 2.4375e-02, |
| -1.1564e-02, 3.4171e-02], |
| [-1.8206e-02, -3.4415e-02, -4.2652e-02, ..., -9.5949e-03, |
| -1.5960e-02, -3.3009e-02], |
| [-7.6831e-02, -1.1773e-01, 3.8692e-02, ..., -4.9303e-02, |
| 9.9592e-03, 2.2451e-02]], device='cuda:0'), grad: tensor([[ 6.8545e-07, 0.0000e+00, 4.5355e-07, ..., 0.0000e+00, |
| -2.2911e-07, 9.3132e-10], |
| [ 6.9514e-06, 0.0000e+00, 3.3323e-06, ..., 0.0000e+00, |
| 6.7391e-06, 0.0000e+00], |
| [ 4.8131e-06, 0.0000e+00, 9.9558e-07, ..., 0.0000e+00, |
| 1.6466e-05, 0.0000e+00], |
| ..., |
| [-2.0906e-05, 0.0000e+00, 4.2111e-05, ..., 0.0000e+00, |
| 1.5534e-06, 9.3132e-10], |
| [ 4.7199e-06, 0.0000e+00, 1.2564e-06, ..., 0.0000e+00, |
| -6.0886e-05, 9.3132e-09], |
| [ 1.3299e-06, 0.0000e+00, 1.7929e-04, ..., 0.0000e+00, |
| -5.2899e-07, 1.8626e-09]], device='cuda:0') |
| Epoch 102, bias, value: tensor([-0.0053, 0.0281, 0.0052, -0.0015, 0.0089, 0.0005, 0.0198, -0.0191, |
| 0.0170, 0.0002], device='cuda:0'), grad: tensor([-7.5847e-06, 5.0664e-05, 5.9903e-05, -2.1353e-05, -3.3379e-04, |
| 1.0902e-04, 2.3350e-05, 3.3051e-05, -1.8108e-04, 2.6751e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 101, time 219.97, cls_loss 0.0055 cls_loss_mapping 0.0081 cls_loss_causal 0.5910 re_mapping 0.0098 re_causal 0.0284 /// teacc 98.82 lr 0.00010000 |
| Epoch 103, weight, value: tensor([[-3.6805e-02, -3.1012e-02, -3.2107e-02, ..., -8.1267e-02, |
| 4.7487e-03, -5.6299e-03], |
| [-2.8118e-02, 2.2084e-02, -1.0252e-01, ..., 3.9501e-02, |
| 4.2437e-02, -8.1695e-02], |
| [ 4.8978e-02, -2.9018e-04, -5.1800e-02, ..., -3.6337e-02, |
| -2.3783e-02, 9.4734e-05], |
| ..., |
| [ 3.7675e-02, -4.3732e-02, 5.3099e-04, ..., 2.4381e-02, |
| -1.1898e-02, 3.4208e-02], |
| [-1.8497e-02, -3.4439e-02, -4.2968e-02, ..., -9.6682e-03, |
| -1.6326e-02, -3.3269e-02], |
| [-7.7103e-02, -1.1787e-01, 3.8125e-02, ..., -4.9589e-02, |
| 1.1430e-02, 2.2593e-02]], device='cuda:0'), grad: tensor([[ 7.4320e-07, 0.0000e+00, 1.9997e-05, ..., 1.2573e-07, |
| 3.8058e-05, 3.7253e-09], |
| [ 3.3472e-06, 0.0000e+00, 3.8370e-06, ..., 3.5763e-07, |
| 1.8710e-06, 3.7253e-09], |
| [-4.1537e-06, 0.0000e+00, 5.2862e-06, ..., 7.7300e-08, |
| 4.3996e-06, -8.4750e-08], |
| ..., |
| [-1.9163e-05, 0.0000e+00, 2.0787e-06, ..., -4.3698e-06, |
| 3.8818e-06, 1.3039e-08], |
| [ 1.2266e-06, 0.0000e+00, 4.3772e-07, ..., 4.9360e-08, |
| 1.7673e-05, 3.0734e-08], |
| [ 1.4603e-05, 0.0000e+00, -8.8096e-05, ..., 3.9563e-06, |
| -1.3924e-04, 3.7253e-09]], device='cuda:0') |
| Epoch 103, bias, value: tensor([-0.0048, 0.0281, 0.0043, -0.0017, 0.0088, 0.0003, 0.0202, -0.0185, |
| 0.0166, 0.0005], device='cuda:0'), grad: tensor([ 2.3580e-04, 3.7789e-05, 1.9640e-05, 2.2024e-05, 6.1655e-04, |
| 1.6823e-05, 2.2396e-05, -6.7279e-06, 4.7803e-05, -1.0118e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 102, time 220.05, cls_loss 0.0062 cls_loss_mapping 0.0108 cls_loss_causal 0.5879 re_mapping 0.0097 re_causal 0.0290 /// teacc 98.87 lr 0.00010000 |
| Epoch 104, weight, value: tensor([[-0.0372, -0.0314, -0.0326, ..., -0.0813, 0.0045, -0.0057], |
| [-0.0287, 0.0221, -0.1038, ..., 0.0397, 0.0423, -0.0833], |
| [ 0.0489, -0.0010, -0.0521, ..., -0.0364, -0.0239, 0.0015], |
| ..., |
| [ 0.0382, -0.0427, 0.0004, ..., 0.0244, -0.0124, 0.0340], |
| [-0.0191, -0.0345, -0.0431, ..., -0.0097, -0.0165, -0.0338], |
| [-0.0773, -0.1185, 0.0383, ..., -0.0499, 0.0114, 0.0225]], |
| device='cuda:0'), grad: tensor([[ 4.8801e-06, 0.0000e+00, 3.4366e-07, ..., 0.0000e+00, |
| 1.7863e-06, -6.0629e-07], |
| [ 4.2021e-05, 0.0000e+00, 1.4063e-07, ..., 0.0000e+00, |
| -2.1011e-06, 3.6322e-08], |
| [-5.3585e-05, 0.0000e+00, 3.0827e-07, ..., 0.0000e+00, |
| 8.3167e-07, 1.5367e-07], |
| ..., |
| [-3.4332e-05, 0.0000e+00, 2.3656e-07, ..., 0.0000e+00, |
| 1.2573e-06, 9.2201e-08], |
| [ 1.2048e-05, 0.0000e+00, 1.1548e-07, ..., 0.0000e+00, |
| 2.6934e-06, 9.5926e-08], |
| [ 1.2487e-05, 0.0000e+00, 6.4913e-07, ..., 0.0000e+00, |
| 3.5763e-07, 4.9360e-08]], device='cuda:0') |
| Epoch 104, bias, value: tensor([-0.0036, 0.0277, 0.0041, -0.0008, 0.0091, -0.0004, 0.0206, -0.0183, |
| 0.0162, -0.0001], device='cuda:0'), grad: tensor([-3.6210e-06, 9.2983e-05, -1.3793e-04, -1.1981e-05, 1.0997e-05, |
| 1.8626e-05, 2.5213e-05, -5.8264e-05, 2.3052e-05, 4.0591e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 103, time 219.73, cls_loss 0.0055 cls_loss_mapping 0.0080 cls_loss_causal 0.5779 re_mapping 0.0097 re_causal 0.0281 /// teacc 98.79 lr 0.00010000 |
| Epoch 105, weight, value: tensor([[-0.0377, -0.0315, -0.0327, ..., -0.0816, 0.0042, -0.0059], |
| [-0.0282, 0.0222, -0.1054, ..., 0.0401, 0.0427, -0.0848], |
| [ 0.0493, -0.0010, -0.0513, ..., -0.0364, -0.0243, 0.0016], |
| ..., |
| [ 0.0382, -0.0427, 0.0005, ..., 0.0247, -0.0134, 0.0341], |
| [-0.0195, -0.0345, -0.0432, ..., -0.0098, -0.0164, -0.0341], |
| [-0.0782, -0.1185, 0.0380, ..., -0.0498, 0.0116, 0.0225]], |
| device='cuda:0'), grad: tensor([[ 9.7603e-07, 0.0000e+00, 6.4075e-07, ..., 0.0000e+00, |
| 3.6448e-05, 2.1420e-08], |
| [-9.6798e-05, 0.0000e+00, 1.9465e-07, ..., 0.0000e+00, |
| -2.4629e-04, 7.4506e-09], |
| [ 7.4357e-06, 0.0000e+00, 3.0641e-07, ..., 0.0000e+00, |
| 1.6451e-05, 3.7253e-09], |
| ..., |
| [ 6.8307e-05, 0.0000e+00, 2.0862e-07, ..., 0.0000e+00, |
| 1.7691e-04, 1.5832e-08], |
| [ 3.5577e-06, 0.0000e+00, 9.3691e-07, ..., 0.0000e+00, |
| 3.7938e-05, 1.9185e-07], |
| [ 8.3745e-06, 0.0000e+00, 2.2054e-06, ..., 0.0000e+00, |
| -4.7654e-05, 1.6764e-07]], device='cuda:0') |
| Epoch 105, bias, value: tensor([-0.0042, 0.0281, 0.0044, -0.0022, 0.0098, 0.0007, 0.0211, -0.0186, |
| 0.0160, -0.0006], device='cuda:0'), grad: tensor([ 3.2902e-04, -7.4434e-04, 1.5211e-04, 9.5963e-05, 1.7428e-04, |
| 9.2313e-06, -8.7991e-06, 5.7507e-04, -1.1958e-05, -5.7125e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 104, time 219.80, cls_loss 0.0053 cls_loss_mapping 0.0086 cls_loss_causal 0.6177 re_mapping 0.0097 re_causal 0.0289 /// teacc 98.84 lr 0.00010000 |
| Epoch 106, weight, value: tensor([[-0.0375, -0.0318, -0.0335, ..., -0.0818, 0.0043, -0.0078], |
| [-0.0294, 0.0224, -0.1056, ..., 0.0403, 0.0426, -0.0851], |
| [ 0.0495, -0.0010, -0.0514, ..., -0.0367, -0.0229, 0.0016], |
| ..., |
| [ 0.0389, -0.0428, 0.0004, ..., 0.0246, -0.0140, 0.0341], |
| [-0.0201, -0.0335, -0.0428, ..., -0.0099, -0.0162, -0.0342], |
| [-0.0784, -0.1199, 0.0380, ..., -0.0498, 0.0120, 0.0244]], |
| device='cuda:0'), grad: tensor([[ 2.0377e-06, 3.2596e-09, 1.2163e-06, ..., 2.6962e-07, |
| 5.1316e-07, 1.3504e-08], |
| [ 1.1802e-05, 4.7963e-08, 1.4426e-06, ..., 1.4808e-06, |
| -2.5108e-06, 2.2817e-08], |
| [-1.8716e-05, -2.3795e-07, 3.4971e-07, ..., 2.9616e-07, |
| 4.3586e-07, -1.3085e-07], |
| ..., |
| [-8.3983e-05, 1.5227e-07, -1.7285e-05, ..., -1.8656e-05, |
| 2.1756e-06, 2.5611e-08], |
| [ 2.4855e-05, 8.3819e-09, 6.1421e-07, ..., 2.2911e-07, |
| -2.1867e-06, 6.0536e-09], |
| [ 4.7535e-05, 9.3132e-10, 8.0317e-06, ..., 7.7635e-06, |
| 2.5565e-07, 5.1223e-09]], device='cuda:0') |
| Epoch 106, bias, value: tensor([-4.4666e-03, 2.7202e-02, 4.9677e-03, -1.8957e-03, 9.6344e-03, |
| -4.6174e-04, 2.1111e-02, -1.8173e-02, 1.6226e-02, -7.8943e-05], |
| device='cuda:0'), grad: tensor([-8.6927e-04, 3.1650e-05, -3.9786e-06, 3.0935e-05, 3.7134e-05, |
| 7.3686e-06, 1.0520e-05, -2.5725e-04, 1.6838e-05, 9.9659e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 105, time 219.41, cls_loss 0.0056 cls_loss_mapping 0.0066 cls_loss_causal 0.5660 re_mapping 0.0099 re_causal 0.0279 /// teacc 98.81 lr 0.00010000 |
| Epoch 107, weight, value: tensor([[-0.0380, -0.0338, -0.0338, ..., -0.0819, 0.0038, -0.0078], |
| [-0.0297, 0.0233, -0.1064, ..., 0.0406, 0.0433, -0.0856], |
| [ 0.0496, -0.0015, -0.0515, ..., -0.0367, -0.0233, 0.0018], |
| ..., |
| [ 0.0392, -0.0427, 0.0006, ..., 0.0247, -0.0144, 0.0343], |
| [-0.0206, -0.0329, -0.0432, ..., -0.0100, -0.0165, -0.0344], |
| [-0.0791, -0.1209, 0.0382, ..., -0.0500, 0.0121, 0.0244]], |
| device='cuda:0'), grad: tensor([[ 7.0035e-07, 2.7474e-08, 4.5775e-07, ..., 0.0000e+00, |
| 2.5127e-06, 2.9802e-08], |
| [ 3.5707e-06, 2.2817e-08, 4.5868e-07, ..., 0.0000e+00, |
| -2.4401e-06, 8.5216e-08], |
| [-3.9548e-05, -5.2452e-06, 9.3598e-08, ..., 0.0000e+00, |
| 2.2724e-07, -1.7229e-07], |
| ..., |
| [ 1.9401e-05, 5.1335e-06, 2.0443e-07, ..., 0.0000e+00, |
| 1.3933e-06, -2.0443e-07], |
| [ 3.3937e-06, 3.6322e-08, 2.0474e-05, ..., 0.0000e+00, |
| 9.7603e-06, 5.7742e-08], |
| [ 7.6666e-06, 1.3970e-09, 5.3365e-07, ..., 0.0000e+00, |
| -1.6429e-06, 6.2864e-08]], device='cuda:0') |
| Epoch 107, bias, value: tensor([-0.0046, 0.0273, 0.0049, -0.0016, 0.0094, -0.0007, 0.0190, -0.0179, |
| 0.0183, -0.0001], device='cuda:0'), grad: tensor([-1.5404e-06, 9.8720e-07, -5.4210e-05, 2.7239e-05, 1.2495e-05, |
| -1.6248e-04, 5.5850e-05, 2.2382e-05, 9.7394e-05, 1.6484e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 106, time 219.79, cls_loss 0.0052 cls_loss_mapping 0.0092 cls_loss_causal 0.5927 re_mapping 0.0095 re_causal 0.0290 /// teacc 98.65 lr 0.00010000 |
| Epoch 108, weight, value: tensor([[-0.0382, -0.0345, -0.0339, ..., -0.0820, 0.0035, -0.0079], |
| [-0.0309, 0.0235, -0.1067, ..., 0.0407, 0.0445, -0.0868], |
| [ 0.0506, -0.0014, -0.0512, ..., -0.0368, -0.0235, 0.0022], |
| ..., |
| [ 0.0392, -0.0429, 0.0004, ..., 0.0250, -0.0150, 0.0342], |
| [-0.0208, -0.0336, -0.0445, ..., -0.0100, -0.0169, -0.0345], |
| [-0.0795, -0.1217, 0.0382, ..., -0.0508, 0.0120, 0.0244]], |
| device='cuda:0'), grad: tensor([[ 1.3961e-06, 0.0000e+00, 4.7497e-08, ..., 0.0000e+00, |
| 2.9981e-05, 1.3292e-05], |
| [ 7.4413e-07, 0.0000e+00, 6.5193e-09, ..., 0.0000e+00, |
| -6.9849e-07, 6.2399e-08], |
| [ 4.8578e-06, 0.0000e+00, 7.4506e-09, ..., 0.0000e+00, |
| 4.6268e-06, 1.6568e-06], |
| ..., |
| [ 1.1353e-06, 0.0000e+00, 3.5390e-08, ..., 0.0000e+00, |
| 1.3700e-06, 9.3132e-09], |
| [ 2.0396e-06, 0.0000e+00, 3.5297e-07, ..., 0.0000e+00, |
| 2.5164e-06, 2.6729e-07], |
| [ 2.6077e-06, 0.0000e+00, 3.6322e-08, ..., 0.0000e+00, |
| -1.6456e-06, 3.5390e-08]], device='cuda:0') |
| Epoch 108, bias, value: tensor([-0.0045, 0.0273, 0.0062, -0.0019, 0.0109, -0.0006, 0.0191, -0.0187, |
| 0.0177, -0.0013], device='cuda:0'), grad: tensor([ 1.6916e-04, 1.0170e-06, 3.1114e-05, -2.7850e-05, 8.5160e-06, |
| 2.6435e-05, -2.2793e-04, 5.0329e-06, 1.3761e-05, 8.7079e-07], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 107, time 219.86, cls_loss 0.0052 cls_loss_mapping 0.0074 cls_loss_causal 0.5872 re_mapping 0.0096 re_causal 0.0278 /// teacc 98.86 lr 0.00010000 |
| Epoch 109, weight, value: tensor([[-3.8403e-02, -3.5559e-02, -3.4272e-02, ..., -8.2115e-02, |
| 2.9623e-03, -8.0154e-03], |
| [-3.1084e-02, 2.2822e-02, -1.0705e-01, ..., 3.8417e-02, |
| 4.4395e-02, -8.6931e-02], |
| [ 5.0280e-02, -1.7909e-03, -5.1552e-02, ..., -3.6777e-02, |
| -2.4747e-02, 2.1256e-03], |
| ..., |
| [ 3.9916e-02, -4.1949e-02, 1.0697e-04, ..., 2.5264e-02, |
| -1.5121e-02, 3.4228e-02], |
| [-2.1378e-02, -3.5603e-02, -4.5783e-02, ..., -1.0111e-02, |
| -1.7135e-02, -3.4575e-02], |
| [-7.9854e-02, -1.2301e-01, 3.7948e-02, ..., -4.8472e-02, |
| 1.2831e-02, 2.4600e-02]], device='cuda:0'), grad: tensor([[ 5.8301e-07, -2.1532e-06, 7.0594e-07, ..., 0.0000e+00, |
| 1.6959e-06, 0.0000e+00], |
| [ 1.4745e-05, 6.0536e-08, 1.3905e-06, ..., 0.0000e+00, |
| 6.9849e-07, 0.0000e+00], |
| [-2.2218e-05, 8.7544e-08, -9.7789e-08, ..., 0.0000e+00, |
| -6.2399e-07, 0.0000e+00], |
| ..., |
| [-2.6971e-06, 3.1292e-07, 2.0228e-06, ..., 0.0000e+00, |
| 6.8918e-07, 0.0000e+00], |
| [ 2.8759e-06, 1.5274e-07, 6.7167e-06, ..., 0.0000e+00, |
| 2.2873e-06, 0.0000e+00], |
| [ 2.0433e-06, 2.3656e-07, -1.3580e-03, ..., 0.0000e+00, |
| -3.6788e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 109, bias, value: tensor([-0.0042, 0.0271, 0.0052, -0.0021, 0.0114, -0.0006, 0.0191, -0.0176, |
| 0.0169, -0.0011], device='cuda:0'), grad: tensor([-1.9848e-05, 3.2067e-05, -3.6597e-05, 2.2620e-05, 3.3932e-03, |
| 2.1175e-05, -4.8071e-05, 1.7164e-06, 3.0056e-05, -3.3989e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 108, time 219.60, cls_loss 0.0059 cls_loss_mapping 0.0100 cls_loss_causal 0.5958 re_mapping 0.0092 re_causal 0.0278 /// teacc 98.78 lr 0.00010000 |
| Epoch 110, weight, value: tensor([[-0.0397, -0.0375, -0.0346, ..., -0.0836, 0.0025, -0.0081], |
| [-0.0318, 0.0207, -0.1075, ..., 0.0384, 0.0443, -0.0871], |
| [ 0.0500, -0.0008, -0.0536, ..., -0.0373, -0.0245, 0.0021], |
| ..., |
| [ 0.0403, -0.0417, -0.0009, ..., 0.0246, -0.0151, 0.0343], |
| [-0.0218, -0.0337, -0.0461, ..., -0.0106, -0.0170, -0.0346], |
| [-0.0807, -0.1270, 0.0384, ..., -0.0486, 0.0131, 0.0246]], |
| device='cuda:0'), grad: tensor([[ 7.8231e-08, 2.0489e-08, 3.4254e-06, ..., 4.1258e-07, |
| 2.0452e-06, 0.0000e+00], |
| [ 2.3656e-07, 1.0245e-08, 7.1153e-07, ..., 2.0489e-08, |
| -3.0994e-06, 0.0000e+00], |
| [-8.2608e-07, -7.0874e-07, 5.8208e-07, ..., 7.4506e-09, |
| 3.9581e-07, 0.0000e+00], |
| ..., |
| [-7.5437e-08, 2.5518e-07, 6.0722e-06, ..., 5.5879e-08, |
| 1.3607e-06, 0.0000e+00], |
| [ 7.8231e-07, 1.6112e-07, 1.4920e-06, ..., 9.3132e-08, |
| 1.9316e-06, 0.0000e+00], |
| [ 7.1526e-07, 1.2107e-08, 1.4193e-05, ..., 1.0617e-07, |
| 4.4703e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 110, bias, value: tensor([-0.0047, 0.0257, 0.0052, 0.0003, 0.0111, -0.0010, 0.0188, -0.0171, |
| 0.0169, -0.0009], device='cuda:0'), grad: tensor([-2.3380e-05, -2.4252e-06, 3.3677e-06, 3.6340e-06, -8.4877e-05, |
| -2.2605e-05, 3.2932e-06, 4.2081e-05, 1.0341e-05, 7.0393e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 109, time 219.83, cls_loss 0.0051 cls_loss_mapping 0.0073 cls_loss_causal 0.6049 re_mapping 0.0095 re_causal 0.0287 /// teacc 98.85 lr 0.00010000 |
| Epoch 111, weight, value: tensor([[-0.0401, -0.0383, -0.0351, ..., -0.0838, 0.0019, -0.0081], |
| [-0.0320, 0.0204, -0.1081, ..., 0.0385, 0.0450, -0.0875], |
| [ 0.0501, -0.0009, -0.0542, ..., -0.0374, -0.0250, 0.0025], |
| ..., |
| [ 0.0405, -0.0414, -0.0002, ..., 0.0248, -0.0157, 0.0342], |
| [-0.0220, -0.0337, -0.0465, ..., -0.0106, -0.0177, -0.0348], |
| [-0.0817, -0.1276, 0.0384, ..., -0.0487, 0.0133, 0.0247]], |
| device='cuda:0'), grad: tensor([[ 2.1830e-06, 1.3504e-07, 3.1628e-06, ..., 0.0000e+00, |
| 7.7300e-07, 4.3772e-07], |
| [ 1.8273e-06, 5.7742e-08, 7.6368e-07, ..., 0.0000e+00, |
| -2.2873e-06, 1.3411e-07], |
| [-1.3113e-06, -1.6158e-06, 5.5656e-06, ..., 0.0000e+00, |
| 2.1569e-06, -8.2795e-07], |
| ..., |
| [ 1.4260e-05, 1.2852e-07, 1.4743e-06, ..., 0.0000e+00, |
| 7.3388e-06, 2.8033e-07], |
| [ 4.4182e-06, 2.0210e-07, 4.3809e-06, ..., 0.0000e+00, |
| 2.0806e-06, 9.9465e-07], |
| [ 5.1379e-05, 1.5832e-08, 5.2825e-06, ..., 0.0000e+00, |
| 1.9237e-05, 3.3900e-07]], device='cuda:0') |
| Epoch 111, bias, value: tensor([-5.3301e-03, 2.5805e-02, 4.9322e-03, 3.8411e-05, 1.1153e-02, |
| -4.7907e-04, 1.9546e-02, -1.6959e-02, 1.6641e-02, -1.1235e-03], |
| device='cuda:0'), grad: tensor([-5.1051e-05, -3.2093e-06, 2.5973e-05, -1.5163e-04, -3.4308e-04, |
| -1.0085e-04, 4.3035e-04, 5.0932e-05, 3.2067e-05, 1.0979e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 110, time 220.27, cls_loss 0.0051 cls_loss_mapping 0.0084 cls_loss_causal 0.5764 re_mapping 0.0097 re_causal 0.0273 /// teacc 98.81 lr 0.00010000 |
| Epoch 112, weight, value: tensor([[-0.0405, -0.0396, -0.0347, ..., -0.0838, 0.0017, -0.0086], |
| [-0.0311, 0.0213, -0.1087, ..., 0.0385, 0.0462, -0.0880], |
| [ 0.0502, -0.0020, -0.0536, ..., -0.0374, -0.0254, 0.0026], |
| ..., |
| [ 0.0405, -0.0406, 0.0002, ..., 0.0251, -0.0170, 0.0342], |
| [-0.0222, -0.0338, -0.0465, ..., -0.0106, -0.0174, -0.0356], |
| [-0.0823, -0.1288, 0.0387, ..., -0.0487, 0.0137, 0.0247]], |
| device='cuda:0'), grad: tensor([[ 2.3246e-06, 8.3819e-09, 1.0341e-05, ..., 0.0000e+00, |
| 2.5574e-06, 0.0000e+00], |
| [ 8.8010e-07, 1.8626e-09, 3.6843e-06, ..., 0.0000e+00, |
| -2.1979e-07, 0.0000e+00], |
| [-7.9256e-07, 9.3132e-10, 2.4289e-06, ..., 0.0000e+00, |
| 1.6838e-06, 0.0000e+00], |
| ..., |
| [-1.6140e-06, 1.8626e-09, 1.4961e-05, ..., 0.0000e+00, |
| 6.4820e-07, 0.0000e+00], |
| [-1.1418e-06, 5.2154e-08, 4.6864e-06, ..., 0.0000e+00, |
| -1.4819e-05, 0.0000e+00], |
| [ 1.9986e-06, 8.3819e-09, 1.9264e-04, ..., 0.0000e+00, |
| -5.7183e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 112, bias, value: tensor([-0.0049, 0.0272, 0.0048, -0.0007, 0.0104, -0.0015, 0.0205, -0.0174, |
| 0.0169, -0.0010], device='cuda:0'), grad: tensor([ 3.0816e-05, 2.3514e-05, 8.3089e-05, 2.1005e-04, -1.1311e-03, |
| -1.1855e-04, 1.1645e-05, 1.3542e-04, -3.0446e-04, 1.0605e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 111, time 219.73, cls_loss 0.0060 cls_loss_mapping 0.0070 cls_loss_causal 0.6029 re_mapping 0.0092 re_causal 0.0268 /// teacc 98.88 lr 0.00010000 |
| Epoch 113, weight, value: tensor([[-0.0410, -0.0416, -0.0349, ..., -0.0840, 0.0015, -0.0088], |
| [-0.0308, 0.0216, -0.1101, ..., 0.0388, 0.0468, -0.0901], |
| [ 0.0507, -0.0021, -0.0546, ..., -0.0374, -0.0257, 0.0020], |
| ..., |
| [ 0.0405, -0.0407, 0.0014, ..., 0.0249, -0.0180, 0.0341], |
| [-0.0225, -0.0344, -0.0476, ..., -0.0107, -0.0170, -0.0371], |
| [-0.0834, -0.1302, 0.0389, ..., -0.0488, 0.0139, 0.0247]], |
| device='cuda:0'), grad: tensor([[ 1.9558e-07, 0.0000e+00, 4.5821e-06, ..., 0.0000e+00, |
| 1.8282e-06, 0.0000e+00], |
| [ 2.3190e-07, 0.0000e+00, -1.5914e-05, ..., 1.8626e-09, |
| -3.9071e-05, 0.0000e+00], |
| [-3.0361e-06, 0.0000e+00, 1.4994e-07, ..., 9.3132e-10, |
| 1.5991e-06, 0.0000e+00], |
| ..., |
| [ 9.9279e-07, 0.0000e+00, 3.0920e-06, ..., 8.3819e-09, |
| 5.6997e-06, 0.0000e+00], |
| [-5.7146e-06, 0.0000e+00, 5.6982e-05, ..., 9.3132e-10, |
| 2.2531e-05, 0.0000e+00], |
| [ 3.7812e-07, 0.0000e+00, -4.5347e-04, ..., 1.1176e-08, |
| -2.0897e-04, 0.0000e+00]], device='cuda:0') |
| Epoch 113, bias, value: tensor([-0.0051, 0.0274, 0.0048, -0.0015, 0.0101, -0.0007, 0.0202, -0.0174, |
| 0.0173, -0.0010], device='cuda:0'), grad: tensor([ 1.4007e-06, -1.2732e-04, 1.7891e-06, 7.3624e-04, 1.2189e-04, |
| 3.4404e-04, 4.4614e-05, 2.5123e-05, 9.4354e-05, -1.2417e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 112---------------------------------------------------- |
| epoch 112, time 220.48, cls_loss 0.0040 cls_loss_mapping 0.0059 cls_loss_causal 0.5629 re_mapping 0.0095 re_causal 0.0286 /// teacc 98.89 lr 0.00010000 |
| Epoch 114, weight, value: tensor([[-0.0416, -0.0419, -0.0361, ..., -0.0842, 0.0012, -0.0088], |
| [-0.0308, 0.0217, -0.1108, ..., 0.0389, 0.0475, -0.0909], |
| [ 0.0509, -0.0021, -0.0534, ..., -0.0377, -0.0263, 0.0035], |
| ..., |
| [ 0.0407, -0.0408, 0.0013, ..., 0.0249, -0.0181, 0.0340], |
| [-0.0225, -0.0343, -0.0476, ..., -0.0108, -0.0173, -0.0379], |
| [-0.0839, -0.1303, 0.0393, ..., -0.0489, 0.0144, 0.0246]], |
| device='cuda:0'), grad: tensor([[ 3.4459e-07, 4.6566e-09, 3.3993e-07, ..., 0.0000e+00, |
| 4.8801e-07, 6.3330e-08], |
| [ 4.1574e-06, 9.3132e-09, 2.7064e-06, ..., 0.0000e+00, |
| -8.3968e-06, 2.4214e-08], |
| [-1.7732e-05, -1.5739e-07, 1.5255e-06, ..., 0.0000e+00, |
| 9.2015e-07, 3.5390e-08], |
| ..., |
| [-3.6098e-06, 1.5832e-08, 1.9614e-06, ..., 0.0000e+00, |
| 1.7537e-06, 9.3132e-10], |
| [ 7.2606e-06, 7.0781e-08, 1.9046e-06, ..., 0.0000e+00, |
| 4.4405e-06, 2.9802e-08], |
| [ 1.5832e-06, 1.8626e-09, -8.3260e-07, ..., 0.0000e+00, |
| -5.6550e-06, 1.8626e-09]], device='cuda:0') |
| Epoch 114, bias, value: tensor([-0.0057, 0.0279, 0.0047, -0.0011, 0.0098, -0.0015, 0.0200, -0.0174, |
| 0.0176, -0.0008], device='cuda:0'), grad: tensor([ 2.4587e-06, -5.3048e-06, -2.0295e-05, 8.5086e-06, -1.9427e-06, |
| 9.2015e-07, 2.9840e-06, 1.7984e-06, 2.1785e-05, -1.0937e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 113, time 219.65, cls_loss 0.0048 cls_loss_mapping 0.0087 cls_loss_causal 0.5868 re_mapping 0.0096 re_causal 0.0280 /// teacc 98.83 lr 0.00010000 |
| Epoch 115, weight, value: tensor([[-0.0419, -0.0422, -0.0371, ..., -0.0843, 0.0009, -0.0088], |
| [-0.0309, 0.0223, -0.1121, ..., 0.0390, 0.0478, -0.0915], |
| [ 0.0512, -0.0020, -0.0527, ..., -0.0381, -0.0266, 0.0045], |
| ..., |
| [ 0.0407, -0.0411, 0.0011, ..., 0.0253, -0.0188, 0.0340], |
| [-0.0232, -0.0344, -0.0470, ..., -0.0108, -0.0183, -0.0381], |
| [-0.0845, -0.1310, 0.0389, ..., -0.0491, 0.0151, 0.0246]], |
| device='cuda:0'), grad: tensor([[ 4.8801e-07, 0.0000e+00, 5.6550e-06, ..., 4.0978e-08, |
| 4.6287e-07, 2.7940e-09], |
| [ 8.4117e-06, 0.0000e+00, 7.6666e-06, ..., 2.4214e-08, |
| 3.3155e-07, 9.3132e-10], |
| [ 1.0967e-05, 0.0000e+00, 8.8885e-06, ..., 3.7253e-09, |
| 8.9686e-07, -5.4017e-08], |
| ..., |
| [-1.7602e-06, 0.0000e+00, 7.7710e-06, ..., -4.6566e-08, |
| 7.3053e-06, 7.4506e-09], |
| [ 2.0191e-05, 0.0000e+00, 9.5963e-06, ..., 8.3819e-08, |
| -6.0461e-06, 2.4214e-08], |
| [ 8.5868e-07, 0.0000e+00, 1.5557e-05, ..., 4.6566e-08, |
| -5.7220e-06, 9.3132e-10]], device='cuda:0') |
| Epoch 115, bias, value: tensor([-0.0051, 0.0280, 0.0047, -0.0013, 0.0102, -0.0013, 0.0198, -0.0177, |
| 0.0173, -0.0006], device='cuda:0'), grad: tensor([-1.1814e-04, 4.5747e-05, 5.8800e-05, -5.1022e-05, -3.1257e-04, |
| -6.2406e-05, 2.9612e-04, 1.3113e-04, 5.0783e-05, -3.8475e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 114---------------------------------------------------- |
| epoch 114, time 220.54, cls_loss 0.0040 cls_loss_mapping 0.0069 cls_loss_causal 0.5432 re_mapping 0.0096 re_causal 0.0265 /// teacc 98.92 lr 0.00010000 |
| Epoch 116, weight, value: tensor([[-0.0423, -0.0426, -0.0373, ..., -0.0843, 0.0007, -0.0092], |
| [-0.0312, 0.0222, -0.1125, ..., 0.0390, 0.0470, -0.0932], |
| [ 0.0517, -0.0015, -0.0525, ..., -0.0384, -0.0267, 0.0045], |
| ..., |
| [ 0.0407, -0.0413, 0.0012, ..., 0.0253, -0.0177, 0.0339], |
| [-0.0237, -0.0342, -0.0470, ..., -0.0109, -0.0186, -0.0386], |
| [-0.0853, -0.1316, 0.0386, ..., -0.0491, 0.0151, 0.0247]], |
| device='cuda:0'), grad: tensor([[ 4.0606e-07, 1.8626e-09, 3.8370e-07, ..., 9.3132e-10, |
| 3.8557e-07, 0.0000e+00], |
| [ 9.6038e-06, 1.6764e-08, 2.0832e-05, ..., 1.8626e-09, |
| 1.8403e-05, 0.0000e+00], |
| [ 3.4273e-06, -8.5682e-08, 5.6159e-07, ..., 4.6566e-09, |
| 1.6233e-06, 0.0000e+00], |
| ..., |
| [-7.3835e-06, 1.3970e-08, -9.6619e-05, ..., 1.6764e-08, |
| -6.5923e-05, 0.0000e+00], |
| [-9.4891e-05, 3.8184e-08, -4.2260e-05, ..., 1.8626e-09, |
| -8.6606e-05, 0.0000e+00], |
| [ 2.0843e-06, 1.8626e-09, 7.0691e-05, ..., 2.7940e-09, |
| 4.8637e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 116, bias, value: tensor([-0.0052, 0.0264, 0.0050, -0.0018, 0.0099, -0.0006, 0.0199, -0.0164, |
| 0.0176, -0.0011], device='cuda:0'), grad: tensor([ 1.8645e-06, 1.5104e-04, 1.2010e-05, 2.0695e-04, 3.3528e-05, |
| 5.3078e-05, 1.7393e-04, -5.8126e-04, -4.6754e-04, 4.1628e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 115, time 219.85, cls_loss 0.0056 cls_loss_mapping 0.0099 cls_loss_causal 0.5739 re_mapping 0.0091 re_causal 0.0261 /// teacc 98.77 lr 0.00010000 |
| Epoch 117, weight, value: tensor([[-0.0407, -0.0432, -0.0374, ..., -0.0845, 0.0027, -0.0093], |
| [-0.0325, 0.0221, -0.1119, ..., 0.0402, 0.0472, -0.0946], |
| [ 0.0514, -0.0006, -0.0528, ..., -0.0384, -0.0271, 0.0048], |
| ..., |
| [ 0.0411, -0.0412, 0.0010, ..., 0.0249, -0.0189, 0.0341], |
| [-0.0247, -0.0362, -0.0474, ..., -0.0110, -0.0186, -0.0394], |
| [-0.0860, -0.1327, 0.0384, ..., -0.0494, 0.0161, 0.0249]], |
| device='cuda:0'), grad: tensor([[ 1.3076e-06, 1.2107e-08, 1.0677e-05, ..., 0.0000e+00, |
| 1.1912e-06, 3.4180e-07], |
| [ 1.3467e-06, 1.0245e-08, 2.3469e-07, ..., 0.0000e+00, |
| -6.3404e-06, 6.9849e-08], |
| [-7.3854e-07, -1.8347e-07, 3.0026e-06, ..., 0.0000e+00, |
| 1.0692e-06, -1.3132e-07], |
| ..., |
| [-3.8855e-06, 2.4214e-08, 5.8487e-07, ..., 0.0000e+00, |
| 1.8552e-06, 5.3085e-08], |
| [ 1.3523e-06, 4.4703e-08, 1.0077e-06, ..., 0.0000e+00, |
| 5.0105e-06, 2.4587e-07], |
| [ 1.0654e-05, 2.7940e-09, -2.0608e-05, ..., 0.0000e+00, |
| 1.6108e-05, 1.3970e-08]], device='cuda:0') |
| Epoch 117, bias, value: tensor([-0.0032, 0.0256, 0.0052, -0.0014, 0.0094, -0.0002, 0.0200, -0.0167, |
| 0.0170, -0.0013], device='cuda:0'), grad: tensor([ 4.9114e-05, -9.1717e-06, 1.4983e-05, -4.3780e-05, 1.0595e-05, |
| 1.4469e-05, -6.2212e-06, -1.5805e-06, 1.5303e-05, -4.3660e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 116, time 219.65, cls_loss 0.0058 cls_loss_mapping 0.0087 cls_loss_causal 0.5783 re_mapping 0.0091 re_causal 0.0268 /// teacc 98.89 lr 0.00010000 |
| Epoch 118, weight, value: tensor([[-0.0412, -0.0444, -0.0362, ..., -0.0850, 0.0034, -0.0121], |
| [-0.0339, 0.0230, -0.1138, ..., 0.0404, 0.0474, -0.0954], |
| [ 0.0513, -0.0005, -0.0519, ..., -0.0386, -0.0276, 0.0050], |
| ..., |
| [ 0.0422, -0.0417, 0.0007, ..., 0.0250, -0.0193, 0.0343], |
| [-0.0249, -0.0364, -0.0479, ..., -0.0112, -0.0190, -0.0396], |
| [-0.0869, -0.1341, 0.0369, ..., -0.0498, 0.0159, 0.0277]], |
| device='cuda:0'), grad: tensor([[ 9.4399e-06, 0.0000e+00, 2.0802e-05, ..., 0.0000e+00, |
| 6.4261e-08, 0.0000e+00], |
| [ 1.6140e-06, 0.0000e+00, 2.6338e-06, ..., 0.0000e+00, |
| -2.1979e-07, 0.0000e+00], |
| [ 4.3437e-06, 0.0000e+00, 7.7337e-06, ..., 0.0000e+00, |
| 2.7101e-07, -1.8626e-09], |
| ..., |
| [ 8.1304e-07, 0.0000e+00, 9.3505e-06, ..., 0.0000e+00, |
| 2.5332e-07, 0.0000e+00], |
| [ 1.5739e-06, 0.0000e+00, 1.3579e-06, ..., 0.0000e+00, |
| 7.8231e-08, 0.0000e+00], |
| [ 5.5805e-06, 0.0000e+00, 1.0952e-05, ..., 0.0000e+00, |
| -9.4995e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 118, bias, value: tensor([-0.0037, 0.0250, 0.0049, -0.0013, 0.0106, -0.0005, 0.0202, -0.0159, |
| 0.0166, -0.0014], device='cuda:0'), grad: tensor([ 4.8786e-05, 8.2701e-06, 2.3931e-05, 1.1943e-05, -1.4257e-04, |
| -1.7434e-05, 3.3349e-05, 6.0797e-06, -1.7202e-06, 2.9370e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 117, time 219.61, cls_loss 0.0035 cls_loss_mapping 0.0055 cls_loss_causal 0.5805 re_mapping 0.0087 re_causal 0.0272 /// teacc 98.87 lr 0.00010000 |
| Epoch 119, weight, value: tensor([[-0.0414, -0.0448, -0.0364, ..., -0.0850, 0.0033, -0.0122], |
| [-0.0341, 0.0231, -0.1143, ..., 0.0404, 0.0475, -0.0961], |
| [ 0.0523, -0.0004, -0.0503, ..., -0.0386, -0.0279, 0.0050], |
| ..., |
| [ 0.0422, -0.0416, 0.0007, ..., 0.0251, -0.0198, 0.0347], |
| [-0.0262, -0.0368, -0.0491, ..., -0.0112, -0.0194, -0.0397], |
| [-0.0871, -0.1347, 0.0367, ..., -0.0498, 0.0169, 0.0277]], |
| device='cuda:0'), grad: tensor([[ 6.3144e-07, 0.0000e+00, 5.0198e-07, ..., 0.0000e+00, |
| 2.3749e-07, 4.6566e-09], |
| [ 5.6066e-06, 0.0000e+00, 9.0525e-07, ..., 0.0000e+00, |
| -4.2841e-08, 4.6566e-09], |
| [-4.0114e-05, 0.0000e+00, 1.9064e-06, ..., 0.0000e+00, |
| 7.3947e-07, 2.7940e-09], |
| ..., |
| [ 4.6939e-05, 0.0000e+00, 1.7196e-05, ..., 0.0000e+00, |
| 2.2911e-07, 2.2352e-08], |
| [ 7.9349e-06, 0.0000e+00, 1.1995e-06, ..., 0.0000e+00, |
| 9.8720e-07, 2.7940e-09], |
| [ 4.9993e-06, 0.0000e+00, 6.9998e-06, ..., 0.0000e+00, |
| -1.8999e-07, 7.4506e-09]], device='cuda:0') |
| Epoch 119, bias, value: tensor([-0.0032, 0.0248, 0.0056, -0.0015, 0.0106, -0.0004, 0.0204, -0.0163, |
| 0.0156, -0.0013], device='cuda:0'), grad: tensor([ 2.1793e-06, 1.2830e-05, -3.2961e-05, -3.2187e-05, -9.1553e-05, |
| 3.6769e-06, 7.6108e-06, 8.7440e-05, 1.3083e-05, 2.9802e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 118, time 219.85, cls_loss 0.0037 cls_loss_mapping 0.0055 cls_loss_causal 0.5427 re_mapping 0.0089 re_causal 0.0256 /// teacc 98.83 lr 0.00010000 |
| Epoch 120, weight, value: tensor([[-0.0416, -0.0450, -0.0365, ..., -0.0850, 0.0025, -0.0136], |
| [-0.0344, 0.0231, -0.1147, ..., 0.0404, 0.0476, -0.0969], |
| [ 0.0525, -0.0004, -0.0504, ..., -0.0387, -0.0277, 0.0049], |
| ..., |
| [ 0.0423, -0.0416, 0.0007, ..., 0.0251, -0.0199, 0.0344], |
| [-0.0266, -0.0368, -0.0503, ..., -0.0112, -0.0201, -0.0400], |
| [-0.0880, -0.1349, 0.0374, ..., -0.0498, 0.0172, 0.0292]], |
| device='cuda:0'), grad: tensor([[ 1.9968e-06, 0.0000e+00, 2.7493e-06, ..., 6.7707e-07, |
| 1.9558e-06, 1.8626e-09], |
| [-2.8300e-04, 0.0000e+00, 8.0094e-07, ..., -1.1557e-04, |
| -8.0764e-05, 9.3132e-10], |
| [ 2.4331e-04, 0.0000e+00, 3.8818e-06, ..., 9.9182e-05, |
| 6.3241e-05, -2.5146e-08], |
| ..., |
| [ 2.2314e-06, 0.0000e+00, 8.8196e-07, ..., 8.5160e-06, |
| 5.8040e-06, 3.7253e-09], |
| [ 3.5837e-06, 0.0000e+00, 1.9558e-06, ..., 1.8626e-07, |
| 1.0006e-05, 1.8626e-09], |
| [ 1.0699e-05, 0.0000e+00, 2.6608e-04, ..., 1.7043e-07, |
| 2.2724e-07, 1.8626e-09]], device='cuda:0') |
| Epoch 120, bias, value: tensor([-0.0037, 0.0247, 0.0058, -0.0014, 0.0098, -0.0005, 0.0215, -0.0163, |
| 0.0147, -0.0005], device='cuda:0'), grad: tensor([-2.1830e-06, -8.9788e-04, 7.7772e-04, 2.5436e-05, -1.2054e-03, |
| 2.3901e-05, -9.4324e-06, 2.6017e-05, 4.8608e-05, 1.2121e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 119, time 219.49, cls_loss 0.0045 cls_loss_mapping 0.0060 cls_loss_causal 0.5151 re_mapping 0.0089 re_causal 0.0245 /// teacc 98.91 lr 0.00010000 |
| Epoch 121, weight, value: tensor([[-4.1891e-02, -4.5554e-02, -3.6993e-02, ..., -8.5104e-02, |
| 2.3789e-03, -1.3753e-02], |
| [-3.4267e-02, 2.3360e-02, -1.1493e-01, ..., 4.0970e-02, |
| 4.8083e-02, -9.7092e-02], |
| [ 5.2753e-02, -7.1994e-06, -5.0606e-02, ..., -3.9629e-02, |
| -2.8378e-02, 5.0982e-03], |
| ..., |
| [ 4.1737e-02, -4.2261e-02, 3.1517e-04, ..., 2.4773e-02, |
| -2.1074e-02, 3.4351e-02], |
| [-2.6481e-02, -3.6210e-02, -5.0717e-02, ..., -1.1237e-02, |
| -1.9154e-02, -4.0000e-02], |
| [-8.8695e-02, -1.3654e-01, 3.6906e-02, ..., -4.9851e-02, |
| 1.7697e-02, 2.9283e-02]], device='cuda:0'), grad: tensor([[ 7.0315e-07, 0.0000e+00, 1.1744e-06, ..., 0.0000e+00, |
| 2.7101e-06, 0.0000e+00], |
| [ 4.6380e-07, 0.0000e+00, 6.3423e-07, ..., 0.0000e+00, |
| -2.8815e-06, 0.0000e+00], |
| [-1.1571e-05, 0.0000e+00, -8.7917e-07, ..., 0.0000e+00, |
| 3.9861e-06, 0.0000e+00], |
| ..., |
| [-4.3958e-07, 0.0000e+00, 1.5646e-06, ..., 0.0000e+00, |
| 1.2740e-06, 0.0000e+00], |
| [ 2.5108e-06, 0.0000e+00, 5.3108e-05, ..., 0.0000e+00, |
| -5.8532e-05, 0.0000e+00], |
| [ 2.4866e-07, 0.0000e+00, -7.8738e-05, ..., 0.0000e+00, |
| 5.6505e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 121, bias, value: tensor([-0.0034, 0.0249, 0.0060, -0.0009, 0.0099, -0.0005, 0.0210, -0.0172, |
| 0.0154, -0.0007], device='cuda:0'), grad: tensor([ 4.8786e-05, -6.2473e-06, 3.3826e-05, 1.0192e-04, 3.7503e-04, |
| -2.4390e-04, 5.5581e-05, 1.4514e-05, 3.8177e-05, -4.1676e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 120, time 219.65, cls_loss 0.0042 cls_loss_mapping 0.0060 cls_loss_causal 0.5517 re_mapping 0.0085 re_causal 0.0251 /// teacc 98.84 lr 0.00010000 |
| Epoch 122, weight, value: tensor([[-0.0423, -0.0464, -0.0373, ..., -0.0852, 0.0020, -0.0138], |
| [-0.0338, 0.0234, -0.1153, ..., 0.0410, 0.0483, -0.0971], |
| [ 0.0527, 0.0008, -0.0507, ..., -0.0397, -0.0288, 0.0051], |
| ..., |
| [ 0.0415, -0.0431, 0.0005, ..., 0.0248, -0.0214, 0.0344], |
| [-0.0267, -0.0355, -0.0510, ..., -0.0113, -0.0192, -0.0401], |
| [-0.0895, -0.1379, 0.0366, ..., -0.0499, 0.0181, 0.0293]], |
| device='cuda:0'), grad: tensor([[ 3.2224e-07, 0.0000e+00, 2.0862e-07, ..., 0.0000e+00, |
| 3.1851e-07, 0.0000e+00], |
| [ 3.3900e-06, 0.0000e+00, 1.2843e-06, ..., 0.0000e+00, |
| 2.7381e-07, 0.0000e+00], |
| [ 5.3346e-06, 0.0000e+00, 1.3132e-07, ..., 0.0000e+00, |
| 7.0259e-06, 0.0000e+00], |
| ..., |
| [-1.0841e-05, 0.0000e+00, 4.3120e-07, ..., 0.0000e+00, |
| 3.8557e-07, 0.0000e+00], |
| [-1.0632e-05, 0.0000e+00, 7.9256e-07, ..., 0.0000e+00, |
| -2.5615e-05, 0.0000e+00], |
| [ 6.6869e-07, 0.0000e+00, 9.8526e-05, ..., 0.0000e+00, |
| 3.0287e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 122, bias, value: tensor([-0.0036, 0.0253, 0.0059, -0.0012, 0.0097, 0.0006, 0.0206, -0.0174, |
| 0.0154, -0.0008], device='cuda:0'), grad: tensor([-1.7909e-06, 1.2085e-05, 4.6521e-05, 8.1480e-05, -3.6263e-04, |
| 1.0304e-05, 9.7752e-06, -1.3247e-05, -1.5104e-04, 3.6764e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 121, time 218.27, cls_loss 0.0038 cls_loss_mapping 0.0058 cls_loss_causal 0.5796 re_mapping 0.0081 re_causal 0.0251 /// teacc 98.73 lr 0.00010000 |
| Epoch 123, weight, value: tensor([[-0.0426, -0.0470, -0.0376, ..., -0.0853, 0.0020, -0.0138], |
| [-0.0337, 0.0231, -0.1159, ..., 0.0410, 0.0484, -0.0973], |
| [ 0.0524, 0.0007, -0.0509, ..., -0.0398, -0.0290, 0.0054], |
| ..., |
| [ 0.0417, -0.0429, 0.0005, ..., 0.0249, -0.0216, 0.0344], |
| [-0.0266, -0.0341, -0.0521, ..., -0.0113, -0.0193, -0.0402], |
| [-0.0903, -0.1387, 0.0365, ..., -0.0499, 0.0185, 0.0293]], |
| device='cuda:0'), grad: tensor([[ 6.6124e-08, 0.0000e+00, 1.0338e-07, ..., 0.0000e+00, |
| 1.7602e-07, 6.8918e-08], |
| [ 7.8324e-07, 0.0000e+00, 1.8068e-07, ..., 2.7940e-09, |
| -6.5845e-07, 1.7695e-08], |
| [-1.5525e-06, 0.0000e+00, 1.4622e-07, ..., 9.3132e-10, |
| 2.9244e-07, -4.8429e-08], |
| ..., |
| [-4.4703e-07, 0.0000e+00, 6.1560e-07, ..., 8.3819e-09, |
| 1.0319e-06, 3.6322e-08], |
| [ 2.1048e-07, 0.0000e+00, 1.1146e-05, ..., 0.0000e+00, |
| 5.3924e-07, 4.0978e-08], |
| [ 7.4599e-07, 0.0000e+00, -9.6336e-06, ..., 9.3132e-10, |
| -1.3925e-05, 1.7136e-07]], device='cuda:0') |
| Epoch 123, bias, value: tensor([-0.0035, 0.0253, 0.0053, -0.0015, 0.0101, 0.0009, 0.0206, -0.0171, |
| 0.0153, -0.0009], device='cuda:0'), grad: tensor([-1.8924e-05, 2.5779e-06, 1.6484e-06, 3.5524e-05, 8.7440e-05, |
| -2.3335e-05, 4.7311e-06, 5.7630e-06, 2.2635e-05, -1.1802e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 122, time 217.89, cls_loss 0.0045 cls_loss_mapping 0.0068 cls_loss_causal 0.5693 re_mapping 0.0082 re_causal 0.0244 /// teacc 98.82 lr 0.00010000 |
| Epoch 124, weight, value: tensor([[-0.0426, -0.0489, -0.0372, ..., -0.0853, 0.0019, -0.0138], |
| [-0.0349, 0.0233, -0.1164, ..., 0.0410, 0.0484, -0.0974], |
| [ 0.0517, 0.0010, -0.0507, ..., -0.0398, -0.0296, 0.0055], |
| ..., |
| [ 0.0434, -0.0433, 0.0004, ..., 0.0249, -0.0216, 0.0345], |
| [-0.0270, -0.0343, -0.0527, ..., -0.0113, -0.0191, -0.0402], |
| [-0.0913, -0.1407, 0.0360, ..., -0.0499, 0.0186, 0.0294]], |
| device='cuda:0'), grad: tensor([[ 6.2678e-07, 5.0291e-08, 3.0190e-05, ..., 0.0000e+00, |
| 4.0555e-04, 1.0395e-03], |
| [ 2.1532e-06, 6.4448e-07, 3.4459e-07, ..., 0.0000e+00, |
| -6.1281e-06, 1.1707e-06], |
| [ 6.0797e-06, 2.2855e-06, 4.6194e-07, ..., 0.0000e+00, |
| 7.1488e-06, 1.3039e-05], |
| ..., |
| [-1.4164e-05, -4.1462e-06, 1.9465e-07, ..., 0.0000e+00, |
| 4.1015e-06, 6.4299e-06], |
| [ 1.7900e-06, 5.0571e-07, 9.4436e-07, ..., 0.0000e+00, |
| 3.4794e-06, 4.2096e-07], |
| [ 2.3730e-06, 2.8871e-08, -3.0905e-05, ..., 0.0000e+00, |
| -4.2057e-04, -1.0691e-03]], device='cuda:0') |
| Epoch 124, bias, value: tensor([-0.0031, 0.0245, 0.0046, -0.0014, 0.0098, 0.0005, 0.0216, -0.0160, |
| 0.0155, -0.0015], device='cuda:0'), grad: tensor([ 3.3512e-03, -6.5193e-06, 7.2837e-05, 1.5944e-05, 2.4661e-05, |
| 9.9763e-06, -6.9775e-06, -3.8624e-05, 2.0161e-05, -3.4447e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 123, time 217.94, cls_loss 0.0048 cls_loss_mapping 0.0090 cls_loss_causal 0.5585 re_mapping 0.0081 re_causal 0.0244 /// teacc 98.73 lr 0.00010000 |
| Epoch 125, weight, value: tensor([[-4.2627e-02, -5.0596e-02, -3.7549e-02, ..., -8.5329e-02, |
| 1.2055e-03, -1.4830e-02], |
| [-3.6399e-02, 2.5624e-02, -1.1669e-01, ..., 4.1077e-02, |
| 4.8046e-02, -9.7861e-02], |
| [ 5.3142e-02, 5.3213e-05, -5.0821e-02, ..., -3.9793e-02, |
| -2.7590e-02, 5.0983e-03], |
| ..., |
| [ 4.3337e-02, -4.3581e-02, 1.4466e-03, ..., 2.4874e-02, |
| -2.3025e-02, 3.4299e-02], |
| [-2.7399e-02, -3.4579e-02, -5.3095e-02, ..., -1.1379e-02, |
| -1.9393e-02, -4.0246e-02], |
| [-9.2012e-02, -1.4119e-01, 3.4798e-02, ..., -4.9943e-02, |
| 1.9133e-02, 3.0366e-02]], device='cuda:0'), grad: tensor([[ 2.9523e-07, 9.3132e-10, 7.2643e-08, ..., 0.0000e+00, |
| 4.5002e-06, 0.0000e+00], |
| [ 9.2573e-07, 5.5879e-09, 2.2259e-07, ..., 9.3132e-10, |
| -5.0198e-07, 0.0000e+00], |
| [ 1.3959e-04, 1.1176e-08, 6.3330e-08, ..., 0.0000e+00, |
| 5.2482e-05, 0.0000e+00], |
| ..., |
| [ 1.1493e-06, -2.7940e-08, 1.4473e-06, ..., 1.3970e-08, |
| 3.7681e-06, 0.0000e+00], |
| [ 5.0142e-06, -2.7940e-09, 1.2480e-07, ..., 0.0000e+00, |
| 1.8567e-05, 0.0000e+00], |
| [ 1.0822e-06, 9.3132e-10, -2.6263e-07, ..., 1.8626e-09, |
| -1.7047e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 125, bias, value: tensor([-0.0032, 0.0235, 0.0065, -0.0008, 0.0110, -0.0002, 0.0218, -0.0160, |
| 0.0151, -0.0026], device='cuda:0'), grad: tensor([ 9.6262e-06, 1.2014e-06, 2.0218e-04, -2.0838e-04, 2.0593e-05, |
| 1.2290e-04, -1.7917e-04, 1.5825e-05, 1.0389e-04, -8.8751e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 124, time 218.06, cls_loss 0.0034 cls_loss_mapping 0.0056 cls_loss_causal 0.5950 re_mapping 0.0083 re_causal 0.0250 /// teacc 98.79 lr 0.00010000 |
| Epoch 126, weight, value: tensor([[-0.0428, -0.0509, -0.0376, ..., -0.0866, 0.0010, -0.0148], |
| [-0.0362, 0.0260, -0.1168, ..., 0.0423, 0.0494, -0.0980], |
| [ 0.0527, -0.0002, -0.0503, ..., -0.0405, -0.0284, 0.0052], |
| ..., |
| [ 0.0435, -0.0438, 0.0012, ..., 0.0238, -0.0230, 0.0343], |
| [-0.0281, -0.0339, -0.0535, ..., -0.0116, -0.0212, -0.0403], |
| [-0.0922, -0.1420, 0.0347, ..., -0.0500, 0.0193, 0.0304]], |
| device='cuda:0'), grad: tensor([[ 6.7875e-06, 0.0000e+00, 4.1537e-07, ..., 3.5390e-08, |
| 5.9139e-07, 1.6764e-08], |
| [ 3.1050e-06, 9.3132e-10, 2.1774e-06, ..., 6.9290e-07, |
| -1.7852e-05, 1.8626e-09], |
| [ 1.0759e-05, 9.3132e-10, 4.9081e-07, ..., 3.0734e-08, |
| 1.4435e-07, 1.8626e-09], |
| ..., |
| [ 1.6838e-05, 0.0000e+00, 1.6131e-06, ..., 5.0291e-08, |
| 8.5309e-07, 0.0000e+00], |
| [ 5.2489e-06, -5.5879e-09, 2.3097e-07, ..., 2.7940e-09, |
| 6.3889e-07, 3.6322e-08], |
| [ 6.8024e-06, 0.0000e+00, 3.1479e-06, ..., 1.2666e-07, |
| -5.5879e-09, 1.8626e-09]], device='cuda:0') |
| Epoch 126, bias, value: tensor([-0.0029, 0.0244, 0.0063, -0.0001, 0.0111, -0.0007, 0.0218, -0.0160, |
| 0.0137, -0.0028], device='cuda:0'), grad: tensor([ 3.1926e-06, -8.8871e-05, 2.6375e-05, -7.8917e-05, 7.8261e-05, |
| 6.2212e-06, 3.8929e-06, 2.9132e-05, -3.8520e-06, 2.4319e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 125, time 217.57, cls_loss 0.0039 cls_loss_mapping 0.0057 cls_loss_causal 0.5592 re_mapping 0.0083 re_causal 0.0252 /// teacc 98.84 lr 0.00010000 |
| Epoch 127, weight, value: tensor([[-0.0426, -0.0521, -0.0376, ..., -0.0870, 0.0008, -0.0148], |
| [-0.0368, 0.0249, -0.1158, ..., 0.0442, 0.0492, -0.0984], |
| [ 0.0525, -0.0003, -0.0502, ..., -0.0409, -0.0288, 0.0055], |
| ..., |
| [ 0.0444, -0.0424, 0.0007, ..., 0.0229, -0.0225, 0.0340], |
| [-0.0284, -0.0344, -0.0547, ..., -0.0118, -0.0204, -0.0404], |
| [-0.0932, -0.1440, 0.0345, ..., -0.0502, 0.0194, 0.0304]], |
| device='cuda:0'), grad: tensor([[ 1.8310e-06, 0.0000e+00, 1.2293e-07, ..., 0.0000e+00, |
| 1.4342e-07, 0.0000e+00], |
| [ 1.1725e-06, 0.0000e+00, 1.0338e-07, ..., 0.0000e+00, |
| -2.3246e-06, 9.3132e-10], |
| [-1.0657e-04, 0.0000e+00, 1.7695e-08, ..., 0.0000e+00, |
| 3.2410e-07, -6.5193e-09], |
| ..., |
| [ 8.5354e-05, 0.0000e+00, 3.0454e-07, ..., 0.0000e+00, |
| 1.3057e-06, 1.8626e-09], |
| [ 4.9137e-06, 0.0000e+00, 5.5693e-07, ..., 0.0000e+00, |
| 1.3243e-06, 1.8626e-09], |
| [ 1.0684e-05, 0.0000e+00, -2.9616e-06, ..., 0.0000e+00, |
| -7.9125e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 127, bias, value: tensor([-2.7319e-03, 2.4492e-02, 6.0260e-03, -1.9379e-05, 1.0846e-02, |
| -5.6764e-04, 2.1557e-02, -1.5500e-02, 1.3877e-02, -3.1854e-03], |
| device='cuda:0'), grad: tensor([-6.3851e-06, -2.0042e-06, -1.6820e-04, 1.6257e-05, 1.7643e-05, |
| -8.6054e-06, 7.4655e-06, 1.4496e-04, 1.2785e-05, -1.4186e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 126, time 217.37, cls_loss 0.0034 cls_loss_mapping 0.0062 cls_loss_causal 0.5403 re_mapping 0.0081 re_causal 0.0253 /// teacc 98.90 lr 0.00010000 |
| Epoch 128, weight, value: tensor([[-0.0425, -0.0528, -0.0379, ..., -0.0872, 0.0007, -0.0148], |
| [-0.0369, 0.0253, -0.1164, ..., 0.0444, 0.0497, -0.0987], |
| [ 0.0513, -0.0005, -0.0504, ..., -0.0414, -0.0289, 0.0055], |
| ..., |
| [ 0.0457, -0.0424, 0.0005, ..., 0.0229, -0.0225, 0.0340], |
| [-0.0285, -0.0345, -0.0547, ..., -0.0119, -0.0210, -0.0404], |
| [-0.0939, -0.1444, 0.0344, ..., -0.0503, 0.0195, 0.0304]], |
| device='cuda:0'), grad: tensor([[ 1.8422e-06, 0.0000e+00, 1.8338e-06, ..., 4.6566e-10, |
| 1.2480e-07, 2.7940e-09], |
| [ 1.0021e-05, 1.8626e-09, 2.9337e-06, ..., 2.3283e-09, |
| -1.0785e-06, 2.0955e-08], |
| [ 2.4457e-06, 9.3132e-10, 1.3243e-06, ..., 1.8626e-09, |
| 1.7649e-07, 3.4925e-08], |
| ..., |
| [-2.2396e-05, -6.5193e-09, -3.8091e-06, ..., 4.6566e-09, |
| 4.9965e-07, -1.0431e-07], |
| [ 1.6659e-05, 4.6566e-10, 1.1057e-05, ..., 7.9162e-09, |
| -3.0734e-08, 4.1910e-09], |
| [ 2.5872e-06, 4.6566e-10, 8.7731e-07, ..., 1.1642e-08, |
| 4.6100e-08, 2.5611e-08]], device='cuda:0') |
| Epoch 128, bias, value: tensor([-0.0025, 0.0247, 0.0048, -0.0004, 0.0110, -0.0006, 0.0213, -0.0144, |
| 0.0139, -0.0034], device='cuda:0'), grad: tensor([ 1.6913e-05, 3.0816e-05, 1.3649e-05, 3.4094e-05, 1.5363e-05, |
| 2.3520e-04, -3.9649e-04, -5.6863e-05, 9.9123e-05, 8.1956e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 127, time 217.16, cls_loss 0.0054 cls_loss_mapping 0.0067 cls_loss_causal 0.5697 re_mapping 0.0086 re_causal 0.0245 /// teacc 98.79 lr 0.00010000 |
| Epoch 129, weight, value: tensor([[-0.0431, -0.0539, -0.0384, ..., -0.0876, 0.0008, -0.0148], |
| [-0.0363, 0.0256, -0.1174, ..., 0.0447, 0.0508, -0.0988], |
| [ 0.0513, 0.0008, -0.0493, ..., -0.0422, -0.0292, 0.0056], |
| ..., |
| [ 0.0455, -0.0440, 0.0005, ..., 0.0229, -0.0243, 0.0340], |
| [-0.0289, -0.0348, -0.0554, ..., -0.0121, -0.0216, -0.0404], |
| [-0.0930, -0.1463, 0.0342, ..., -0.0503, 0.0203, 0.0304]], |
| device='cuda:0'), grad: tensor([[ 4.1258e-07, 0.0000e+00, 2.1420e-07, ..., 0.0000e+00, |
| 6.1467e-08, 9.3132e-10], |
| [ 9.4533e-05, 0.0000e+00, 6.4727e-07, ..., 4.6566e-09, |
| -2.1420e-08, 5.5879e-09], |
| [ 5.4762e-07, 0.0000e+00, 3.4459e-07, ..., 9.3132e-10, |
| 1.0896e-07, 4.6566e-09], |
| ..., |
| [-7.2527e-04, 0.0000e+00, 6.6161e-05, ..., 2.7008e-08, |
| 8.5831e-06, -2.6077e-08], |
| [ 1.7481e-06, 0.0000e+00, 5.9828e-06, ..., 0.0000e+00, |
| 1.6633e-06, 3.7253e-09], |
| [ 5.7745e-04, 0.0000e+00, -9.6083e-05, ..., 0.0000e+00, |
| -1.3061e-05, 8.3819e-09]], device='cuda:0') |
| Epoch 129, bias, value: tensor([-0.0022, 0.0255, 0.0050, -0.0011, 0.0105, -0.0012, 0.0227, -0.0155, |
| 0.0130, -0.0025], device='cuda:0'), grad: tensor([ 1.6550e-06, 2.4092e-04, 4.4219e-06, 1.1832e-04, 8.6546e-05, |
| -1.9282e-05, 4.3474e-06, -1.5459e-03, 1.9282e-05, 1.0891e-03], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 128, time 217.40, cls_loss 0.0045 cls_loss_mapping 0.0068 cls_loss_causal 0.5977 re_mapping 0.0082 re_causal 0.0249 /// teacc 98.87 lr 0.00010000 |
| Epoch 130, weight, value: tensor([[-0.0434, -0.0553, -0.0403, ..., -0.0878, -0.0003, -0.0160], |
| [-0.0365, 0.0258, -0.1181, ..., 0.0450, 0.0512, -0.0995], |
| [ 0.0518, 0.0008, -0.0492, ..., -0.0426, -0.0294, 0.0051], |
| ..., |
| [ 0.0457, -0.0443, 0.0003, ..., 0.0227, -0.0246, 0.0338], |
| [-0.0284, -0.0349, -0.0574, ..., -0.0122, -0.0218, -0.0405], |
| [-0.0943, -0.1489, 0.0365, ..., -0.0505, 0.0213, 0.0316]], |
| device='cuda:0'), grad: tensor([[ 1.4072e-06, 2.8219e-07, 1.3411e-07, ..., 0.0000e+00, |
| 3.3267e-06, 7.4506e-09], |
| [ 1.6105e-04, 3.3975e-05, 1.5926e-07, ..., 0.0000e+00, |
| -1.9222e-06, 9.3132e-10], |
| [-2.0921e-04, -4.4346e-05, 3.9209e-07, ..., 0.0000e+00, |
| 1.2405e-06, -4.0978e-08], |
| ..., |
| [ 8.6427e-06, 2.0191e-06, 1.5460e-07, ..., 0.0000e+00, |
| 4.3306e-07, 1.1176e-08], |
| [ 4.1910e-06, 7.6555e-07, -2.9746e-06, ..., 0.0000e+00, |
| 1.5181e-07, 9.3132e-10], |
| [ 1.1116e-05, 2.2855e-06, 2.7418e-06, ..., 0.0000e+00, |
| 1.0040e-06, 1.4901e-08]], device='cuda:0') |
| Epoch 130, bias, value: tensor([-0.0031, 0.0255, 0.0054, -0.0015, 0.0102, -0.0016, 0.0219, -0.0154, |
| 0.0128, -0.0010], device='cuda:0'), grad: tensor([ 9.9242e-06, 2.6059e-04, -3.4046e-04, 2.0459e-05, 4.7684e-06, |
| 1.5073e-05, -1.1809e-05, 1.6093e-05, -4.4331e-06, 2.9758e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 129, time 217.34, cls_loss 0.0038 cls_loss_mapping 0.0056 cls_loss_causal 0.5527 re_mapping 0.0082 re_causal 0.0253 /// teacc 98.82 lr 0.00010000 |
| Epoch 131, weight, value: tensor([[-0.0437, -0.0563, -0.0406, ..., -0.0878, -0.0007, -0.0165], |
| [-0.0375, 0.0238, -0.1184, ..., 0.0451, 0.0515, -0.0998], |
| [ 0.0517, 0.0022, -0.0495, ..., -0.0427, -0.0294, 0.0050], |
| ..., |
| [ 0.0467, -0.0440, 0.0016, ..., 0.0227, -0.0247, 0.0336], |
| [-0.0289, -0.0351, -0.0576, ..., -0.0122, -0.0222, -0.0405], |
| [-0.0954, -0.1510, 0.0365, ..., -0.0505, 0.0218, 0.0321]], |
| device='cuda:0'), grad: tensor([[ 7.7300e-07, 1.0151e-07, 1.3877e-07, ..., 0.0000e+00, |
| 3.6974e-07, 0.0000e+00], |
| [ 8.5533e-06, 1.4249e-07, 3.5763e-07, ..., 0.0000e+00, |
| -3.4198e-06, 0.0000e+00], |
| [-5.0664e-06, -2.1160e-06, 1.4063e-07, ..., 0.0000e+00, |
| 8.2608e-07, -5.5879e-09], |
| ..., |
| [ 7.8380e-06, 5.4482e-07, 1.8522e-05, ..., 0.0000e+00, |
| 4.0568e-06, 2.7940e-09], |
| [ 9.5144e-06, 1.1539e-06, 1.5851e-06, ..., 0.0000e+00, |
| 3.0361e-07, 9.3132e-10], |
| [ 7.9036e-05, 8.3819e-09, 1.2070e-04, ..., 0.0000e+00, |
| -4.2245e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 131, bias, value: tensor([-0.0032, 0.0251, 0.0053, -0.0016, 0.0097, -0.0019, 0.0216, -0.0143, |
| 0.0131, -0.0011], device='cuda:0'), grad: tensor([ 2.6338e-06, 9.9242e-06, -4.4778e-06, 1.5542e-05, -4.4227e-04, |
| 3.9674e-06, 9.7975e-07, 7.2122e-05, 2.2560e-05, 3.1805e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 130, time 217.21, cls_loss 0.0041 cls_loss_mapping 0.0064 cls_loss_causal 0.5552 re_mapping 0.0082 re_causal 0.0238 /// teacc 98.89 lr 0.00010000 |
| Epoch 132, weight, value: tensor([[-0.0466, -0.0579, -0.0408, ..., -0.0879, -0.0012, -0.0166], |
| [-0.0385, 0.0230, -0.1188, ..., 0.0449, 0.0518, -0.1009], |
| [ 0.0519, 0.0021, -0.0489, ..., -0.0428, -0.0295, 0.0092], |
| ..., |
| [ 0.0473, -0.0432, 0.0017, ..., 0.0227, -0.0250, 0.0335], |
| [-0.0294, -0.0348, -0.0575, ..., -0.0123, -0.0221, -0.0409], |
| [-0.0963, -0.1540, 0.0359, ..., -0.0500, 0.0220, 0.0320]], |
| device='cuda:0'), grad: tensor([[ 1.5646e-07, 2.7940e-09, 4.4424e-07, ..., 0.0000e+00, |
| 1.2480e-07, 2.4214e-08], |
| [ 4.7311e-07, 4.0047e-08, 1.7583e-06, ..., 0.0000e+00, |
| -3.4831e-06, 7.4506e-09], |
| [-5.1036e-07, 2.8871e-08, 7.6182e-07, ..., 0.0000e+00, |
| 2.4866e-07, -2.4866e-07], |
| ..., |
| [-3.5688e-06, -1.7975e-07, 6.4913e-07, ..., 0.0000e+00, |
| 1.5320e-06, 1.8068e-07], |
| [ 4.8243e-07, 2.7940e-09, 4.0233e-07, ..., 0.0000e+00, |
| 9.6671e-07, 2.1420e-08], |
| [ 1.5311e-06, 5.2154e-08, 4.7326e-05, ..., 0.0000e+00, |
| -9.8348e-07, 1.8626e-09]], device='cuda:0') |
| Epoch 132, bias, value: tensor([-0.0034, 0.0246, 0.0053, -0.0010, 0.0100, -0.0024, 0.0197, -0.0138, |
| 0.0151, -0.0015], device='cuda:0'), grad: tensor([ 2.4825e-05, -1.5842e-06, 8.5384e-06, 4.1090e-06, -8.9109e-05, |
| 3.8072e-06, -8.0407e-05, 2.0396e-07, 8.9332e-06, 1.2082e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 131, time 217.19, cls_loss 0.0032 cls_loss_mapping 0.0047 cls_loss_causal 0.5390 re_mapping 0.0083 re_causal 0.0238 /// teacc 98.80 lr 0.00010000 |
| Epoch 133, weight, value: tensor([[-0.0463, -0.0582, -0.0412, ..., -0.0892, -0.0010, -0.0166], |
| [-0.0383, 0.0232, -0.1192, ..., 0.0449, 0.0529, -0.1012], |
| [ 0.0523, 0.0023, -0.0494, ..., -0.0434, -0.0297, 0.0095], |
| ..., |
| [ 0.0471, -0.0435, 0.0015, ..., 0.0225, -0.0259, 0.0332], |
| [-0.0303, -0.0349, -0.0581, ..., -0.0126, -0.0239, -0.0412], |
| [-0.0964, -0.1553, 0.0365, ..., -0.0498, 0.0228, 0.0320]], |
| device='cuda:0'), grad: tensor([[ 4.6566e-07, 9.3132e-10, 8.1956e-08, ..., 2.7195e-07, |
| 6.0629e-07, 0.0000e+00], |
| [-2.2069e-05, 2.4214e-08, 9.7789e-08, ..., -2.5630e-05, |
| -5.0306e-05, 0.0000e+00], |
| [ 1.6153e-05, 4.6566e-09, 2.3376e-07, ..., 2.0161e-05, |
| 3.9101e-05, 0.0000e+00], |
| ..., |
| [ 1.3942e-06, -7.1712e-08, -9.7416e-07, ..., 2.8163e-06, |
| 5.8860e-06, 0.0000e+00], |
| [ 8.4471e-07, 3.7253e-09, 6.8266e-07, ..., 5.1223e-08, |
| 4.0885e-07, 0.0000e+00], |
| [ 7.3668e-07, 1.5832e-08, 2.9337e-07, ..., 1.0990e-07, |
| -6.2492e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 133, bias, value: tensor([-0.0031, 0.0250, 0.0054, -0.0017, 0.0094, -0.0017, 0.0194, -0.0142, |
| 0.0142, -0.0007], device='cuda:0'), grad: tensor([-4.5240e-05, -2.7251e-04, 2.1565e-04, 4.6715e-06, 2.0668e-05, |
| 2.1495e-06, 4.8101e-05, 1.2137e-05, 4.9211e-06, 9.2387e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 132, time 217.25, cls_loss 0.0039 cls_loss_mapping 0.0056 cls_loss_causal 0.5650 re_mapping 0.0080 re_causal 0.0237 /// teacc 98.77 lr 0.00010000 |
| Epoch 134, weight, value: tensor([[-0.0463, -0.0613, -0.0417, ..., -0.0893, -0.0013, -0.0187], |
| [-0.0384, 0.0235, -0.1194, ..., 0.0455, 0.0533, -0.1029], |
| [ 0.0525, 0.0023, -0.0488, ..., -0.0442, -0.0299, 0.0089], |
| ..., |
| [ 0.0473, -0.0432, 0.0013, ..., 0.0221, -0.0259, 0.0329], |
| [-0.0310, -0.0349, -0.0587, ..., -0.0128, -0.0245, -0.0417], |
| [-0.0979, -0.1640, 0.0360, ..., -0.0499, 0.0225, 0.0341]], |
| device='cuda:0'), grad: tensor([[ 3.4198e-06, 1.8626e-09, 1.8626e-07, ..., 7.4506e-09, |
| 6.5193e-08, 3.3434e-06], |
| [ 1.7323e-07, 1.8626e-09, 1.8394e-06, ..., 3.1665e-07, |
| -1.7062e-06, 4.3772e-08], |
| [-5.4911e-06, -4.1910e-08, 4.7218e-07, ..., 4.2841e-08, |
| 1.4994e-07, -5.1446e-06], |
| ..., |
| [ 7.6462e-07, 3.5390e-08, 9.9931e-07, ..., 6.5193e-08, |
| 7.1619e-07, 6.5845e-07], |
| [ 2.3376e-07, -2.7940e-09, 9.4250e-07, ..., 4.6566e-09, |
| 7.9256e-07, 1.2200e-07], |
| [ 4.0885e-07, 9.3132e-10, 1.3942e-06, ..., 5.5879e-08, |
| -5.8766e-07, 2.5798e-07]], device='cuda:0') |
| Epoch 134, bias, value: tensor([-0.0042, 0.0252, 0.0058, -0.0018, 0.0096, -0.0012, 0.0194, -0.0140, |
| 0.0137, -0.0006], device='cuda:0'), grad: tensor([-3.5260e-06, 2.3991e-06, -1.3016e-05, 1.7034e-06, -1.1824e-05, |
| 1.2685e-06, 8.4788e-06, 6.2846e-06, 3.9376e-06, 4.2804e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 133, time 217.18, cls_loss 0.0043 cls_loss_mapping 0.0053 cls_loss_causal 0.5260 re_mapping 0.0081 re_causal 0.0236 /// teacc 98.89 lr 0.00010000 |
| Epoch 135, weight, value: tensor([[-0.0463, -0.0633, -0.0420, ..., -0.0893, -0.0027, -0.0206], |
| [-0.0399, 0.0244, -0.1198, ..., 0.0464, 0.0537, -0.1077], |
| [ 0.0537, 0.0029, -0.0491, ..., -0.0443, -0.0300, 0.0102], |
| ..., |
| [ 0.0469, -0.0448, 0.0012, ..., 0.0213, -0.0263, 0.0317], |
| [-0.0315, -0.0351, -0.0591, ..., -0.0130, -0.0244, -0.0432], |
| [-0.0995, -0.1688, 0.0354, ..., -0.0501, 0.0233, 0.0359]], |
| device='cuda:0'), grad: tensor([[ 2.6636e-07, 1.6764e-08, 3.1665e-08, ..., 0.0000e+00, |
| -1.0364e-05, 0.0000e+00], |
| [ 9.0618e-07, 3.4273e-07, 8.3819e-08, ..., 0.0000e+00, |
| -3.4496e-06, 0.0000e+00], |
| [-2.2911e-06, 3.1944e-07, 5.3085e-08, ..., 9.3132e-10, |
| 1.6820e-06, -9.3132e-10], |
| ..., |
| [ 9.4995e-06, -1.3513e-06, 3.7216e-06, ..., 9.3132e-10, |
| 4.5169e-07, 9.3132e-10], |
| [ 1.9260e-06, 6.8918e-08, 1.1735e-07, ..., 9.3132e-10, |
| 3.4153e-05, 0.0000e+00], |
| [-5.7369e-05, 1.4901e-08, -1.3188e-05, ..., 0.0000e+00, |
| 1.6633e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 135, bias, value: tensor([-0.0055, 0.0245, 0.0065, -0.0011, 0.0104, -0.0005, 0.0197, -0.0143, |
| 0.0131, -0.0003], device='cuda:0'), grad: tensor([-4.0740e-05, -4.6194e-06, 3.1628e-06, 8.1360e-06, 3.3259e-04, |
| 5.8270e-04, -6.5899e-04, 1.1539e-04, 1.3030e-04, -4.6730e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 134, time 217.67, cls_loss 0.0039 cls_loss_mapping 0.0075 cls_loss_causal 0.5445 re_mapping 0.0081 re_causal 0.0231 /// teacc 98.92 lr 0.00010000 |
| Epoch 136, weight, value: tensor([[-0.0446, -0.0650, -0.0424, ..., -0.0896, -0.0028, -0.0208], |
| [-0.0403, 0.0243, -0.1201, ..., 0.0471, 0.0548, -0.1087], |
| [ 0.0539, 0.0028, -0.0515, ..., -0.0444, -0.0302, 0.0104], |
| ..., |
| [ 0.0469, -0.0449, 0.0010, ..., 0.0210, -0.0274, 0.0314], |
| [-0.0317, -0.0344, -0.0599, ..., -0.0131, -0.0246, -0.0438], |
| [-0.1025, -0.1704, 0.0346, ..., -0.0506, 0.0234, 0.0363]], |
| device='cuda:0'), grad: tensor([[ 1.3322e-05, 0.0000e+00, 5.4911e-06, ..., 0.0000e+00, |
| 1.3970e-08, 0.0000e+00], |
| [ 5.3458e-06, 0.0000e+00, 2.2128e-06, ..., 0.0000e+00, |
| -1.4529e-07, 9.3132e-10], |
| [ 1.5602e-05, 0.0000e+00, 6.6981e-06, ..., 0.0000e+00, |
| 5.5879e-08, 4.6566e-09], |
| ..., |
| [ 4.7497e-08, 0.0000e+00, 1.9744e-07, ..., 0.0000e+00, |
| 4.2841e-08, -7.4506e-09], |
| [ 1.3858e-06, 0.0000e+00, 8.2701e-07, ..., 0.0000e+00, |
| 8.3819e-09, 0.0000e+00], |
| [-1.6898e-05, 0.0000e+00, -6.8434e-06, ..., 0.0000e+00, |
| -2.0489e-08, 9.3132e-10]], device='cuda:0') |
| Epoch 136, bias, value: tensor([-4.9428e-03, 2.4967e-02, 6.1505e-03, -4.1212e-04, 1.0852e-02, |
| 9.9237e-05, 1.9116e-02, -1.4833e-02, 1.2888e-02, -9.0948e-04], |
| device='cuda:0'), grad: tensor([ 4.9740e-05, 1.3009e-05, 4.1425e-05, 1.8209e-05, 3.4142e-04, |
| 1.0274e-05, -4.0984e-04, 1.1306e-06, 3.4682e-06, -6.9201e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 135, time 217.64, cls_loss 0.0030 cls_loss_mapping 0.0049 cls_loss_causal 0.5488 re_mapping 0.0080 re_causal 0.0247 /// teacc 98.76 lr 0.00010000 |
| Epoch 137, weight, value: tensor([[-0.0448, -0.0658, -0.0425, ..., -0.0896, -0.0030, -0.0208], |
| [-0.0404, 0.0244, -0.1203, ..., 0.0471, 0.0550, -0.1090], |
| [ 0.0538, 0.0026, -0.0521, ..., -0.0444, -0.0305, 0.0103], |
| ..., |
| [ 0.0471, -0.0449, 0.0011, ..., 0.0211, -0.0276, 0.0316], |
| [-0.0315, -0.0326, -0.0603, ..., -0.0132, -0.0247, -0.0440], |
| [-0.1029, -0.1710, 0.0333, ..., -0.0515, 0.0235, 0.0363]], |
| device='cuda:0'), grad: tensor([[ 2.5332e-07, 0.0000e+00, 4.4703e-08, ..., 0.0000e+00, |
| 2.0396e-07, 1.5739e-07], |
| [ 1.1683e-05, 0.0000e+00, 1.1083e-07, ..., 0.0000e+00, |
| -2.6245e-06, 1.6810e-06], |
| [ 2.2486e-05, 0.0000e+00, -1.3867e-06, ..., 0.0000e+00, |
| 1.5507e-06, 3.6452e-06], |
| ..., |
| [-5.7071e-05, 0.0000e+00, 4.0978e-08, ..., 0.0000e+00, |
| 6.8173e-07, -8.5905e-06], |
| [ 3.2987e-06, 0.0000e+00, 1.6764e-07, ..., 0.0000e+00, |
| 4.4145e-07, 3.7067e-07], |
| [ 1.4596e-05, 0.0000e+00, 7.7300e-08, ..., 0.0000e+00, |
| -2.3954e-06, 2.1141e-06]], device='cuda:0') |
| Epoch 137, bias, value: tensor([-4.9095e-03, 2.4978e-02, 5.8470e-03, 8.5718e-05, 1.1930e-02, |
| -4.4005e-04, 1.9489e-02, -1.4733e-02, 1.3005e-02, -1.7870e-03], |
| device='cuda:0'), grad: tensor([-1.5870e-05, 3.3855e-05, 9.0301e-05, 1.9558e-06, 1.3597e-05, |
| 7.1041e-06, 3.5763e-06, -2.0278e-04, 1.1377e-05, 5.6952e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 136, time 217.36, cls_loss 0.0033 cls_loss_mapping 0.0059 cls_loss_causal 0.5427 re_mapping 0.0076 re_causal 0.0228 /// teacc 98.84 lr 0.00010000 |
| Epoch 138, weight, value: tensor([[-0.0449, -0.0669, -0.0431, ..., -0.0898, -0.0024, -0.0211], |
| [-0.0418, 0.0247, -0.1206, ..., 0.0471, 0.0555, -0.1102], |
| [ 0.0541, 0.0025, -0.0514, ..., -0.0444, -0.0306, 0.0113], |
| ..., |
| [ 0.0481, -0.0450, 0.0007, ..., 0.0214, -0.0281, 0.0316], |
| [-0.0318, -0.0334, -0.0608, ..., -0.0134, -0.0248, -0.0449], |
| [-0.1037, -0.1713, 0.0326, ..., -0.0523, 0.0237, 0.0365]], |
| device='cuda:0'), grad: tensor([[-1.4043e-04, 9.3132e-10, 6.2212e-07, ..., 1.8626e-09, |
| -2.4009e-06, 9.8720e-08], |
| [ 3.7961e-06, -4.7497e-08, 3.4086e-07, ..., 3.4459e-08, |
| -3.8370e-06, 5.4948e-08], |
| [-1.2718e-05, 8.3819e-09, -4.4443e-06, ..., 3.7253e-09, |
| -1.1260e-06, -1.3718e-06], |
| ..., |
| [ 4.2692e-06, 1.0245e-08, 4.8708e-07, ..., 1.2107e-08, |
| 6.2026e-07, 5.4017e-08], |
| [ 6.9402e-06, 1.3970e-08, 2.5183e-06, ..., 1.8626e-09, |
| 2.3618e-06, 3.9116e-08], |
| [ 1.2279e-04, 9.3132e-10, 1.3933e-06, ..., 8.3819e-09, |
| 2.7195e-06, 7.5437e-08]], device='cuda:0') |
| Epoch 138, bias, value: tensor([-4.3489e-03, 2.3705e-02, 5.6880e-03, -1.3175e-04, 1.2197e-02, |
| -5.3660e-05, 1.9176e-02, -1.3496e-02, 1.3220e-02, -2.2984e-03], |
| device='cuda:0'), grad: tensor([-7.7009e-04, 1.1409e-06, -1.7881e-05, 5.5611e-05, 6.0759e-06, |
| 5.9903e-06, 9.7975e-06, 7.3388e-06, 3.2365e-05, 6.6900e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 137, time 217.46, cls_loss 0.0042 cls_loss_mapping 0.0060 cls_loss_causal 0.5544 re_mapping 0.0078 re_causal 0.0232 /// teacc 98.87 lr 0.00010000 |
| Epoch 139, weight, value: tensor([[-0.0448, -0.0679, -0.0437, ..., -0.0912, -0.0026, -0.0211], |
| [-0.0423, 0.0250, -0.1206, ..., 0.0474, 0.0550, -0.1124], |
| [ 0.0543, 0.0024, -0.0517, ..., -0.0447, -0.0295, 0.0122], |
| ..., |
| [ 0.0486, -0.0451, 0.0005, ..., 0.0205, -0.0287, 0.0302], |
| [-0.0324, -0.0338, -0.0610, ..., -0.0138, -0.0253, -0.0470], |
| [-0.1049, -0.1715, 0.0322, ..., -0.0523, 0.0242, 0.0366]], |
| device='cuda:0'), grad: tensor([[ 2.2072e-07, 0.0000e+00, 1.9558e-08, ..., 0.0000e+00, |
| 2.0899e-06, 1.8626e-08], |
| [ 5.0757e-07, 5.5879e-09, 2.4214e-08, ..., 0.0000e+00, |
| 5.5879e-08, 1.3970e-08], |
| [ 7.1526e-06, 9.3132e-10, 4.1910e-08, ..., 0.0000e+00, |
| 1.5870e-06, 3.7253e-09], |
| ..., |
| [ 1.2174e-05, -9.3132e-09, 1.0710e-07, ..., 0.0000e+00, |
| 1.0626e-06, 3.9116e-08], |
| [ 6.2585e-07, 9.3132e-10, 2.1141e-07, ..., 0.0000e+00, |
| 2.4289e-06, 1.2014e-07], |
| [ 6.1095e-07, 9.3132e-10, -6.9197e-07, ..., 0.0000e+00, |
| -6.6385e-06, -3.9302e-07]], device='cuda:0') |
| Epoch 139, bias, value: tensor([-0.0043, 0.0236, 0.0059, 0.0019, 0.0121, -0.0017, 0.0196, -0.0133, |
| 0.0130, -0.0027], device='cuda:0'), grad: tensor([ 4.6156e-06, 1.5106e-06, 1.3456e-05, -2.1562e-05, 2.1249e-05, |
| 3.4962e-06, -1.4141e-05, 2.0027e-05, 7.1302e-06, -3.5793e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 138, time 217.48, cls_loss 0.0047 cls_loss_mapping 0.0060 cls_loss_causal 0.5589 re_mapping 0.0077 re_causal 0.0233 /// teacc 98.84 lr 0.00010000 |
| Epoch 140, weight, value: tensor([[-0.0451, -0.0694, -0.0438, ..., -0.0915, -0.0029, -0.0212], |
| [-0.0424, 0.0250, -0.1212, ..., 0.0492, 0.0561, -0.1182], |
| [ 0.0536, 0.0024, -0.0518, ..., -0.0449, -0.0290, 0.0139], |
| ..., |
| [ 0.0495, -0.0451, 0.0025, ..., 0.0231, -0.0302, 0.0292], |
| [-0.0327, -0.0336, -0.0619, ..., -0.0139, -0.0265, -0.0526], |
| [-0.1056, -0.1719, 0.0334, ..., -0.0525, 0.0227, 0.0365]], |
| device='cuda:0'), grad: tensor([[-2.9244e-07, 9.3132e-10, 3.9581e-07, ..., 0.0000e+00, |
| 5.7742e-08, 2.1420e-08], |
| [ 1.7285e-06, 3.7253e-09, 2.3823e-06, ..., 1.7788e-07, |
| -7.9162e-08, 1.3132e-07], |
| [-9.8813e-07, 7.4506e-09, 6.0350e-07, ..., 9.3132e-10, |
| 1.6298e-07, 3.4459e-08], |
| ..., |
| [-2.2240e-06, -1.9558e-08, 7.5027e-06, ..., -2.0675e-07, |
| 4.1723e-07, 4.1444e-07], |
| [-7.4413e-07, 9.3132e-10, 6.8359e-07, ..., 9.3132e-10, |
| -3.6769e-06, 2.7008e-08], |
| [ 7.7859e-07, 9.3132e-10, 3.5837e-06, ..., 2.1420e-08, |
| -2.4047e-06, 2.9150e-07]], device='cuda:0') |
| Epoch 140, bias, value: tensor([-0.0044, 0.0243, 0.0056, 0.0009, 0.0124, -0.0006, 0.0198, -0.0127, |
| 0.0120, -0.0035], device='cuda:0'), grad: tensor([ 1.6764e-07, 8.1733e-06, 1.2890e-06, 1.4327e-05, -2.9519e-05, |
| 8.2105e-06, 1.5181e-06, 1.1414e-05, -2.0623e-05, 5.0142e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 139---------------------------------------------------- |
| epoch 139, time 218.25, cls_loss 0.0035 cls_loss_mapping 0.0048 cls_loss_causal 0.5540 re_mapping 0.0076 re_causal 0.0230 /// teacc 98.93 lr 0.00010000 |
| Epoch 141, weight, value: tensor([[-0.0454, -0.0696, -0.0441, ..., -0.0916, -0.0033, -0.0213], |
| [-0.0428, 0.0251, -0.1215, ..., 0.0494, 0.0568, -0.1189], |
| [ 0.0539, 0.0017, -0.0528, ..., -0.0449, -0.0294, 0.0139], |
| ..., |
| [ 0.0491, -0.0452, 0.0023, ..., 0.0229, -0.0309, 0.0282], |
| [-0.0330, -0.0320, -0.0622, ..., -0.0140, -0.0268, -0.0538], |
| [-0.1058, -0.1722, 0.0339, ..., -0.0523, 0.0233, 0.0367]], |
| device='cuda:0'), grad: tensor([[-2.4121e-06, 1.8626e-09, 3.4180e-07, ..., 0.0000e+00, |
| 2.7101e-07, 4.1910e-08], |
| [ 3.8277e-07, 2.7940e-09, 1.4901e-07, ..., 0.0000e+00, |
| -5.5879e-09, 1.0245e-08], |
| [-3.5297e-07, -2.9802e-08, 2.8536e-06, ..., 0.0000e+00, |
| 5.9325e-07, -4.6566e-09], |
| ..., |
| [-5.1036e-07, 9.3132e-09, 1.4435e-07, ..., 0.0000e+00, |
| 3.7625e-07, 7.4506e-09], |
| [ 3.1851e-07, 4.6566e-09, -8.1584e-07, ..., 0.0000e+00, |
| 2.1271e-06, 3.3900e-07], |
| [ 1.5451e-06, 9.3132e-10, 2.3656e-07, ..., 0.0000e+00, |
| -1.5395e-06, 1.8626e-09]], device='cuda:0') |
| Epoch 141, bias, value: tensor([-0.0046, 0.0242, 0.0054, 0.0019, 0.0122, -0.0005, 0.0196, -0.0133, |
| 0.0120, -0.0029], device='cuda:0'), grad: tensor([-5.8636e-06, 1.3923e-06, 1.5125e-05, 3.0808e-06, 3.4496e-06, |
| 7.8797e-05, -1.0431e-04, 1.5749e-06, 8.1062e-06, -1.5376e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 140, time 217.41, cls_loss 0.0032 cls_loss_mapping 0.0053 cls_loss_causal 0.5313 re_mapping 0.0077 re_causal 0.0235 /// teacc 98.85 lr 0.00010000 |
| Epoch 142, weight, value: tensor([[-0.0447, -0.0701, -0.0445, ..., -0.0917, -0.0038, -0.0215], |
| [-0.0425, 0.0258, -0.1222, ..., 0.0500, 0.0573, -0.1193], |
| [ 0.0538, 0.0013, -0.0520, ..., -0.0469, -0.0297, 0.0162], |
| ..., |
| [ 0.0490, -0.0456, 0.0023, ..., 0.0228, -0.0317, 0.0284], |
| [-0.0329, -0.0321, -0.0626, ..., -0.0140, -0.0270, -0.0592], |
| [-0.1066, -0.1723, 0.0340, ..., -0.0524, 0.0241, 0.0365]], |
| device='cuda:0'), grad: tensor([[-2.8033e-07, 0.0000e+00, 8.9034e-07, ..., 0.0000e+00, |
| 2.0675e-07, 6.5193e-08], |
| [ 1.0421e-06, 0.0000e+00, 1.2983e-06, ..., 0.0000e+00, |
| -1.0375e-06, 9.2201e-08], |
| [ 4.2692e-06, 0.0000e+00, 4.3064e-06, ..., 0.0000e+00, |
| 1.2480e-07, 3.8464e-07], |
| ..., |
| [-1.5497e-06, 0.0000e+00, 2.4289e-06, ..., 0.0000e+00, |
| 4.1258e-07, -8.6706e-07], |
| [ 1.0962e-06, 0.0000e+00, 1.9185e-07, ..., 0.0000e+00, |
| 5.1223e-08, 2.5984e-07], |
| [ 7.5717e-07, 0.0000e+00, 6.8545e-06, ..., 0.0000e+00, |
| 4.9733e-07, 6.8918e-08]], device='cuda:0') |
| Epoch 142, bias, value: tensor([-0.0046, 0.0246, 0.0050, 0.0019, 0.0117, -0.0004, 0.0194, -0.0136, |
| 0.0123, -0.0024], device='cuda:0'), grad: tensor([-4.0792e-06, 1.3718e-06, 1.4804e-05, 2.8163e-06, -3.7611e-05, |
| -7.2550e-07, 5.1558e-06, 1.1157e-06, -5.2247e-07, 1.7643e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 141, time 217.42, cls_loss 0.0035 cls_loss_mapping 0.0060 cls_loss_causal 0.5447 re_mapping 0.0077 re_causal 0.0229 /// teacc 98.83 lr 0.00010000 |
| Epoch 143, weight, value: tensor([[-0.0448, -0.0706, -0.0453, ..., -0.0917, -0.0041, -0.0215], |
| [-0.0414, 0.0260, -0.1232, ..., 0.0500, 0.0595, -0.1196], |
| [ 0.0543, 0.0015, -0.0515, ..., -0.0469, -0.0298, 0.0175], |
| ..., |
| [ 0.0485, -0.0459, 0.0021, ..., 0.0228, -0.0325, 0.0282], |
| [-0.0333, -0.0323, -0.0634, ..., -0.0140, -0.0274, -0.0595], |
| [-0.1097, -0.1725, 0.0332, ..., -0.0524, 0.0228, 0.0364]], |
| device='cuda:0'), grad: tensor([[ 1.1083e-06, 4.6566e-09, 2.0768e-07, ..., 0.0000e+00, |
| 4.4703e-08, 0.0000e+00], |
| [ 7.3314e-06, 2.0768e-07, 1.2480e-07, ..., 0.0000e+00, |
| 3.3807e-07, 0.0000e+00], |
| [-3.9369e-05, 4.2841e-08, -5.3197e-06, ..., 0.0000e+00, |
| 1.3039e-08, 0.0000e+00], |
| ..., |
| [-9.8944e-06, -5.2340e-07, 1.1297e-06, ..., 0.0000e+00, |
| -1.5935e-06, 0.0000e+00], |
| [ 2.1383e-05, 3.7253e-09, 1.2234e-05, ..., 0.0000e+00, |
| 7.1116e-06, 0.0000e+00], |
| [ 6.4820e-06, 1.1548e-07, 2.1048e-07, ..., 0.0000e+00, |
| 6.9663e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 143, bias, value: tensor([-0.0046, 0.0259, 0.0053, 0.0023, 0.0116, -0.0006, 0.0198, -0.0144, |
| 0.0118, -0.0031], device='cuda:0'), grad: tensor([ 1.6829e-06, 1.2398e-05, -7.4327e-05, 1.2413e-05, 8.0094e-06, |
| 1.1623e-04, -1.9383e-04, -1.9014e-05, 1.2445e-04, 1.1809e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 142, time 217.58, cls_loss 0.0028 cls_loss_mapping 0.0043 cls_loss_causal 0.5187 re_mapping 0.0077 re_causal 0.0228 /// teacc 98.89 lr 0.00010000 |
| Epoch 144, weight, value: tensor([[-0.0450, -0.0708, -0.0456, ..., -0.0917, -0.0038, -0.0215], |
| [-0.0414, 0.0262, -0.1243, ..., 0.0500, 0.0585, -0.1197], |
| [ 0.0548, 0.0012, -0.0511, ..., -0.0470, -0.0299, 0.0174], |
| ..., |
| [ 0.0485, -0.0460, 0.0019, ..., 0.0228, -0.0327, 0.0282], |
| [-0.0336, -0.0314, -0.0640, ..., -0.0140, -0.0252, -0.0596], |
| [-0.1108, -0.1727, 0.0348, ..., -0.0524, 0.0231, 0.0364]], |
| device='cuda:0'), grad: tensor([[-6.2678e-07, 1.8626e-09, 2.8498e-07, ..., 2.3283e-08, |
| 1.2573e-07, 0.0000e+00], |
| [ 6.0070e-07, -8.2888e-07, 1.0720e-06, ..., 8.8476e-08, |
| -6.0052e-06, 0.0000e+00], |
| [ 6.2361e-06, 5.9884e-07, 5.0385e-07, ..., 3.3528e-08, |
| 4.2655e-06, 0.0000e+00], |
| ..., |
| [ 1.3024e-05, 5.5879e-08, -4.5568e-05, ..., -1.4752e-05, |
| 4.6939e-06, 0.0000e+00], |
| [ 2.4572e-05, 1.3877e-07, -1.7630e-06, ..., 2.9802e-08, |
| 1.2793e-05, 0.0000e+00], |
| [ 1.3085e-06, 1.1176e-08, 3.5018e-06, ..., 1.3504e-07, |
| -2.5518e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 144, bias, value: tensor([-0.0042, 0.0249, 0.0056, 0.0021, 0.0100, -0.0005, 0.0199, -0.0144, |
| 0.0132, -0.0019], device='cuda:0'), grad: tensor([-4.2543e-06, -1.4521e-05, 3.1054e-05, -7.2896e-05, 2.6584e-04, |
| 1.2696e-05, -7.4983e-05, -2.7537e-04, 1.1462e-04, 1.8358e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 143, time 217.66, cls_loss 0.0027 cls_loss_mapping 0.0044 cls_loss_causal 0.5373 re_mapping 0.0075 re_causal 0.0226 /// teacc 98.84 lr 0.00010000 |
| Epoch 145, weight, value: tensor([[-0.0444, -0.0713, -0.0461, ..., -0.0917, -0.0041, -0.0215], |
| [-0.0419, 0.0264, -0.1247, ..., 0.0500, 0.0575, -0.1198], |
| [ 0.0551, 0.0011, -0.0513, ..., -0.0470, -0.0301, 0.0174], |
| ..., |
| [ 0.0489, -0.0460, 0.0010, ..., 0.0229, -0.0309, 0.0276], |
| [-0.0344, -0.0313, -0.0642, ..., -0.0140, -0.0254, -0.0596], |
| [-0.1114, -0.1728, 0.0348, ..., -0.0524, 0.0231, 0.0364]], |
| device='cuda:0'), grad: tensor([[-4.3306e-07, 2.2352e-08, 1.2293e-07, ..., 0.0000e+00, |
| 1.9372e-07, 0.0000e+00], |
| [ 2.6915e-07, -2.9523e-06, 3.0082e-07, ..., 0.0000e+00, |
| -5.8301e-06, 0.0000e+00], |
| [ 3.0547e-06, 2.6356e-07, 1.5274e-07, ..., 0.0000e+00, |
| 2.2277e-06, 0.0000e+00], |
| ..., |
| [-1.4156e-05, 8.3819e-08, 7.1712e-08, ..., 0.0000e+00, |
| -3.4869e-06, 0.0000e+00], |
| [ 2.6524e-06, 2.1607e-06, 1.0151e-07, ..., 0.0000e+00, |
| 5.1558e-06, 9.3132e-10], |
| [ 7.3574e-06, 1.8626e-08, 2.2724e-06, ..., 0.0000e+00, |
| 9.8627e-07, 9.3132e-10]], device='cuda:0') |
| Epoch 145, bias, value: tensor([-4.0226e-03, 2.4210e-02, 5.7599e-03, 1.7476e-03, 1.0116e-02, |
| 1.7832e-05, 1.9924e-02, -1.3649e-02, 1.2958e-02, -2.1085e-03], |
| device='cuda:0'), grad: tensor([-3.8035e-06, -1.1526e-05, 1.4603e-05, 5.5581e-06, -5.6103e-06, |
| -2.2613e-06, -2.2873e-06, -4.0263e-05, 1.9819e-05, 2.5690e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 144, time 217.57, cls_loss 0.0024 cls_loss_mapping 0.0042 cls_loss_causal 0.5331 re_mapping 0.0073 re_causal 0.0216 /// teacc 98.89 lr 0.00010000 |
| Epoch 146, weight, value: tensor([[-0.0445, -0.0720, -0.0463, ..., -0.0917, -0.0044, -0.0215], |
| [-0.0419, 0.0265, -0.1249, ..., 0.0500, 0.0578, -0.1200], |
| [ 0.0554, 0.0008, -0.0515, ..., -0.0470, -0.0303, 0.0174], |
| ..., |
| [ 0.0488, -0.0460, 0.0014, ..., 0.0229, -0.0311, 0.0273], |
| [-0.0350, -0.0306, -0.0640, ..., -0.0140, -0.0262, -0.0597], |
| [-0.1121, -0.1734, 0.0342, ..., -0.0524, 0.0235, 0.0364]], |
| device='cuda:0'), grad: tensor([[ 2.8126e-07, 0.0000e+00, 1.6671e-07, ..., 0.0000e+00, |
| 1.1176e-07, 0.0000e+00], |
| [ 9.6112e-07, 0.0000e+00, 7.8324e-07, ..., 0.0000e+00, |
| -5.8673e-08, 0.0000e+00], |
| [-2.6673e-06, 0.0000e+00, 1.0710e-07, ..., 0.0000e+00, |
| 2.9430e-07, 0.0000e+00], |
| ..., |
| [ 9.1866e-06, 0.0000e+00, 1.7554e-05, ..., 0.0000e+00, |
| 4.7535e-06, 0.0000e+00], |
| [ 2.7195e-06, 0.0000e+00, 4.5449e-07, ..., 0.0000e+00, |
| 8.2236e-07, 0.0000e+00], |
| [ 1.2228e-06, 0.0000e+00, 4.8369e-05, ..., 0.0000e+00, |
| 7.8231e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 146, bias, value: tensor([-3.6336e-03, 2.4358e-02, 5.8552e-03, 1.8479e-03, 1.0120e-02, |
| 1.4250e-05, 2.0323e-02, -1.3637e-02, 1.2483e-02, -2.4146e-03], |
| device='cuda:0'), grad: tensor([-2.7902e-06, 3.3528e-06, -4.1872e-06, 3.2540e-06, -1.9431e-04, |
| -1.9789e-05, -1.9651e-06, 6.4611e-05, 8.3670e-06, 1.4329e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 145, time 217.47, cls_loss 0.0028 cls_loss_mapping 0.0051 cls_loss_causal 0.5687 re_mapping 0.0077 re_causal 0.0228 /// teacc 98.85 lr 0.00010000 |
| Epoch 147, weight, value: tensor([[-0.0445, -0.0724, -0.0466, ..., -0.0919, -0.0047, -0.0215], |
| [-0.0420, 0.0266, -0.1255, ..., 0.0500, 0.0585, -0.1201], |
| [ 0.0554, 0.0007, -0.0519, ..., -0.0470, -0.0305, 0.0174], |
| ..., |
| [ 0.0490, -0.0461, 0.0010, ..., 0.0228, -0.0315, 0.0276], |
| [-0.0355, -0.0295, -0.0640, ..., -0.0141, -0.0263, -0.0598], |
| [-0.1123, -0.1737, 0.0340, ..., -0.0524, 0.0241, 0.0364]], |
| device='cuda:0'), grad: tensor([[ 6.8918e-08, 0.0000e+00, 5.7649e-07, ..., 0.0000e+00, |
| 7.3947e-07, 2.0489e-07], |
| [ 1.0477e-06, -2.8871e-08, 5.6811e-08, ..., 0.0000e+00, |
| 1.5367e-06, 4.5728e-07], |
| [-3.5316e-06, 1.3970e-08, 3.2503e-07, ..., 0.0000e+00, |
| -2.1290e-06, 3.4459e-08], |
| ..., |
| [-7.1339e-07, 5.5879e-09, 4.7497e-08, ..., 0.0000e+00, |
| -9.7826e-06, -3.1497e-06], |
| [ 2.7083e-06, 3.7253e-09, 1.8626e-09, ..., 0.0000e+00, |
| 2.0787e-06, 3.5390e-08], |
| [ 2.0675e-07, 0.0000e+00, 1.0738e-06, ..., 0.0000e+00, |
| 6.0797e-06, 2.2184e-06]], device='cuda:0') |
| Epoch 147, bias, value: tensor([-0.0035, 0.0246, 0.0056, 0.0007, 0.0101, 0.0006, 0.0204, -0.0136, |
| 0.0126, -0.0024], device='cuda:0'), grad: tensor([ 4.9248e-06, 9.6038e-06, -8.0019e-06, 7.2643e-07, -3.5390e-08, |
| 7.0706e-06, -4.2841e-06, -5.3465e-05, 8.5384e-06, 3.4958e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 146, time 217.52, cls_loss 0.0028 cls_loss_mapping 0.0055 cls_loss_causal 0.5228 re_mapping 0.0077 re_causal 0.0230 /// teacc 98.90 lr 0.00010000 |
| Epoch 148, weight, value: tensor([[-0.0446, -0.0726, -0.0467, ..., -0.0919, -0.0049, -0.0215], |
| [-0.0424, 0.0266, -0.1268, ..., 0.0500, 0.0589, -0.1203], |
| [ 0.0559, 0.0009, -0.0518, ..., -0.0470, -0.0298, 0.0173], |
| ..., |
| [ 0.0491, -0.0462, 0.0009, ..., 0.0228, -0.0323, 0.0277], |
| [-0.0360, -0.0287, -0.0648, ..., -0.0141, -0.0271, -0.0599], |
| [-0.1123, -0.1738, 0.0340, ..., -0.0524, 0.0246, 0.0364]], |
| device='cuda:0'), grad: tensor([[ 6.2399e-08, 0.0000e+00, -9.7096e-05, ..., 0.0000e+00, |
| 1.6484e-07, 3.7253e-09], |
| [-7.2271e-07, 0.0000e+00, 1.0459e-06, ..., 0.0000e+00, |
| -1.3262e-06, 9.3132e-10], |
| [ 1.0934e-06, 0.0000e+00, 5.0180e-06, ..., 0.0000e+00, |
| 1.6801e-06, 9.3132e-10], |
| ..., |
| [-1.6009e-06, 0.0000e+00, 3.1572e-07, ..., 0.0000e+00, |
| -9.3132e-08, 0.0000e+00], |
| [ 1.2759e-07, 0.0000e+00, -4.2439e-05, ..., 0.0000e+00, |
| -2.7016e-05, 2.7940e-09], |
| [ 7.0874e-07, 0.0000e+00, 8.7798e-05, ..., 0.0000e+00, |
| 2.2233e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 148, bias, value: tensor([-0.0034, 0.0245, 0.0062, 0.0007, 0.0101, 0.0008, 0.0203, -0.0139, |
| 0.0123, -0.0022], device='cuda:0'), grad: tensor([-5.8651e-04, -1.1176e-08, 3.8773e-05, 1.9059e-05, 5.6326e-06, |
| 3.4153e-05, 2.1720e-04, -1.6810e-06, -2.0349e-04, 4.7636e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 147---------------------------------------------------- |
| epoch 147, time 218.16, cls_loss 0.0030 cls_loss_mapping 0.0047 cls_loss_causal 0.5137 re_mapping 0.0077 re_causal 0.0221 /// teacc 98.94 lr 0.00010000 |
| Epoch 149, weight, value: tensor([[-0.0451, -0.0729, -0.0466, ..., -0.0920, -0.0053, -0.0216], |
| [-0.0416, 0.0270, -0.1280, ..., 0.0502, 0.0595, -0.1204], |
| [ 0.0555, 0.0004, -0.0523, ..., -0.0478, -0.0306, 0.0174], |
| ..., |
| [ 0.0492, -0.0463, 0.0007, ..., 0.0228, -0.0320, 0.0277], |
| [-0.0368, -0.0287, -0.0642, ..., -0.0141, -0.0265, -0.0599], |
| [-0.1132, -0.1739, 0.0333, ..., -0.0525, 0.0244, 0.0364]], |
| device='cuda:0'), grad: tensor([[ 4.1444e-08, 0.0000e+00, 1.2247e-07, ..., 0.0000e+00, |
| 4.8950e-06, 2.1607e-07], |
| [-6.0573e-06, 0.0000e+00, 1.4761e-07, ..., 0.0000e+00, |
| -7.9051e-06, 1.8626e-09], |
| [ 1.8906e-07, 0.0000e+00, 3.5390e-08, ..., 0.0000e+00, |
| 2.4866e-07, 0.0000e+00], |
| ..., |
| [ 5.6587e-06, 0.0000e+00, 2.9802e-07, ..., 0.0000e+00, |
| 7.3090e-06, 4.6566e-10], |
| [ 2.9337e-07, 0.0000e+00, 5.3644e-07, ..., 0.0000e+00, |
| 1.0747e-06, 3.2596e-08], |
| [ 1.3970e-07, 0.0000e+00, 1.0170e-06, ..., 0.0000e+00, |
| -2.0061e-06, 4.6566e-10]], device='cuda:0') |
| Epoch 149, bias, value: tensor([-0.0035, 0.0253, 0.0051, -0.0010, 0.0101, 0.0020, 0.0202, -0.0137, |
| 0.0128, -0.0027], device='cuda:0'), grad: tensor([ 2.5809e-05, -2.0534e-05, 8.4052e-07, 1.2070e-06, 2.6580e-06, |
| 3.0659e-06, -3.2723e-05, 2.0325e-05, 5.8934e-06, -6.6124e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 148, time 217.82, cls_loss 0.0030 cls_loss_mapping 0.0062 cls_loss_causal 0.5130 re_mapping 0.0080 re_causal 0.0223 /// teacc 98.84 lr 0.00010000 |
| Epoch 150, weight, value: tensor([[-0.0456, -0.0736, -0.0468, ..., -0.0921, -0.0058, -0.0216], |
| [-0.0415, 0.0272, -0.1290, ..., 0.0505, 0.0600, -0.1207], |
| [ 0.0559, 0.0014, -0.0526, ..., -0.0492, -0.0309, 0.0180], |
| ..., |
| [ 0.0490, -0.0470, 0.0006, ..., 0.0228, -0.0323, 0.0276], |
| [-0.0379, -0.0287, -0.0651, ..., -0.0141, -0.0267, -0.0600], |
| [-0.1134, -0.1741, 0.0333, ..., -0.0525, 0.0251, 0.0365]], |
| device='cuda:0'), grad: tensor([[ 9.3598e-08, 0.0000e+00, 2.8405e-08, ..., 0.0000e+00, |
| 1.2387e-07, 4.6566e-10], |
| [ 1.7229e-08, 0.0000e+00, 6.5658e-08, ..., 0.0000e+00, |
| -2.5313e-06, 4.6566e-10], |
| [ 1.1213e-05, 0.0000e+00, 1.3392e-06, ..., 0.0000e+00, |
| 2.0443e-07, -2.3283e-09], |
| ..., |
| [-1.2457e-05, 0.0000e+00, -1.4557e-06, ..., 0.0000e+00, |
| 1.4324e-06, 4.6566e-10], |
| [ 7.2131e-07, 0.0000e+00, 1.0477e-07, ..., 0.0000e+00, |
| 2.4401e-07, 9.3132e-10], |
| [ 2.0862e-07, 0.0000e+00, 3.4040e-07, ..., 0.0000e+00, |
| 8.8476e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 150, bias, value: tensor([-0.0036, 0.0254, 0.0049, -0.0022, 0.0102, 0.0039, 0.0198, -0.0137, |
| 0.0123, -0.0026], device='cuda:0'), grad: tensor([ 5.7230e-07, -7.2159e-06, 1.9610e-05, -1.2107e-07, 7.1526e-07, |
| 9.0711e-07, -3.8138e-07, -1.6734e-05, 1.9874e-06, 6.5379e-07], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 149, time 217.74, cls_loss 0.0027 cls_loss_mapping 0.0036 cls_loss_causal 0.5477 re_mapping 0.0077 re_causal 0.0229 /// teacc 98.92 lr 0.00010000 |
| Epoch 151, weight, value: tensor([[-0.0455, -0.0740, -0.0453, ..., -0.0921, -0.0061, -0.0216], |
| [-0.0412, 0.0272, -0.1294, ..., 0.0505, 0.0605, -0.1209], |
| [ 0.0557, 0.0012, -0.0527, ..., -0.0492, -0.0311, 0.0180], |
| ..., |
| [ 0.0488, -0.0470, 0.0019, ..., 0.0228, -0.0343, 0.0275], |
| [-0.0375, -0.0283, -0.0651, ..., -0.0141, -0.0265, -0.0600], |
| [-0.1138, -0.1742, 0.0332, ..., -0.0525, 0.0268, 0.0365]], |
| device='cuda:0'), grad: tensor([[ 1.5320e-07, 0.0000e+00, -1.2154e-07, ..., 0.0000e+00, |
| 8.1956e-08, 0.0000e+00], |
| [ 6.7987e-08, 0.0000e+00, 1.1390e-06, ..., 0.0000e+00, |
| 8.0233e-07, 0.0000e+00], |
| [-1.0170e-06, 0.0000e+00, -4.3772e-08, ..., 0.0000e+00, |
| 1.0058e-07, -9.3132e-10], |
| ..., |
| [ 4.6147e-07, 0.0000e+00, 3.6974e-07, ..., 0.0000e+00, |
| 4.5588e-07, 0.0000e+00], |
| [ 1.5134e-07, 0.0000e+00, 5.9884e-07, ..., 0.0000e+00, |
| 5.1549e-07, 0.0000e+00], |
| [ 3.4226e-07, 0.0000e+00, -3.8072e-06, ..., 0.0000e+00, |
| -5.7556e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 151, bias, value: tensor([-0.0030, 0.0257, 0.0046, -0.0019, 0.0100, 0.0037, 0.0193, -0.0145, |
| 0.0127, -0.0022], device='cuda:0'), grad: tensor([-5.2117e-06, 9.8422e-06, -8.2934e-07, 2.6803e-06, 2.2963e-05, |
| 4.6901e-06, 6.5845e-07, 3.6135e-06, 2.5406e-06, -4.0919e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 150, time 217.61, cls_loss 0.0034 cls_loss_mapping 0.0058 cls_loss_causal 0.5523 re_mapping 0.0072 re_causal 0.0217 /// teacc 98.87 lr 0.00010000 |
| Epoch 152, weight, value: tensor([[-0.0445, -0.0740, -0.0457, ..., -0.0921, -0.0081, -0.0241], |
| [-0.0425, 0.0272, -0.1311, ..., 0.0505, 0.0605, -0.1214], |
| [ 0.0560, 0.0012, -0.0515, ..., -0.0492, -0.0312, 0.0180], |
| ..., |
| [ 0.0498, -0.0470, 0.0016, ..., 0.0228, -0.0335, 0.0272], |
| [-0.0373, -0.0284, -0.0660, ..., -0.0141, -0.0267, -0.0603], |
| [-0.1154, -0.1743, 0.0332, ..., -0.0525, 0.0274, 0.0390]], |
| device='cuda:0'), grad: tensor([[ 5.5321e-07, 0.0000e+00, 1.2293e-07, ..., 0.0000e+00, |
| 8.2888e-08, 1.8626e-09], |
| [ 1.1642e-06, 0.0000e+00, 1.1735e-07, ..., 0.0000e+00, |
| -6.0163e-07, 1.8626e-09], |
| [ 1.7092e-05, 0.0000e+00, 4.2506e-06, ..., 0.0000e+00, |
| 1.6298e-07, -8.3819e-09], |
| ..., |
| [ 5.6289e-06, 0.0000e+00, 1.4324e-06, ..., 0.0000e+00, |
| 1.8068e-07, 9.3132e-10], |
| [ 4.2468e-06, 0.0000e+00, 6.9663e-06, ..., 0.0000e+00, |
| 2.7977e-06, 0.0000e+00], |
| [ 1.0999e-06, 0.0000e+00, 8.0373e-07, ..., 0.0000e+00, |
| -3.2037e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 152, bias, value: tensor([-0.0042, 0.0249, 0.0047, -0.0017, 0.0101, 0.0030, 0.0206, -0.0134, |
| 0.0126, -0.0022], device='cuda:0'), grad: tensor([-5.5023e-06, 2.7940e-09, 1.6540e-05, -2.8536e-05, -6.6217e-07, |
| -3.9376e-06, -3.1441e-05, 3.6415e-06, 4.4644e-05, 5.1484e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 151, time 217.86, cls_loss 0.0033 cls_loss_mapping 0.0058 cls_loss_causal 0.5633 re_mapping 0.0074 re_causal 0.0217 /// teacc 98.83 lr 0.00010000 |
| Epoch 153, weight, value: tensor([[-0.0442, -0.0743, -0.0464, ..., -0.0921, -0.0063, -0.0247], |
| [-0.0426, 0.0273, -0.1324, ..., 0.0505, 0.0607, -0.1218], |
| [ 0.0571, 0.0012, -0.0523, ..., -0.0493, -0.0312, 0.0172], |
| ..., |
| [ 0.0493, -0.0470, 0.0018, ..., 0.0228, -0.0339, 0.0266], |
| [-0.0379, -0.0284, -0.0657, ..., -0.0142, -0.0270, -0.0605], |
| [-0.1168, -0.1744, 0.0330, ..., -0.0525, 0.0267, 0.0395]], |
| device='cuda:0'), grad: tensor([[-8.3819e-09, 0.0000e+00, 4.0047e-08, ..., 0.0000e+00, |
| 8.4750e-08, 2.1420e-08], |
| [ 2.3209e-06, 0.0000e+00, 1.2601e-06, ..., 1.8626e-09, |
| -1.3690e-07, 2.1420e-08], |
| [-1.2904e-05, 0.0000e+00, 3.6322e-07, ..., 1.8626e-09, |
| 1.9278e-07, -2.5146e-08], |
| ..., |
| [-4.9453e-07, 0.0000e+00, 7.5437e-07, ..., 1.8626e-09, |
| 7.5996e-07, 1.8813e-07], |
| [-4.2841e-08, 0.0000e+00, 6.7875e-06, ..., 1.8626e-09, |
| 1.3355e-06, 1.1083e-07], |
| [ 3.1386e-07, 0.0000e+00, -4.7591e-07, ..., 9.3132e-10, |
| -4.3288e-06, -7.0035e-07]], device='cuda:0') |
| Epoch 153, bias, value: tensor([-0.0027, 0.0247, 0.0057, -0.0012, 0.0097, 0.0036, 0.0201, -0.0138, |
| 0.0126, -0.0031], device='cuda:0'), grad: tensor([-2.0005e-06, 9.4622e-06, -1.2368e-05, 1.4052e-05, -5.3942e-06, |
| -1.1943e-05, 3.2075e-06, 4.1127e-06, 2.1353e-05, -2.0593e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 152, time 218.05, cls_loss 0.0026 cls_loss_mapping 0.0047 cls_loss_causal 0.5490 re_mapping 0.0073 re_causal 0.0216 /// teacc 98.87 lr 0.00010000 |
| Epoch 154, weight, value: tensor([[-0.0443, -0.0746, -0.0463, ..., -0.0921, -0.0066, -0.0247], |
| [-0.0428, 0.0278, -0.1330, ..., 0.0505, 0.0608, -0.1222], |
| [ 0.0596, 0.0007, -0.0517, ..., -0.0493, -0.0312, 0.0171], |
| ..., |
| [ 0.0474, -0.0472, 0.0017, ..., 0.0229, -0.0339, 0.0264], |
| [-0.0384, -0.0285, -0.0663, ..., -0.0142, -0.0270, -0.0610], |
| [-0.1173, -0.1745, 0.0331, ..., -0.0526, 0.0269, 0.0395]], |
| device='cuda:0'), grad: tensor([[ 3.0547e-07, 0.0000e+00, 1.3504e-07, ..., 9.3132e-10, |
| 5.8450e-06, 1.5926e-06], |
| [-4.0978e-08, 0.0000e+00, 4.6566e-09, ..., 5.5879e-09, |
| -1.5795e-06, 2.6077e-08], |
| [-7.4506e-09, 0.0000e+00, 5.5879e-09, ..., 2.7940e-09, |
| 9.6764e-07, 5.3085e-08], |
| ..., |
| [-2.9802e-07, 0.0000e+00, 0.0000e+00, ..., -1.3039e-08, |
| 3.2224e-07, 1.8626e-09], |
| [ 4.6473e-07, 0.0000e+00, 5.5879e-09, ..., 0.0000e+00, |
| 3.2224e-07, 5.0291e-08], |
| [ 2.9523e-07, 0.0000e+00, 1.8626e-09, ..., 1.8626e-09, |
| 4.9360e-08, 1.3970e-08]], device='cuda:0') |
| Epoch 154, bias, value: tensor([-0.0023, 0.0246, 0.0080, -0.0016, 0.0094, 0.0042, 0.0197, -0.0155, |
| 0.0126, -0.0033], device='cuda:0'), grad: tensor([ 3.6716e-05, -3.2857e-06, 1.7313e-06, -2.9318e-06, 3.0212e-06, |
| 4.8950e-06, -4.3869e-05, -9.3132e-08, 2.6077e-06, 1.1381e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 153, time 217.75, cls_loss 0.0029 cls_loss_mapping 0.0044 cls_loss_causal 0.5572 re_mapping 0.0070 re_causal 0.0218 /// teacc 98.85 lr 0.00010000 |
| Epoch 155, weight, value: tensor([[-0.0440, -0.0747, -0.0463, ..., -0.0923, -0.0064, -0.0247], |
| [-0.0425, 0.0278, -0.1318, ..., 0.0505, 0.0614, -0.1225], |
| [ 0.0598, 0.0006, -0.0519, ..., -0.0494, -0.0320, 0.0171], |
| ..., |
| [ 0.0472, -0.0472, 0.0020, ..., 0.0238, -0.0336, 0.0259], |
| [-0.0389, -0.0284, -0.0666, ..., -0.0142, -0.0272, -0.0611], |
| [-0.1190, -0.1746, 0.0330, ..., -0.0527, 0.0269, 0.0397]], |
| device='cuda:0'), grad: tensor([[ 8.6240e-07, 0.0000e+00, 3.3528e-08, ..., 0.0000e+00, |
| 9.9931e-07, 0.0000e+00], |
| [ 4.8988e-07, 1.8626e-09, 1.8440e-07, ..., 0.0000e+00, |
| -1.5181e-07, 2.7940e-09], |
| [ 4.7684e-05, -3.7253e-09, 4.3772e-08, ..., 0.0000e+00, |
| 1.3970e-08, 9.3132e-10], |
| ..., |
| [-4.5076e-06, 1.8626e-09, 5.5321e-07, ..., 0.0000e+00, |
| 7.5437e-08, 8.3819e-09], |
| [-4.6343e-05, 0.0000e+00, 2.0768e-07, ..., 0.0000e+00, |
| 1.2200e-07, 0.0000e+00], |
| [ 2.4773e-07, 0.0000e+00, -1.4901e-07, ..., 0.0000e+00, |
| -5.4482e-07, 2.7940e-09]], device='cuda:0') |
| Epoch 155, bias, value: tensor([-0.0017, 0.0251, 0.0078, -0.0010, 0.0092, 0.0033, 0.0202, -0.0153, |
| 0.0124, -0.0038], device='cuda:0'), grad: tensor([ 6.0201e-06, 1.1977e-06, 1.4019e-04, 4.9621e-06, 3.3155e-07, |
| 8.0187e-07, -4.4703e-06, -3.9376e-06, -1.4472e-04, -4.8988e-07], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 154, time 217.72, cls_loss 0.0034 cls_loss_mapping 0.0058 cls_loss_causal 0.5530 re_mapping 0.0069 re_causal 0.0213 /// teacc 98.91 lr 0.00010000 |
| Epoch 156, weight, value: tensor([[-0.0441, -0.0755, -0.0469, ..., -0.0925, -0.0068, -0.0247], |
| [-0.0421, 0.0279, -0.1352, ..., 0.0510, 0.0634, -0.1232], |
| [ 0.0586, -0.0022, -0.0530, ..., -0.0507, -0.0325, 0.0169], |
| ..., |
| [ 0.0483, -0.0449, 0.0015, ..., 0.0237, -0.0354, 0.0235], |
| [-0.0393, -0.0283, -0.0660, ..., -0.0143, -0.0255, -0.0614], |
| [-0.1201, -0.1748, 0.0330, ..., -0.0529, 0.0270, 0.0398]], |
| device='cuda:0'), grad: tensor([[ 6.5472e-07, 1.8626e-09, 3.2876e-07, ..., 0.0000e+00, |
| -2.4699e-06, 9.3132e-10], |
| [ 1.1828e-07, 5.5879e-09, 9.9652e-08, ..., 0.0000e+00, |
| -1.4529e-07, 0.0000e+00], |
| [-3.1412e-05, 3.9116e-08, -2.3603e-05, ..., 0.0000e+00, |
| 2.4866e-07, -2.7940e-09], |
| ..., |
| [ 1.5367e-07, 1.7695e-08, 1.7602e-07, ..., 0.0000e+00, |
| 2.9337e-07, 9.3132e-10], |
| [ 2.7269e-05, -1.5646e-07, 2.1785e-05, ..., 0.0000e+00, |
| 5.4725e-06, 9.3132e-10], |
| [ 4.0978e-07, 9.3132e-10, -8.5961e-07, ..., 0.0000e+00, |
| -4.4852e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 156, bias, value: tensor([-0.0023, 0.0256, 0.0065, -0.0002, 0.0093, 0.0019, 0.0202, -0.0149, |
| 0.0141, -0.0037], device='cuda:0'), grad: tensor([-1.4089e-05, 4.7125e-07, -9.9063e-05, 6.6571e-06, 5.4203e-06, |
| 2.8536e-06, -9.4809e-07, 1.4780e-06, 1.1295e-04, -1.5602e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 155, time 217.80, cls_loss 0.0030 cls_loss_mapping 0.0058 cls_loss_causal 0.5448 re_mapping 0.0072 re_causal 0.0215 /// teacc 98.93 lr 0.00010000 |
| Epoch 157, weight, value: tensor([[-0.0442, -0.0758, -0.0471, ..., -0.0928, -0.0067, -0.0247], |
| [-0.0446, 0.0279, -0.1359, ..., 0.0509, 0.0622, -0.1236], |
| [ 0.0598, -0.0022, -0.0530, ..., -0.0508, -0.0304, 0.0169], |
| ..., |
| [ 0.0487, -0.0449, 0.0009, ..., 0.0237, -0.0353, 0.0232], |
| [-0.0398, -0.0279, -0.0662, ..., -0.0144, -0.0263, -0.0615], |
| [-0.1212, -0.1750, 0.0325, ..., -0.0528, 0.0269, 0.0398]], |
| device='cuda:0'), grad: tensor([[-8.3353e-07, 0.0000e+00, -2.5611e-07, ..., 0.0000e+00, |
| 2.4959e-07, 0.0000e+00], |
| [-2.0601e-06, 0.0000e+00, 1.2815e-06, ..., 0.0000e+00, |
| -7.8827e-06, 0.0000e+00], |
| [ 2.9802e-06, 0.0000e+00, 1.8394e-06, ..., 0.0000e+00, |
| 4.3586e-06, 0.0000e+00], |
| ..., |
| [ 8.6613e-07, 0.0000e+00, 9.4064e-08, ..., 0.0000e+00, |
| 4.1686e-06, 0.0000e+00], |
| [-4.0233e-06, 0.0000e+00, -5.5581e-06, ..., 0.0000e+00, |
| -3.3081e-06, 0.0000e+00], |
| [ 4.6287e-07, 0.0000e+00, 1.1818e-06, ..., 0.0000e+00, |
| 1.7416e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 157, bias, value: tensor([-0.0022, 0.0239, 0.0079, -0.0004, 0.0096, 0.0022, 0.0199, -0.0145, |
| 0.0135, -0.0039], device='cuda:0'), grad: tensor([-8.4639e-06, -2.0847e-05, 3.1441e-05, 1.3344e-05, 1.7360e-06, |
| 1.5423e-05, 2.3842e-06, 1.7628e-05, -6.0380e-05, 7.6443e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 156, time 217.43, cls_loss 0.0032 cls_loss_mapping 0.0045 cls_loss_causal 0.5750 re_mapping 0.0072 re_causal 0.0213 /// teacc 98.92 lr 0.00010000 |
| Epoch 158, weight, value: tensor([[-0.0445, -0.0761, -0.0478, ..., -0.0931, -0.0068, -0.0247], |
| [-0.0424, 0.0283, -0.1368, ..., 0.0509, 0.0646, -0.1236], |
| [ 0.0591, -0.0025, -0.0529, ..., -0.0508, -0.0307, 0.0169], |
| ..., |
| [ 0.0478, -0.0450, 0.0009, ..., 0.0237, -0.0373, 0.0232], |
| [-0.0399, -0.0279, -0.0667, ..., -0.0145, -0.0266, -0.0615], |
| [-0.1223, -0.1751, 0.0321, ..., -0.0528, 0.0263, 0.0398]], |
| device='cuda:0'), grad: tensor([[ 5.9605e-08, 8.3819e-09, 4.1444e-07, ..., 0.0000e+00, |
| 4.8429e-08, 0.0000e+00], |
| [-9.1735e-07, -1.6214e-06, 1.2435e-05, ..., 0.0000e+00, |
| -2.4028e-07, 0.0000e+00], |
| [ 1.1344e-06, 1.1222e-06, 2.9057e-07, ..., 0.0000e+00, |
| 9.4995e-08, 0.0000e+00], |
| ..., |
| [-9.2294e-07, 4.2841e-08, 8.0187e-07, ..., 0.0000e+00, |
| 1.9837e-07, 0.0000e+00], |
| [-4.9360e-08, 7.9162e-08, -1.8626e-08, ..., 0.0000e+00, |
| -8.0280e-07, 0.0000e+00], |
| [ 3.3807e-07, 1.8626e-09, 2.1353e-05, ..., 0.0000e+00, |
| 5.6997e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 158, bias, value: tensor([-0.0020, 0.0261, 0.0071, -0.0005, 0.0094, 0.0025, 0.0201, -0.0156, |
| 0.0134, -0.0044], device='cuda:0'), grad: tensor([-8.5160e-06, 2.6196e-05, 7.8380e-06, 6.6683e-07, -1.3053e-04, |
| 6.0834e-06, 3.9905e-05, 1.7984e-06, -4.0755e-06, 6.0737e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 157---------------------------------------------------- |
| epoch 157, time 218.21, cls_loss 0.0034 cls_loss_mapping 0.0053 cls_loss_causal 0.5586 re_mapping 0.0067 re_causal 0.0209 /// teacc 98.96 lr 0.00010000 |
| Epoch 159, weight, value: tensor([[-0.0452, -0.0764, -0.0487, ..., -0.0932, -0.0071, -0.0247], |
| [-0.0435, 0.0286, -0.1380, ..., 0.0509, 0.0646, -0.1237], |
| [ 0.0579, -0.0022, -0.0525, ..., -0.0509, -0.0309, 0.0169], |
| ..., |
| [ 0.0498, -0.0454, 0.0009, ..., 0.0237, -0.0373, 0.0232], |
| [-0.0412, -0.0280, -0.0670, ..., -0.0145, -0.0266, -0.0615], |
| [-0.1213, -0.1752, 0.0311, ..., -0.0528, 0.0275, 0.0398]], |
| device='cuda:0'), grad: tensor([[ 1.1735e-07, 0.0000e+00, 8.8476e-08, ..., 0.0000e+00, |
| 2.1327e-07, 0.0000e+00], |
| [ 3.7253e-08, 0.0000e+00, 1.9185e-07, ..., 0.0000e+00, |
| -2.2098e-05, 0.0000e+00], |
| [-7.9162e-08, 0.0000e+00, 3.6322e-08, ..., 0.0000e+00, |
| 1.7295e-06, 0.0000e+00], |
| ..., |
| [-5.5879e-08, 0.0000e+00, 4.5169e-07, ..., 0.0000e+00, |
| 1.1539e-06, 0.0000e+00], |
| [ 3.6508e-06, 0.0000e+00, 1.2517e-05, ..., 0.0000e+00, |
| 1.9237e-05, 0.0000e+00], |
| [ 2.6077e-08, 0.0000e+00, -2.4773e-07, ..., 0.0000e+00, |
| -1.0366e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 159, bias, value: tensor([-0.0024, 0.0252, 0.0058, -0.0021, 0.0101, 0.0026, 0.0212, -0.0139, |
| 0.0133, -0.0044], device='cuda:0'), grad: tensor([ 6.1747e-07, -6.8307e-05, 4.9807e-06, 7.4580e-06, 1.5581e-06, |
| -4.1008e-05, 2.7660e-07, 8.9258e-06, 9.3997e-05, -8.7172e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 158---------------------------------------------------- |
| epoch 158, time 218.38, cls_loss 0.0036 cls_loss_mapping 0.0043 cls_loss_causal 0.5536 re_mapping 0.0071 re_causal 0.0211 /// teacc 98.97 lr 0.00010000 |
| Epoch 160, weight, value: tensor([[-0.0455, -0.0767, -0.0508, ..., -0.0932, -0.0089, -0.0255], |
| [-0.0436, 0.0289, -0.1382, ..., 0.0509, 0.0648, -0.1238], |
| [ 0.0579, -0.0023, -0.0526, ..., -0.0509, -0.0311, 0.0169], |
| ..., |
| [ 0.0478, -0.0455, 0.0009, ..., 0.0237, -0.0372, 0.0231], |
| [-0.0415, -0.0277, -0.0684, ..., -0.0145, -0.0270, -0.0615], |
| [-0.1229, -0.1755, 0.0311, ..., -0.0528, 0.0283, 0.0406]], |
| device='cuda:0'), grad: tensor([[ 5.8077e-06, 0.0000e+00, 2.2762e-06, ..., 0.0000e+00, |
| 8.5216e-07, 0.0000e+00], |
| [ 2.7582e-05, 0.0000e+00, 9.1456e-07, ..., 0.0000e+00, |
| 6.4448e-07, 0.0000e+00], |
| [-9.2015e-06, 0.0000e+00, 3.0734e-08, ..., 0.0000e+00, |
| -2.0396e-07, 0.0000e+00], |
| ..., |
| [-5.7548e-05, 0.0000e+00, 3.1292e-06, ..., 0.0000e+00, |
| 1.1148e-06, 0.0000e+00], |
| [ 6.4597e-06, 0.0000e+00, 2.8014e-06, ..., 0.0000e+00, |
| 3.8520e-06, 0.0000e+00], |
| [ 2.0847e-05, 0.0000e+00, 2.8461e-06, ..., 0.0000e+00, |
| -1.0237e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 160, bias, value: tensor([-0.0036, 0.0253, 0.0057, -0.0008, 0.0101, 0.0041, 0.0214, -0.0156, |
| 0.0128, -0.0043], device='cuda:0'), grad: tensor([-1.9324e-04, 5.5164e-05, 2.5973e-05, 8.0407e-05, 2.9713e-05, |
| 3.0939e-06, 4.7356e-05, -1.2052e-04, 5.6177e-05, 1.5780e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 159, time 217.57, cls_loss 0.0030 cls_loss_mapping 0.0042 cls_loss_causal 0.5736 re_mapping 0.0069 re_causal 0.0218 /// teacc 98.88 lr 0.00010000 |
| Epoch 161, weight, value: tensor([[-4.5857e-02, -7.6750e-02, -5.1790e-02, ..., -9.3160e-02, |
| -9.3876e-03, -2.5657e-02], |
| [-4.3888e-02, 2.8944e-02, -1.4007e-01, ..., 5.0875e-02, |
| 6.4678e-02, -1.2382e-01], |
| [ 5.7464e-02, -2.3744e-03, -5.2539e-02, ..., -5.0855e-02, |
| -3.1399e-02, 1.6877e-02], |
| ..., |
| [ 4.8401e-02, -4.5495e-02, 1.2720e-04, ..., 2.3744e-02, |
| -3.6993e-02, 2.3039e-02], |
| [-4.1970e-02, -2.7749e-02, -7.1137e-02, ..., -1.4514e-02, |
| -2.7987e-02, -6.1501e-02], |
| [-1.2395e-01, -1.7548e-01, 3.0484e-02, ..., -5.2835e-02, |
| 2.8416e-02, 4.0770e-02]], device='cuda:0'), grad: tensor([[ 1.0151e-06, 7.4506e-09, 9.3132e-08, ..., 0.0000e+00, |
| 2.9393e-06, 0.0000e+00], |
| [-4.5672e-06, -5.4762e-06, 3.1833e-06, ..., 0.0000e+00, |
| -2.2188e-05, 0.0000e+00], |
| [ 2.8685e-06, 4.0717e-06, 1.9930e-07, ..., 0.0000e+00, |
| 7.0706e-06, 0.0000e+00], |
| ..., |
| [ 4.5337e-06, 1.1846e-06, 2.2110e-06, ..., 0.0000e+00, |
| 1.1772e-05, 0.0000e+00], |
| [ 8.3074e-06, 8.3819e-09, 8.1770e-07, ..., 0.0000e+00, |
| 6.1467e-07, 0.0000e+00], |
| [-1.1269e-06, 2.7940e-09, 4.8503e-06, ..., 0.0000e+00, |
| -1.8068e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 161, bias, value: tensor([-0.0040, 0.0249, 0.0052, -0.0010, 0.0104, 0.0048, 0.0224, -0.0150, |
| 0.0115, -0.0046], device='cuda:0'), grad: tensor([ 2.6301e-06, -4.5180e-05, 2.3752e-05, -2.2769e-05, -3.8683e-05, |
| 3.5968e-06, 1.8431e-06, 4.0680e-05, 2.0668e-05, 1.3426e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 160---------------------------------------------------- |
| epoch 160, time 218.11, cls_loss 0.0026 cls_loss_mapping 0.0038 cls_loss_causal 0.5599 re_mapping 0.0070 re_causal 0.0206 /// teacc 98.99 lr 0.00010000 |
| Epoch 162, weight, value: tensor([[-4.5117e-02, -7.6832e-02, -5.1337e-02, ..., -9.3163e-02, |
| -9.3608e-03, -2.5657e-02], |
| [-4.4441e-02, 2.8980e-02, -1.4044e-01, ..., 5.0872e-02, |
| 6.4983e-02, -1.2392e-01], |
| [ 5.7667e-02, -2.3949e-03, -5.3077e-02, ..., -5.0859e-02, |
| -3.1665e-02, 1.6904e-02], |
| ..., |
| [ 4.8568e-02, -4.5502e-02, 1.5208e-04, ..., 2.3794e-02, |
| -3.7037e-02, 2.3019e-02], |
| [-4.2857e-02, -2.7810e-02, -7.1684e-02, ..., -1.4516e-02, |
| -2.8775e-02, -6.1504e-02], |
| [-1.2449e-01, -1.7553e-01, 3.0229e-02, ..., -5.2838e-02, |
| 2.8715e-02, 4.0770e-02]], device='cuda:0'), grad: tensor([[ 1.3709e-06, 0.0000e+00, 7.4506e-09, ..., 0.0000e+00, |
| 6.4168e-07, 0.0000e+00], |
| [ 5.2974e-06, 0.0000e+00, 2.9802e-08, ..., 0.0000e+00, |
| 5.8766e-07, 0.0000e+00], |
| [ 6.5845e-07, 0.0000e+00, 1.3039e-08, ..., 0.0000e+00, |
| 9.8813e-07, 0.0000e+00], |
| ..., |
| [-2.5891e-07, 0.0000e+00, 2.6077e-08, ..., 0.0000e+00, |
| 1.1558e-06, 0.0000e+00], |
| [ 3.4049e-06, 0.0000e+00, 2.4214e-08, ..., 0.0000e+00, |
| -4.3493e-07, 0.0000e+00], |
| [ 9.9838e-06, 0.0000e+00, 2.9802e-08, ..., 0.0000e+00, |
| 3.1516e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 162, bias, value: tensor([-0.0034, 0.0246, 0.0054, -0.0009, 0.0102, 0.0047, 0.0226, -0.0148, |
| 0.0106, -0.0046], device='cuda:0'), grad: tensor([ 4.8019e-06, 1.5199e-05, 5.3905e-06, -7.6711e-05, 5.9754e-06, |
| 1.5259e-05, -6.6385e-06, 1.4622e-07, -1.3849e-06, 3.7879e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 161, time 217.40, cls_loss 0.0030 cls_loss_mapping 0.0047 cls_loss_causal 0.5517 re_mapping 0.0069 re_causal 0.0209 /// teacc 98.89 lr 0.00010000 |
| Epoch 163, weight, value: tensor([[-4.5946e-02, -7.7002e-02, -5.2057e-02, ..., -9.3185e-02, |
| -9.5410e-03, -2.5657e-02], |
| [-4.3536e-02, 2.9060e-02, -1.4104e-01, ..., 5.0846e-02, |
| 6.6817e-02, -1.2396e-01], |
| [ 5.7733e-02, -2.4102e-03, -5.4186e-02, ..., -5.0873e-02, |
| -3.1812e-02, 1.6908e-02], |
| ..., |
| [ 4.8195e-02, -4.5548e-02, 1.2527e-04, ..., 2.3880e-02, |
| -3.9076e-02, 2.3016e-02], |
| [-4.4463e-02, -2.7671e-02, -7.3438e-02, ..., -1.4531e-02, |
| -2.8740e-02, -6.1509e-02], |
| [-1.2521e-01, -1.7565e-01, 3.0465e-02, ..., -5.2845e-02, |
| 2.8975e-02, 4.0770e-02]], device='cuda:0'), grad: tensor([[ 1.9837e-07, 0.0000e+00, 1.5832e-08, ..., 0.0000e+00, |
| 1.0896e-07, 0.0000e+00], |
| [ 5.7463e-07, 0.0000e+00, 1.0245e-08, ..., 0.0000e+00, |
| -5.2303e-06, 0.0000e+00], |
| [-2.1923e-06, 0.0000e+00, 3.7253e-09, ..., 0.0000e+00, |
| 1.2573e-07, 0.0000e+00], |
| ..., |
| [ 7.8231e-08, 0.0000e+00, 6.5193e-09, ..., 0.0000e+00, |
| 4.8149e-07, 0.0000e+00], |
| [ 5.2713e-07, 4.6566e-09, 1.9185e-07, ..., 0.0000e+00, |
| 4.2878e-06, 0.0000e+00], |
| [ 6.7987e-08, 0.0000e+00, 1.2387e-07, ..., 0.0000e+00, |
| 1.3132e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 163, bias, value: tensor([-0.0037, 0.0260, 0.0052, -0.0010, 0.0106, 0.0054, 0.0222, -0.0158, |
| 0.0100, -0.0045], device='cuda:0'), grad: tensor([ 2.0303e-07, -1.0371e-05, -2.8480e-06, 1.0291e-06, 9.9652e-08, |
| 4.3027e-07, -3.8184e-07, 1.2992e-06, 1.0043e-05, 4.8894e-07], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 162, time 217.49, cls_loss 0.0033 cls_loss_mapping 0.0041 cls_loss_causal 0.5250 re_mapping 0.0072 re_causal 0.0202 /// teacc 98.97 lr 0.00010000 |
| Epoch 164, weight, value: tensor([[-4.6336e-02, -7.7186e-02, -5.3196e-02, ..., -9.3215e-02, |
| -1.2330e-02, -2.8327e-02], |
| [-4.4323e-02, 2.9254e-02, -1.4132e-01, ..., 5.0783e-02, |
| 6.7174e-02, -1.2457e-01], |
| [ 5.7638e-02, -2.2848e-03, -5.4184e-02, ..., -5.0889e-02, |
| -3.2044e-02, 1.6707e-02], |
| ..., |
| [ 4.8958e-02, -4.5794e-02, -6.4190e-05, ..., 2.3987e-02, |
| -3.9013e-02, 2.2711e-02], |
| [-4.6292e-02, -2.7411e-02, -7.3815e-02, ..., -1.4554e-02, |
| -2.9622e-02, -6.1524e-02], |
| [-1.2614e-01, -1.7583e-01, 3.0799e-02, ..., -5.2888e-02, |
| 3.1135e-02, 4.3435e-02]], device='cuda:0'), grad: tensor([[ 2.2016e-06, 0.0000e+00, 9.3132e-10, ..., 0.0000e+00, |
| 1.4901e-08, 0.0000e+00], |
| [ 6.7592e-05, 5.5879e-09, 1.4901e-08, ..., 0.0000e+00, |
| -4.9639e-07, 0.0000e+00], |
| [-2.7880e-05, 0.0000e+00, 9.3132e-10, ..., 0.0000e+00, |
| 1.5832e-08, 0.0000e+00], |
| ..., |
| [-5.8919e-05, -8.3819e-09, 2.7008e-08, ..., 0.0000e+00, |
| 1.5460e-07, 0.0000e+00], |
| [ 4.2021e-06, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| 2.2911e-07, 0.0000e+00], |
| [ 1.5069e-06, 0.0000e+00, 2.4214e-08, ..., 0.0000e+00, |
| -2.3935e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 164, bias, value: tensor([-0.0054, 0.0258, 0.0051, -0.0011, 0.0105, 0.0055, 0.0230, -0.0152, |
| 0.0087, -0.0037], device='cuda:0'), grad: tensor([-4.3511e-06, 1.3995e-04, -4.7803e-05, 1.2711e-05, 2.4773e-06, |
| 5.5246e-06, 5.1185e-06, -1.2827e-04, 8.6054e-06, 6.0871e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 163---------------------------------------------------- |
| epoch 163, time 218.26, cls_loss 0.0022 cls_loss_mapping 0.0037 cls_loss_causal 0.5264 re_mapping 0.0069 re_causal 0.0208 /// teacc 99.05 lr 0.00010000 |
| Epoch 165, weight, value: tensor([[-0.0463, -0.0772, -0.0532, ..., -0.0934, -0.0123, -0.0283], |
| [-0.0447, 0.0293, -0.1418, ..., 0.0502, 0.0672, -0.1246], |
| [ 0.0577, -0.0023, -0.0545, ..., -0.0509, -0.0323, 0.0167], |
| ..., |
| [ 0.0492, -0.0458, -0.0002, ..., 0.0250, -0.0388, 0.0227], |
| [-0.0464, -0.0274, -0.0735, ..., -0.0146, -0.0297, -0.0615], |
| [-0.1270, -0.1759, 0.0305, ..., -0.0544, 0.0312, 0.0434]], |
| device='cuda:0'), grad: tensor([[ 5.1502e-07, 0.0000e+00, 5.2620e-07, ..., 8.3819e-09, |
| 2.3376e-07, 0.0000e+00], |
| [-8.2999e-06, 0.0000e+00, 4.5635e-06, ..., 4.0978e-08, |
| -1.5542e-05, 0.0000e+00], |
| [ 4.5151e-05, 0.0000e+00, 1.9968e-04, ..., 1.3970e-08, |
| 8.4657e-07, -2.7940e-09], |
| ..., |
| [ 6.4671e-06, 0.0000e+00, 3.3695e-06, ..., 5.5879e-08, |
| 1.2413e-05, 9.3132e-10], |
| [ 1.2284e-06, 0.0000e+00, 2.8443e-06, ..., 2.7940e-09, |
| 2.0675e-07, 0.0000e+00], |
| [ 1.8030e-06, 0.0000e+00, 7.8790e-07, ..., 8.7544e-08, |
| 1.3337e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 165, bias, value: tensor([-0.0048, 0.0256, 0.0050, -0.0013, 0.0105, 0.0055, 0.0232, -0.0149, |
| 0.0089, -0.0041], device='cuda:0'), grad: tensor([-3.7365e-06, -4.0889e-05, 3.4976e-04, -2.4773e-06, -3.7122e-04, |
| 1.1437e-06, 3.6266e-06, 4.4227e-05, 6.7800e-06, 1.3463e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 164, time 217.47, cls_loss 0.0029 cls_loss_mapping 0.0033 cls_loss_causal 0.5503 re_mapping 0.0070 re_causal 0.0209 /// teacc 98.99 lr 0.00010000 |
| Epoch 166, weight, value: tensor([[-0.0468, -0.0773, -0.0539, ..., -0.0943, -0.0124, -0.0283], |
| [-0.0449, 0.0296, -0.1427, ..., 0.0519, 0.0680, -0.1246], |
| [ 0.0571, -0.0024, -0.0553, ..., -0.0510, -0.0325, 0.0167], |
| ..., |
| [ 0.0497, -0.0459, -0.0006, ..., 0.0241, -0.0395, 0.0227], |
| [-0.0472, -0.0274, -0.0739, ..., -0.0151, -0.0301, -0.0615], |
| [-0.1277, -0.1759, 0.0296, ..., -0.0546, 0.0314, 0.0434]], |
| device='cuda:0'), grad: tensor([[ 1.6671e-07, 0.0000e+00, 6.5472e-07, ..., 9.5926e-08, |
| 3.2596e-07, 0.0000e+00], |
| [ 1.8766e-06, 0.0000e+00, 1.8179e-06, ..., 3.0175e-07, |
| 3.1665e-08, 0.0000e+00], |
| [-1.8999e-06, 0.0000e+00, -2.9244e-07, ..., 6.5193e-08, |
| 7.9162e-08, 0.0000e+00], |
| ..., |
| [-1.3970e-08, 0.0000e+00, 2.4438e-06, ..., 4.0047e-07, |
| 1.9837e-07, 0.0000e+00], |
| [ 7.9721e-07, 0.0000e+00, 4.8727e-06, ..., 5.8953e-07, |
| 3.4153e-05, 0.0000e+00], |
| [ 6.2063e-06, 0.0000e+00, 5.7429e-05, ..., 5.1595e-06, |
| 1.0207e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 166, bias, value: tensor([-0.0045, 0.0259, 0.0043, -0.0012, 0.0108, 0.0054, 0.0237, -0.0147, |
| 0.0084, -0.0046], device='cuda:0'), grad: tensor([ 3.2187e-06, 1.0543e-05, -3.2801e-06, 1.3709e-06, -1.9014e-04, |
| 4.7708e-04, -6.4754e-04, 3.2224e-06, 1.8990e-04, 1.5438e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 165, time 217.56, cls_loss 0.0024 cls_loss_mapping 0.0047 cls_loss_causal 0.5648 re_mapping 0.0069 re_causal 0.0207 /// teacc 98.99 lr 0.00010000 |
| Epoch 167, weight, value: tensor([[-0.0462, -0.0773, -0.0544, ..., -0.0945, -0.0121, -0.0283], |
| [-0.0451, 0.0296, -0.1436, ..., 0.0521, 0.0683, -0.1247], |
| [ 0.0571, -0.0024, -0.0551, ..., -0.0510, -0.0327, 0.0166], |
| ..., |
| [ 0.0499, -0.0459, -0.0008, ..., 0.0240, -0.0399, 0.0227], |
| [-0.0475, -0.0274, -0.0741, ..., -0.0153, -0.0304, -0.0615], |
| [-0.1284, -0.1759, 0.0292, ..., -0.0548, 0.0316, 0.0434]], |
| device='cuda:0'), grad: tensor([[ 2.4773e-07, 0.0000e+00, -1.5497e-06, ..., 1.1176e-08, |
| 1.8533e-07, 0.0000e+00], |
| [ 4.2375e-07, 0.0000e+00, 5.3179e-07, ..., 4.0978e-08, |
| 7.4506e-09, 0.0000e+00], |
| [-1.9297e-06, 0.0000e+00, 7.7207e-07, ..., 1.1176e-08, |
| 1.4529e-07, 0.0000e+00], |
| ..., |
| [-3.6228e-07, 0.0000e+00, 5.0385e-07, ..., 1.0058e-07, |
| 1.0068e-06, 0.0000e+00], |
| [ 5.1875e-07, 0.0000e+00, 5.8860e-07, ..., 9.3132e-09, |
| 5.0887e-06, 0.0000e+00], |
| [ 3.9581e-07, 0.0000e+00, 2.1905e-06, ..., 1.5553e-07, |
| -1.2159e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 167, bias, value: tensor([-0.0041, 0.0257, 0.0043, -0.0019, 0.0108, 0.0064, 0.0238, -0.0147, |
| 0.0082, -0.0046], device='cuda:0'), grad: tensor([-1.0706e-05, 2.3842e-06, -3.2131e-07, 2.1532e-06, 6.7279e-06, |
| 6.0461e-06, 4.2543e-06, 2.3358e-06, 2.2218e-05, -3.5197e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 166, time 217.68, cls_loss 0.0028 cls_loss_mapping 0.0046 cls_loss_causal 0.5522 re_mapping 0.0065 re_causal 0.0200 /// teacc 98.98 lr 0.00010000 |
| Epoch 168, weight, value: tensor([[-0.0465, -0.0773, -0.0549, ..., -0.0960, -0.0125, -0.0283], |
| [-0.0452, 0.0297, -0.1446, ..., 0.0515, 0.0685, -0.1249], |
| [ 0.0573, -0.0025, -0.0549, ..., -0.0512, -0.0331, 0.0166], |
| ..., |
| [ 0.0499, -0.0459, -0.0009, ..., 0.0262, -0.0399, 0.0225], |
| [-0.0479, -0.0275, -0.0744, ..., -0.0158, -0.0305, -0.0615], |
| [-0.1296, -0.1759, 0.0293, ..., -0.0575, 0.0319, 0.0435]], |
| device='cuda:0'), grad: tensor([[ 6.1467e-08, 2.1420e-08, 2.9150e-07, ..., 4.6566e-09, |
| 6.6534e-06, 0.0000e+00], |
| [ 1.5553e-07, 6.7987e-08, 1.4249e-07, ..., 2.9802e-08, |
| -8.5402e-07, 0.0000e+00], |
| [ 9.9093e-07, 6.5193e-09, 3.1199e-07, ..., 1.3970e-08, |
| 5.0850e-06, 0.0000e+00], |
| ..., |
| [ 2.1234e-06, 4.6566e-09, 1.7043e-07, ..., 6.0536e-08, |
| 1.1092e-06, 0.0000e+00], |
| [ 1.1921e-07, 4.6566e-09, -1.3048e-06, ..., 2.7940e-09, |
| -3.4750e-05, 0.0000e+00], |
| [ 1.1483e-06, 4.6566e-09, 1.8943e-06, ..., 2.1327e-07, |
| 3.7979e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 168, bias, value: tensor([-0.0040, 0.0256, 0.0044, -0.0017, 0.0106, 0.0061, 0.0245, -0.0145, |
| 0.0083, -0.0051], device='cuda:0'), grad: tensor([ 2.5034e-05, -6.6496e-07, 2.5898e-05, 1.3337e-05, -4.2319e-06, |
| -1.2740e-06, 6.6221e-05, 3.5781e-06, -1.4889e-04, 2.0996e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 167, time 217.51, cls_loss 0.0034 cls_loss_mapping 0.0045 cls_loss_causal 0.5187 re_mapping 0.0066 re_causal 0.0194 /// teacc 98.93 lr 0.00010000 |
| Epoch 169, weight, value: tensor([[-0.0469, -0.0774, -0.0553, ..., -0.0961, -0.0143, -0.0287], |
| [-0.0476, 0.0297, -0.1454, ..., 0.0514, 0.0674, -0.1250], |
| [ 0.0574, -0.0024, -0.0551, ..., -0.0512, -0.0335, 0.0165], |
| ..., |
| [ 0.0512, -0.0459, -0.0023, ..., 0.0264, -0.0385, 0.0224], |
| [-0.0487, -0.0275, -0.0748, ..., -0.0159, -0.0305, -0.0615], |
| [-0.1304, -0.1759, 0.0293, ..., -0.0578, 0.0336, 0.0438]], |
| device='cuda:0'), grad: tensor([[-2.4233e-06, 0.0000e+00, 1.2107e-08, ..., 0.0000e+00, |
| -1.5125e-05, 1.1176e-08], |
| [ 3.9786e-06, 3.7253e-09, 1.7695e-08, ..., 0.0000e+00, |
| -1.9893e-06, 2.7940e-09], |
| [-2.8834e-05, 9.3132e-10, -1.7695e-08, ..., 0.0000e+00, |
| 2.2445e-06, -1.0431e-07], |
| ..., |
| [ 3.2317e-07, -8.3819e-09, 2.6077e-08, ..., 0.0000e+00, |
| 1.1111e-06, 2.1420e-08], |
| [ 1.3247e-05, 9.3132e-10, 4.4703e-08, ..., 0.0000e+00, |
| 3.7625e-07, 7.4506e-09], |
| [ 4.9546e-06, 0.0000e+00, 1.7509e-07, ..., 0.0000e+00, |
| 1.2144e-05, 9.3132e-09]], device='cuda:0') |
| Epoch 169, bias, value: tensor([-0.0052, 0.0233, 0.0042, -0.0015, 0.0106, 0.0059, 0.0247, -0.0128, |
| 0.0085, -0.0044], device='cuda:0'), grad: tensor([-1.3173e-04, 6.2864e-07, -4.6223e-05, 1.4842e-05, 3.0678e-06, |
| 1.9725e-06, 8.0094e-06, 3.7253e-06, 3.2663e-05, 1.1313e-04], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 168, time 217.62, cls_loss 0.0027 cls_loss_mapping 0.0058 cls_loss_causal 0.5603 re_mapping 0.0067 re_causal 0.0196 /// teacc 98.95 lr 0.00010000 |
| Epoch 170, weight, value: tensor([[-0.0471, -0.0774, -0.0554, ..., -0.0962, -0.0147, -0.0288], |
| [-0.0479, 0.0298, -0.1472, ..., 0.0513, 0.0672, -0.1251], |
| [ 0.0579, -0.0025, -0.0546, ..., -0.0512, -0.0326, 0.0165], |
| ..., |
| [ 0.0510, -0.0460, -0.0052, ..., 0.0265, -0.0400, 0.0224], |
| [-0.0491, -0.0275, -0.0763, ..., -0.0159, -0.0311, -0.0616], |
| [-0.1292, -0.1760, 0.0293, ..., -0.0579, 0.0357, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 2.8312e-07, 0.0000e+00, 1.1176e-08, ..., 0.0000e+00, |
| -3.1665e-08, 0.0000e+00], |
| [ 4.8801e-06, 0.0000e+00, 2.1793e-07, ..., 0.0000e+00, |
| 1.3746e-06, 0.0000e+00], |
| [-4.7348e-06, 0.0000e+00, -2.1420e-08, ..., 0.0000e+00, |
| 1.7509e-07, 0.0000e+00], |
| ..., |
| [ 6.8955e-06, 0.0000e+00, 7.3574e-08, ..., 0.0000e+00, |
| 2.1793e-06, 0.0000e+00], |
| [ 1.3672e-06, 0.0000e+00, 3.8557e-07, ..., 0.0000e+00, |
| 2.7493e-06, 0.0000e+00], |
| [-1.1928e-05, 0.0000e+00, 4.7870e-06, ..., 0.0000e+00, |
| -5.4687e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 170, bias, value: tensor([-0.0050, 0.0230, 0.0049, -0.0016, 0.0106, 0.0060, 0.0244, -0.0135, |
| 0.0080, -0.0033], device='cuda:0'), grad: tensor([-8.3745e-05, 1.4491e-05, -1.7378e-06, 1.4357e-05, -9.7230e-06, |
| 2.9951e-06, -1.1288e-05, 3.9488e-05, 1.7539e-05, 1.7643e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 169, time 217.55, cls_loss 0.0025 cls_loss_mapping 0.0033 cls_loss_causal 0.5709 re_mapping 0.0070 re_causal 0.0216 /// teacc 98.99 lr 0.00010000 |
| Epoch 171, weight, value: tensor([[-0.0474, -0.0775, -0.0557, ..., -0.0965, -0.0148, -0.0288], |
| [-0.0480, 0.0298, -0.1475, ..., 0.0513, 0.0675, -0.1251], |
| [ 0.0585, -0.0025, -0.0548, ..., -0.0513, -0.0328, 0.0165], |
| ..., |
| [ 0.0508, -0.0460, -0.0056, ..., 0.0267, -0.0401, 0.0224], |
| [-0.0490, -0.0275, -0.0768, ..., -0.0161, -0.0308, -0.0616], |
| [-0.1308, -0.1760, 0.0286, ..., -0.0579, 0.0356, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 3.4459e-08, 0.0000e+00, 4.3772e-08, ..., 0.0000e+00, |
| 6.8918e-08, 0.0000e+00], |
| [ 3.3528e-08, 0.0000e+00, 3.2224e-07, ..., 0.0000e+00, |
| 3.8091e-07, 0.0000e+00], |
| [-2.6077e-08, 0.0000e+00, 2.0396e-07, ..., 0.0000e+00, |
| 4.0978e-07, 0.0000e+00], |
| ..., |
| [ 2.4214e-08, 0.0000e+00, 3.5856e-07, ..., 0.0000e+00, |
| 3.7905e-07, 0.0000e+00], |
| [ 3.1665e-08, 0.0000e+00, 3.1199e-07, ..., 0.0000e+00, |
| 5.0291e-08, 0.0000e+00], |
| [ 1.8626e-08, 0.0000e+00, 1.3508e-05, ..., 0.0000e+00, |
| -1.8384e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 171, bias, value: tensor([-0.0050, 0.0232, 0.0051, -0.0016, 0.0115, 0.0060, 0.0249, -0.0138, |
| 0.0082, -0.0040], device='cuda:0'), grad: tensor([-5.3085e-08, 2.4959e-06, 1.7369e-06, 3.7998e-07, -3.4571e-05, |
| -3.2876e-07, -2.5034e-06, 1.9297e-06, 1.6764e-08, 3.0965e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 170, time 217.37, cls_loss 0.0055 cls_loss_mapping 0.0079 cls_loss_causal 0.5374 re_mapping 0.0066 re_causal 0.0194 /// teacc 98.98 lr 0.00010000 |
| Epoch 172, weight, value: tensor([[-0.0486, -0.0777, -0.0559, ..., -0.0975, -0.0149, -0.0288], |
| [-0.0497, 0.0299, -0.1479, ..., 0.0565, 0.0661, -0.1251], |
| [ 0.0581, -0.0025, -0.0548, ..., -0.0515, -0.0343, 0.0165], |
| ..., |
| [ 0.0525, -0.0460, -0.0057, ..., 0.0216, -0.0383, 0.0224], |
| [-0.0497, -0.0268, -0.0770, ..., -0.0170, -0.0318, -0.0616], |
| [-0.1322, -0.1763, 0.0288, ..., -0.0568, 0.0354, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 4.7497e-08, 0.0000e+00, -4.0559e-07, ..., 8.8476e-09, |
| 1.0245e-07, 0.0000e+00], |
| [ 8.3726e-07, 0.0000e+00, 1.1362e-07, ..., 9.7789e-09, |
| -4.4936e-07, 0.0000e+00], |
| [-1.4361e-06, 0.0000e+00, 1.1688e-07, ..., 2.1886e-08, |
| 9.0338e-08, 0.0000e+00], |
| ..., |
| [ 6.9663e-07, 0.0000e+00, 1.1921e-07, ..., 5.4482e-08, |
| 2.4028e-07, 0.0000e+00], |
| [ 1.2852e-07, 0.0000e+00, 1.9511e-07, ..., 1.9092e-08, |
| 3.2363e-07, 0.0000e+00], |
| [ 4.5681e-07, 0.0000e+00, 8.3912e-07, ..., 3.1665e-08, |
| -3.1292e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 172, bias, value: tensor([-0.0048, 0.0222, 0.0039, -0.0015, 0.0110, 0.0055, 0.0248, -0.0124, |
| 0.0078, -0.0041], device='cuda:0'), grad: tensor([-3.6210e-06, 3.9814e-07, -1.5739e-06, 9.3654e-06, -1.7276e-07, |
| -7.1637e-06, -2.6654e-06, 1.4063e-06, 1.7658e-06, 2.2445e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 171, time 217.23, cls_loss 0.0025 cls_loss_mapping 0.0040 cls_loss_causal 0.5282 re_mapping 0.0066 re_causal 0.0205 /// teacc 98.98 lr 0.00010000 |
| Epoch 173, weight, value: tensor([[-0.0492, -0.0784, -0.0559, ..., -0.0977, -0.0150, -0.0288], |
| [-0.0500, 0.0305, -0.1486, ..., 0.0564, 0.0662, -0.1251], |
| [ 0.0576, -0.0027, -0.0546, ..., -0.0516, -0.0349, 0.0165], |
| ..., |
| [ 0.0531, -0.0461, -0.0057, ..., 0.0218, -0.0383, 0.0224], |
| [-0.0500, -0.0266, -0.0769, ..., -0.0171, -0.0315, -0.0616], |
| [-0.1324, -0.1766, 0.0287, ..., -0.0576, 0.0360, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 3.9581e-08, 0.0000e+00, -1.8207e-07, ..., 0.0000e+00, |
| -4.6846e-07, 0.0000e+00], |
| [ 4.2142e-07, 1.1176e-08, 7.5437e-08, ..., 0.0000e+00, |
| 2.0023e-08, 0.0000e+00], |
| [ 1.2061e-07, 3.9581e-08, 9.2201e-08, ..., 0.0000e+00, |
| 1.6298e-08, 0.0000e+00], |
| ..., |
| [ 2.5146e-06, -5.5879e-08, 6.0070e-08, ..., 0.0000e+00, |
| 1.0878e-06, 0.0000e+00], |
| [ 2.6729e-07, 2.3283e-09, -6.8499e-07, ..., 0.0000e+00, |
| 9.3132e-09, 0.0000e+00], |
| [ 8.3074e-07, 0.0000e+00, 2.3330e-07, ..., 0.0000e+00, |
| 1.4761e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 173, bias, value: tensor([-0.0048, 0.0220, 0.0034, -0.0016, 0.0107, 0.0057, 0.0240, -0.0121, |
| 0.0081, -0.0040], device='cuda:0'), grad: tensor([-6.1929e-05, 9.0478e-07, 5.3085e-07, -4.6268e-06, -1.6671e-07, |
| 7.5735e-06, 5.0366e-05, 3.3751e-06, 2.9197e-07, 3.5688e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 172, time 217.41, cls_loss 0.0021 cls_loss_mapping 0.0035 cls_loss_causal 0.5382 re_mapping 0.0066 re_causal 0.0206 /// teacc 98.99 lr 0.00010000 |
| Epoch 174, weight, value: tensor([[-0.0498, -0.0789, -0.0559, ..., -0.0978, -0.0150, -0.0288], |
| [-0.0502, 0.0306, -0.1494, ..., 0.0565, 0.0661, -0.1251], |
| [ 0.0578, -0.0027, -0.0547, ..., -0.0516, -0.0348, 0.0165], |
| ..., |
| [ 0.0532, -0.0461, -0.0061, ..., 0.0218, -0.0384, 0.0224], |
| [-0.0506, -0.0267, -0.0774, ..., -0.0171, -0.0316, -0.0616], |
| [-0.1330, -0.1775, 0.0287, ..., -0.0577, 0.0363, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 1.5367e-07, 0.0000e+00, 4.1444e-08, ..., 4.6566e-10, |
| 6.8452e-08, 0.0000e+00], |
| [ 4.4843e-07, 0.0000e+00, 2.9802e-08, ..., 4.6566e-10, |
| -1.2349e-06, 0.0000e+00], |
| [ 4.8103e-07, 0.0000e+00, 6.9849e-09, ..., 0.0000e+00, |
| 1.8347e-07, 0.0000e+00], |
| ..., |
| [-3.6843e-06, 0.0000e+00, 1.9092e-08, ..., 4.6566e-10, |
| 7.2410e-07, 0.0000e+00], |
| [ 4.9826e-07, 0.0000e+00, 1.5786e-07, ..., 2.7940e-09, |
| 1.9558e-07, 0.0000e+00], |
| [ 4.2748e-07, 0.0000e+00, 3.6787e-08, ..., 4.6566e-10, |
| 2.1420e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 174, bias, value: tensor([-0.0048, 0.0219, 0.0035, -0.0017, 0.0113, 0.0054, 0.0251, -0.0121, |
| 0.0079, -0.0041], device='cuda:0'), grad: tensor([-3.2969e-07, -1.4976e-06, 1.2200e-06, 1.3344e-05, 3.1153e-07, |
| -1.0528e-05, -1.2852e-07, -4.8652e-06, 1.2126e-06, 1.2703e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 173, time 217.46, cls_loss 0.0018 cls_loss_mapping 0.0045 cls_loss_causal 0.5394 re_mapping 0.0066 re_causal 0.0205 /// teacc 98.95 lr 0.00010000 |
| Epoch 175, weight, value: tensor([[-0.0501, -0.0797, -0.0562, ..., -0.0986, -0.0151, -0.0288], |
| [-0.0502, 0.0317, -0.1499, ..., 0.0564, 0.0662, -0.1251], |
| [ 0.0577, -0.0035, -0.0547, ..., -0.0518, -0.0352, 0.0165], |
| ..., |
| [ 0.0534, -0.0463, -0.0064, ..., 0.0218, -0.0384, 0.0223], |
| [-0.0513, -0.0275, -0.0775, ..., -0.0175, -0.0314, -0.0616], |
| [-0.1336, -0.1782, 0.0285, ..., -0.0570, 0.0366, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 1.4901e-07, 0.0000e+00, 1.4435e-07, ..., 9.3132e-10, |
| 2.0117e-07, 0.0000e+00], |
| [ 3.7067e-07, 0.0000e+00, 2.6077e-08, ..., 0.0000e+00, |
| -4.3772e-08, 0.0000e+00], |
| [ 3.8277e-07, 0.0000e+00, 2.7940e-08, ..., 0.0000e+00, |
| 6.6124e-08, 0.0000e+00], |
| ..., |
| [-2.6897e-06, 0.0000e+00, 2.2352e-08, ..., 0.0000e+00, |
| 1.7695e-08, 0.0000e+00], |
| [ 1.3281e-06, 0.0000e+00, 1.5348e-06, ..., 1.8626e-09, |
| 2.2277e-06, 0.0000e+00], |
| [ 1.6894e-06, 0.0000e+00, -2.5593e-06, ..., 3.7253e-09, |
| -4.1239e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 175, bias, value: tensor([-0.0047, 0.0219, 0.0033, -0.0006, 0.0113, 0.0038, 0.0250, -0.0120, |
| 0.0077, -0.0040], device='cuda:0'), grad: tensor([ 1.6699e-06, 6.5565e-07, 1.9521e-06, 4.6343e-06, 1.2545e-06, |
| 3.6787e-07, 1.9372e-07, -4.6156e-06, 1.6019e-05, -2.2128e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 174, time 217.71, cls_loss 0.0026 cls_loss_mapping 0.0034 cls_loss_causal 0.5116 re_mapping 0.0066 re_causal 0.0198 /// teacc 98.87 lr 0.00010000 |
| Epoch 176, weight, value: tensor([[-0.0506, -0.0815, -0.0566, ..., -0.0989, -0.0152, -0.0288], |
| [-0.0502, 0.0346, -0.1505, ..., 0.0565, 0.0662, -0.1252], |
| [ 0.0579, -0.0043, -0.0549, ..., -0.0530, -0.0354, 0.0164], |
| ..., |
| [ 0.0535, -0.0482, -0.0068, ..., 0.0217, -0.0384, 0.0225], |
| [-0.0517, -0.0277, -0.0783, ..., -0.0177, -0.0320, -0.0616], |
| [-0.1337, -0.1789, 0.0304, ..., -0.0571, 0.0368, 0.0440]], |
| device='cuda:0'), grad: tensor([[ 1.2387e-07, 0.0000e+00, 4.2655e-07, ..., 8.3819e-09, |
| 1.3970e-08, 0.0000e+00], |
| [ 4.2282e-07, 0.0000e+00, 5.6904e-07, ..., 9.3132e-08, |
| -3.9116e-08, 0.0000e+00], |
| [ 3.0641e-07, 0.0000e+00, 4.9546e-07, ..., 6.7987e-08, |
| 9.3132e-09, 0.0000e+00], |
| ..., |
| [ 2.0396e-07, 0.0000e+00, 4.9081e-07, ..., 7.4506e-08, |
| 8.9407e-08, 0.0000e+00], |
| [ 2.8498e-07, 0.0000e+00, 1.1828e-07, ..., 1.8626e-09, |
| 2.5146e-08, 0.0000e+00], |
| [ 2.2072e-07, 0.0000e+00, -1.6734e-05, ..., 2.1420e-08, |
| -1.3132e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 176, bias, value: tensor([-0.0047, 0.0219, 0.0033, -0.0008, 0.0094, 0.0048, 0.0238, -0.0120, |
| 0.0073, -0.0023], device='cuda:0'), grad: tensor([ 8.7637e-07, 1.5022e-06, 1.6857e-06, -4.4107e-05, 2.3648e-05, |
| 4.7028e-05, 1.9073e-06, 1.6000e-06, 5.8115e-07, -3.4839e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 175, time 217.51, cls_loss 0.0023 cls_loss_mapping 0.0032 cls_loss_causal 0.5346 re_mapping 0.0065 re_causal 0.0200 /// teacc 98.95 lr 0.00010000 |
| Epoch 177, weight, value: tensor([[-0.0510, -0.0844, -0.0569, ..., -0.0993, -0.0153, -0.0289], |
| [-0.0501, 0.0381, -0.1520, ..., 0.0567, 0.0666, -0.1252], |
| [ 0.0577, -0.0067, -0.0551, ..., -0.0537, -0.0363, 0.0163], |
| ..., |
| [ 0.0536, -0.0498, -0.0068, ..., 0.0217, -0.0386, 0.0224], |
| [-0.0519, -0.0299, -0.0786, ..., -0.0178, -0.0324, -0.0616], |
| [-0.1342, -0.1804, 0.0313, ..., -0.0572, 0.0367, 0.0441]], |
| device='cuda:0'), grad: tensor([[ 2.3358e-06, 0.0000e+00, 1.5367e-06, ..., 0.0000e+00, |
| 1.4901e-08, 0.0000e+00], |
| [ 1.4555e-04, 0.0000e+00, 3.4645e-07, ..., 0.0000e+00, |
| -1.2191e-06, 0.0000e+00], |
| [ 1.8477e-05, 0.0000e+00, 5.4576e-07, ..., 0.0000e+00, |
| 7.4506e-09, 0.0000e+00], |
| ..., |
| [-2.3496e-04, 0.0000e+00, 7.2643e-08, ..., 0.0000e+00, |
| 1.5553e-07, 0.0000e+00], |
| [ 8.0049e-05, 0.0000e+00, 1.8710e-06, ..., 0.0000e+00, |
| 5.8580e-07, 0.0000e+00], |
| [ 1.0774e-05, 0.0000e+00, 1.1653e-05, ..., 0.0000e+00, |
| 1.1921e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 177, bias, value: tensor([-0.0047, 0.0223, 0.0025, -0.0010, 0.0084, 0.0045, 0.0237, -0.0121, |
| 0.0073, -0.0015], device='cuda:0'), grad: tensor([ 1.3605e-05, 4.6992e-04, 5.9366e-05, 4.4644e-05, 4.5486e-06, |
| -8.6725e-05, -4.1515e-05, -7.9012e-04, 2.5797e-04, 6.7592e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 176, time 217.40, cls_loss 0.0024 cls_loss_mapping 0.0037 cls_loss_causal 0.5344 re_mapping 0.0067 re_causal 0.0197 /// teacc 98.92 lr 0.00010000 |
| Epoch 178, weight, value: tensor([[-0.0513, -0.0846, -0.0570, ..., -0.0994, -0.0154, -0.0289], |
| [-0.0502, 0.0382, -0.1522, ..., 0.0568, 0.0666, -0.1252], |
| [ 0.0580, -0.0068, -0.0552, ..., -0.0538, -0.0356, 0.0163], |
| ..., |
| [ 0.0537, -0.0498, -0.0069, ..., 0.0215, -0.0387, 0.0224], |
| [-0.0527, -0.0304, -0.0791, ..., -0.0179, -0.0325, -0.0616], |
| [-0.1346, -0.1809, 0.0311, ..., -0.0572, 0.0369, 0.0441]], |
| device='cuda:0'), grad: tensor([[ 8.1956e-08, 0.0000e+00, 5.5879e-09, ..., 0.0000e+00, |
| 1.5832e-08, 0.0000e+00], |
| [ 1.9008e-06, 0.0000e+00, 1.0524e-07, ..., 0.0000e+00, |
| -5.1316e-07, 0.0000e+00], |
| [ 7.5437e-08, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| 1.6764e-08, 0.0000e+00], |
| ..., |
| [ 4.9081e-07, 0.0000e+00, -1.0338e-07, ..., 0.0000e+00, |
| 6.6683e-07, 0.0000e+00], |
| [ 4.0978e-08, 0.0000e+00, 7.6089e-07, ..., 0.0000e+00, |
| -4.0978e-08, 0.0000e+00], |
| [-7.0184e-06, 0.0000e+00, 2.9523e-07, ..., 0.0000e+00, |
| -5.6066e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 178, bias, value: tensor([-0.0048, 0.0223, 0.0030, -0.0011, 0.0106, 0.0045, 0.0239, -0.0121, |
| 0.0068, -0.0036], device='cuda:0'), grad: tensor([ 1.0151e-07, 2.7232e-06, 2.9430e-07, 2.7195e-06, 1.2636e-05, |
| -2.7567e-06, 6.4913e-07, 8.3297e-06, 1.2098e-06, -2.5928e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 177, time 217.36, cls_loss 0.0020 cls_loss_mapping 0.0032 cls_loss_causal 0.4922 re_mapping 0.0066 re_causal 0.0197 /// teacc 99.00 lr 0.00010000 |
| Epoch 179, weight, value: tensor([[-0.0528, -0.0848, -0.0571, ..., -0.0995, -0.0156, -0.0289], |
| [-0.0502, 0.0383, -0.1555, ..., 0.0569, 0.0663, -0.1252], |
| [ 0.0580, -0.0068, -0.0551, ..., -0.0546, -0.0356, 0.0163], |
| ..., |
| [ 0.0536, -0.0499, -0.0071, ..., 0.0215, -0.0389, 0.0224], |
| [-0.0532, -0.0306, -0.0793, ..., -0.0179, -0.0336, -0.0616], |
| [-0.1353, -0.1813, 0.0312, ..., -0.0567, 0.0378, 0.0441]], |
| device='cuda:0'), grad: tensor([[-1.0338e-06, 0.0000e+00, 1.1176e-08, ..., 9.3132e-10, |
| 1.0077e-06, 0.0000e+00], |
| [ 2.9057e-07, -9.3132e-10, 1.4342e-07, ..., 1.2107e-08, |
| -5.4296e-07, 0.0000e+00], |
| [-7.0669e-06, 0.0000e+00, 2.4214e-08, ..., 9.3132e-10, |
| 1.4687e-06, 0.0000e+00], |
| ..., |
| [ 8.2254e-06, 9.3132e-10, 2.7940e-08, ..., 1.8626e-09, |
| 4.9453e-07, 0.0000e+00], |
| [ 5.4911e-06, 0.0000e+00, 4.0978e-08, ..., 9.3132e-10, |
| 7.5996e-07, 0.0000e+00], |
| [ 2.1681e-06, 0.0000e+00, 1.3784e-07, ..., 1.1176e-08, |
| 1.1455e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 179, bias, value: tensor([-0.0053, 0.0218, 0.0030, -0.0009, 0.0117, 0.0045, 0.0242, -0.0123, |
| 0.0061, -0.0035], device='cuda:0'), grad: tensor([-1.0490e-05, -2.0899e-06, -1.4510e-06, -1.3478e-05, 6.2101e-06, |
| 7.2829e-07, -1.9118e-05, 1.6659e-05, 1.0513e-05, 1.2442e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 178, time 217.36, cls_loss 0.0021 cls_loss_mapping 0.0038 cls_loss_causal 0.5240 re_mapping 0.0070 re_causal 0.0196 /// teacc 98.83 lr 0.00010000 |
| Epoch 180, weight, value: tensor([[-0.0541, -0.0853, -0.0572, ..., -0.0995, -0.0157, -0.0289], |
| [-0.0502, 0.0384, -0.1557, ..., 0.0569, 0.0664, -0.1253], |
| [ 0.0584, -0.0046, -0.0547, ..., -0.0546, -0.0357, 0.0164], |
| ..., |
| [ 0.0535, -0.0516, -0.0072, ..., 0.0215, -0.0390, 0.0223], |
| [-0.0538, -0.0308, -0.0794, ..., -0.0179, -0.0336, -0.0617], |
| [-0.1358, -0.1820, 0.0310, ..., -0.0568, 0.0379, 0.0441]], |
| device='cuda:0'), grad: tensor([[-9.3132e-09, 0.0000e+00, 3.7253e-08, ..., 0.0000e+00, |
| 7.4863e-05, 0.0000e+00], |
| [ 6.7800e-07, 0.0000e+00, 8.7917e-07, ..., 0.0000e+00, |
| -1.2340e-06, 0.0000e+00], |
| [ 3.1330e-06, 0.0000e+00, 1.5739e-07, ..., 0.0000e+00, |
| 3.6694e-07, 0.0000e+00], |
| ..., |
| [ 4.0978e-06, 0.0000e+00, 2.1141e-07, ..., 0.0000e+00, |
| 2.4959e-06, 0.0000e+00], |
| [-7.8380e-06, 0.0000e+00, 9.4995e-08, ..., 0.0000e+00, |
| 9.9558e-07, 0.0000e+00], |
| [-5.6904e-07, 0.0000e+00, 3.1032e-06, ..., 0.0000e+00, |
| -4.0382e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 180, bias, value: tensor([-0.0050, 0.0217, 0.0031, -0.0003, 0.0118, 0.0042, 0.0246, -0.0124, |
| 0.0060, -0.0036], device='cuda:0'), grad: tensor([ 9.6560e-04, 2.4959e-06, 1.2040e-05, 1.9390e-06, -4.2245e-06, |
| 3.2391e-06, -9.8038e-04, 4.6074e-05, -4.3392e-05, -4.9397e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 179, time 217.26, cls_loss 0.0032 cls_loss_mapping 0.0034 cls_loss_causal 0.5091 re_mapping 0.0062 re_causal 0.0188 /// teacc 98.98 lr 0.00010000 |
| Epoch 181, weight, value: tensor([[-0.0544, -0.0860, -0.0573, ..., -0.0995, -0.0160, -0.0322], |
| [-0.0497, 0.0399, -0.1558, ..., 0.0569, 0.0667, -0.1261], |
| [ 0.0580, -0.0057, -0.0549, ..., -0.0547, -0.0361, 0.0138], |
| ..., |
| [ 0.0532, -0.0525, -0.0076, ..., 0.0215, -0.0393, 0.0215], |
| [-0.0537, -0.0309, -0.0796, ..., -0.0179, -0.0338, -0.0619], |
| [-0.1372, -0.1827, 0.0305, ..., -0.0568, 0.0384, 0.0474]], |
| device='cuda:0'), grad: tensor([[ 2.2352e-08, 0.0000e+00, 4.9360e-08, ..., 5.5879e-09, |
| 1.6112e-07, 9.3132e-10], |
| [ 7.4226e-07, 0.0000e+00, 8.8476e-08, ..., -1.3569e-06, |
| -1.0379e-05, 9.3132e-10], |
| [ 3.1535e-06, 0.0000e+00, 4.8429e-08, ..., 2.7940e-09, |
| 1.0338e-07, -2.4214e-08], |
| ..., |
| [-5.9530e-06, 0.0000e+00, 3.3155e-07, ..., 9.6392e-07, |
| 8.4266e-06, 2.0489e-08], |
| [ 7.8231e-08, 0.0000e+00, -3.7253e-09, ..., 1.3970e-08, |
| -2.3935e-07, 0.0000e+00], |
| [-2.8908e-06, 0.0000e+00, 6.4075e-07, ..., 2.2165e-07, |
| -1.3754e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 181, bias, value: tensor([-0.0074, 0.0221, 0.0024, -0.0004, 0.0120, 0.0045, 0.0251, -0.0127, |
| 0.0062, -0.0027], device='cuda:0'), grad: tensor([-2.8223e-05, -1.3173e-05, 7.3463e-06, 3.0454e-06, 8.0705e-05, |
| 1.8431e-06, 6.7241e-06, 8.4639e-06, -3.4757e-06, -6.3300e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 180, time 217.34, cls_loss 0.0033 cls_loss_mapping 0.0051 cls_loss_causal 0.5516 re_mapping 0.0072 re_causal 0.0200 /// teacc 98.91 lr 0.00010000 |
| Epoch 182, weight, value: tensor([[-0.0548, -0.0890, -0.0563, ..., -0.0997, -0.0162, -0.0323], |
| [-0.0496, 0.0410, -0.1559, ..., 0.0568, 0.0668, -0.1264], |
| [ 0.0580, -0.0062, -0.0546, ..., -0.0548, -0.0364, 0.0142], |
| ..., |
| [ 0.0566, -0.0531, -0.0078, ..., 0.0215, -0.0381, 0.0211], |
| [-0.0545, -0.0310, -0.0795, ..., -0.0181, -0.0330, -0.0622], |
| [-0.1408, -0.1838, 0.0300, ..., -0.0565, 0.0356, 0.0475]], |
| device='cuda:0'), grad: tensor([[ 1.7695e-08, 0.0000e+00, 3.2596e-08, ..., 0.0000e+00, |
| 1.5832e-08, 0.0000e+00], |
| [ 2.6077e-08, 0.0000e+00, 2.6077e-08, ..., 0.0000e+00, |
| -1.5832e-07, 0.0000e+00], |
| [-4.2655e-07, 0.0000e+00, 3.0734e-08, ..., 0.0000e+00, |
| 1.8626e-08, 0.0000e+00], |
| ..., |
| [ 4.7404e-07, 0.0000e+00, 3.7253e-08, ..., 0.0000e+00, |
| 1.3318e-07, 0.0000e+00], |
| [ 4.3772e-08, 0.0000e+00, 3.6135e-07, ..., 0.0000e+00, |
| 1.9092e-07, 0.0000e+00], |
| [ 9.0338e-08, 0.0000e+00, 3.1851e-07, ..., 0.0000e+00, |
| -1.0151e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 182, bias, value: tensor([-0.0066, 0.0222, 0.0023, -0.0042, 0.0120, 0.0041, 0.0255, -0.0102, |
| 0.0067, -0.0053], device='cuda:0'), grad: tensor([-3.2317e-07, -2.5425e-07, -3.8464e-07, -4.3213e-07, 1.0058e-07, |
| -1.5080e-05, 1.3612e-05, 9.4157e-07, 1.3318e-06, 4.6473e-07], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 181, time 217.48, cls_loss 0.0021 cls_loss_mapping 0.0037 cls_loss_causal 0.5057 re_mapping 0.0069 re_causal 0.0202 /// teacc 99.00 lr 0.00010000 |
| Epoch 183, weight, value: tensor([[-0.0550, -0.0909, -0.0564, ..., -0.0999, -0.0163, -0.0323], |
| [-0.0499, 0.0418, -0.1561, ..., 0.0568, 0.0668, -0.1268], |
| [ 0.0580, -0.0065, -0.0556, ..., -0.0548, -0.0367, 0.0136], |
| ..., |
| [ 0.0564, -0.0537, -0.0085, ..., 0.0216, -0.0380, 0.0213], |
| [-0.0557, -0.0313, -0.0794, ..., -0.0182, -0.0336, -0.0627], |
| [-0.1408, -0.1867, 0.0301, ..., -0.0561, 0.0358, 0.0475]], |
| device='cuda:0'), grad: tensor([[ 2.1979e-07, 9.3132e-09, 3.7253e-08, ..., 0.0000e+00, |
| 2.3283e-08, 1.6764e-08], |
| [-1.0796e-05, 6.5193e-09, 1.0245e-08, ..., 0.0000e+00, |
| -1.1310e-05, 3.7253e-09], |
| [-2.8443e-06, -2.1327e-07, 1.3039e-08, ..., 0.0000e+00, |
| 1.7229e-07, 5.5879e-09], |
| ..., |
| [ 1.1265e-05, 2.1700e-07, 5.4948e-08, ..., 0.0000e+00, |
| 1.1042e-05, 2.5146e-08], |
| [ 9.1176e-07, 5.5879e-09, 6.6590e-07, ..., 0.0000e+00, |
| -4.8950e-06, 5.5879e-09], |
| [ 7.1526e-07, 2.2352e-08, 1.1548e-07, ..., 0.0000e+00, |
| 5.3085e-08, 5.1223e-08]], device='cuda:0') |
| Epoch 183, bias, value: tensor([-0.0065, 0.0221, 0.0021, -0.0034, 0.0118, 0.0036, 0.0256, -0.0103, |
| 0.0062, -0.0051], device='cuda:0'), grad: tensor([-8.0094e-07, -2.5913e-05, -3.1013e-06, 2.0325e-05, 3.2689e-07, |
| -1.8045e-05, 1.4335e-05, 2.5362e-05, -1.4544e-05, 2.0210e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 182, time 217.24, cls_loss 0.0029 cls_loss_mapping 0.0048 cls_loss_causal 0.5477 re_mapping 0.0070 re_causal 0.0204 /// teacc 98.82 lr 0.00010000 |
| Epoch 184, weight, value: tensor([[-0.0544, -0.0920, -0.0566, ..., -0.1026, -0.0160, -0.0323], |
| [-0.0502, 0.0422, -0.1562, ..., 0.0568, 0.0670, -0.1272], |
| [ 0.0581, -0.0055, -0.0555, ..., -0.0556, -0.0371, 0.0135], |
| ..., |
| [ 0.0566, -0.0549, -0.0094, ..., 0.0219, -0.0381, 0.0214], |
| [-0.0569, -0.0312, -0.0802, ..., -0.0186, -0.0344, -0.0627], |
| [-0.1411, -0.1877, 0.0294, ..., -0.0572, 0.0357, 0.0475]], |
| device='cuda:0'), grad: tensor([[ 2.7847e-07, 0.0000e+00, 4.3027e-07, ..., 0.0000e+00, |
| 4.3400e-07, 0.0000e+00], |
| [ 1.3225e-07, 0.0000e+00, 3.3341e-07, ..., 3.7253e-09, |
| -2.3376e-07, 0.0000e+00], |
| [-9.7323e-07, 0.0000e+00, 3.4831e-07, ..., 0.0000e+00, |
| 1.1399e-06, 0.0000e+00], |
| ..., |
| [ 7.5251e-07, 0.0000e+00, 1.2005e-06, ..., -5.5879e-09, |
| 2.8759e-06, 0.0000e+00], |
| [ 1.5711e-06, 0.0000e+00, 8.8010e-07, ..., 0.0000e+00, |
| 2.6412e-06, 0.0000e+00], |
| [-2.1495e-06, 0.0000e+00, -1.5972e-06, ..., 9.3132e-10, |
| -9.2760e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 184, bias, value: tensor([-0.0061, 0.0220, 0.0020, -0.0034, 0.0120, 0.0038, 0.0250, -0.0102, |
| 0.0054, -0.0053], device='cuda:0'), grad: tensor([ 3.7812e-06, 1.0710e-07, 1.7788e-06, 1.1902e-06, 5.0850e-07, |
| -2.4047e-06, 1.5311e-06, 1.0528e-05, 1.1221e-05, -2.8238e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 183, time 217.79, cls_loss 0.0021 cls_loss_mapping 0.0038 cls_loss_causal 0.5195 re_mapping 0.0068 re_causal 0.0195 /// teacc 98.89 lr 0.00010000 |
| Epoch 185, weight, value: tensor([[-0.0545, -0.0928, -0.0553, ..., -0.1031, -0.0161, -0.0323], |
| [-0.0503, 0.0431, -0.1564, ..., 0.0567, 0.0671, -0.1274], |
| [ 0.0588, -0.0058, -0.0556, ..., -0.0570, -0.0373, 0.0136], |
| ..., |
| [ 0.0566, -0.0556, -0.0097, ..., 0.0219, -0.0382, 0.0213], |
| [-0.0583, -0.0305, -0.0802, ..., -0.0189, -0.0337, -0.0628], |
| [-0.1410, -0.1885, 0.0290, ..., -0.0563, 0.0361, 0.0475]], |
| device='cuda:0'), grad: tensor([[ 1.4994e-07, 0.0000e+00, -4.2841e-08, ..., 6.6124e-08, |
| 8.4750e-08, 0.0000e+00], |
| [ 2.0731e-06, 0.0000e+00, 3.2596e-08, ..., 1.6810e-06, |
| -1.0934e-06, 0.0000e+00], |
| [ 6.6776e-07, 0.0000e+00, 2.1420e-08, ..., 3.0734e-08, |
| 7.6462e-07, 0.0000e+00], |
| ..., |
| [-4.1425e-06, 0.0000e+00, 5.1223e-08, ..., -2.5332e-06, |
| 2.3190e-07, 0.0000e+00], |
| [ 8.0094e-08, 0.0000e+00, 8.2888e-08, ..., 2.9802e-08, |
| 6.8918e-08, 0.0000e+00], |
| [ 5.2620e-07, 0.0000e+00, 8.8196e-07, ..., 4.2468e-07, |
| -2.0862e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 185, bias, value: tensor([-0.0056, 0.0220, 0.0027, -0.0043, 0.0120, 0.0060, 0.0236, -0.0103, |
| 0.0055, -0.0054], device='cuda:0'), grad: tensor([ 1.6671e-07, 5.0887e-06, 2.9076e-06, 1.7481e-06, -1.5274e-07, |
| -2.8033e-07, -1.3877e-06, -1.1422e-05, 4.6659e-07, 2.8554e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 184, time 217.34, cls_loss 0.0020 cls_loss_mapping 0.0033 cls_loss_causal 0.5266 re_mapping 0.0068 re_causal 0.0198 /// teacc 98.95 lr 0.00010000 |
| Epoch 186, weight, value: tensor([[-0.0548, -0.0931, -0.0579, ..., -0.1046, -0.0160, -0.0323], |
| [-0.0504, 0.0432, -0.1566, ..., 0.0568, 0.0672, -0.1276], |
| [ 0.0590, -0.0058, -0.0560, ..., -0.0575, -0.0375, 0.0136], |
| ..., |
| [ 0.0565, -0.0556, -0.0118, ..., 0.0215, -0.0383, 0.0212], |
| [-0.0586, -0.0308, -0.0785, ..., -0.0201, -0.0335, -0.0629], |
| [-0.1411, -0.1887, 0.0287, ..., -0.0570, 0.0361, 0.0475]], |
| device='cuda:0'), grad: tensor([[ 5.5879e-09, 0.0000e+00, -1.9092e-07, ..., 0.0000e+00, |
| 2.3097e-07, 0.0000e+00], |
| [ 6.1467e-08, 0.0000e+00, 1.2480e-07, ..., 0.0000e+00, |
| -6.1281e-07, 0.0000e+00], |
| [-8.5682e-08, 0.0000e+00, 5.2154e-08, ..., 0.0000e+00, |
| 4.0978e-08, 0.0000e+00], |
| ..., |
| [ 2.1141e-07, 0.0000e+00, 2.0843e-06, ..., 0.0000e+00, |
| 1.1548e-07, 9.3132e-10], |
| [-3.3602e-06, 0.0000e+00, 6.7987e-08, ..., 0.0000e+00, |
| 2.4084e-06, 0.0000e+00], |
| [ 3.0063e-06, 0.0000e+00, 3.0279e-05, ..., 0.0000e+00, |
| -2.7940e-08, -3.7253e-09]], device='cuda:0') |
| Epoch 186, bias, value: tensor([-0.0056, 0.0219, 0.0027, -0.0043, 0.0123, 0.0053, 0.0230, -0.0104, |
| 0.0073, -0.0055], device='cuda:0'), grad: tensor([-1.3756e-06, 5.6140e-06, 1.0943e-06, 7.5251e-07, -5.7906e-05, |
| 2.9095e-06, -1.0684e-05, 6.6906e-06, -2.1085e-05, 7.3910e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 185, time 217.74, cls_loss 0.0022 cls_loss_mapping 0.0035 cls_loss_causal 0.5201 re_mapping 0.0067 re_causal 0.0185 /// teacc 99.01 lr 0.00010000 |
| Epoch 187, weight, value: tensor([[-0.0551, -0.0933, -0.0580, ..., -0.1058, -0.0160, -0.0323], |
| [-0.0503, 0.0434, -0.1567, ..., 0.0578, 0.0675, -0.1280], |
| [ 0.0588, -0.0061, -0.0567, ..., -0.0578, -0.0380, 0.0134], |
| ..., |
| [ 0.0566, -0.0555, -0.0122, ..., 0.0207, -0.0386, 0.0206], |
| [-0.0596, -0.0312, -0.0787, ..., -0.0206, -0.0336, -0.0637], |
| [-0.1412, -0.1891, 0.0284, ..., -0.0587, 0.0364, 0.0476]], |
| device='cuda:0'), grad: tensor([[ 3.4273e-07, 0.0000e+00, 6.4261e-08, ..., 0.0000e+00, |
| 1.7043e-07, 0.0000e+00], |
| [ 6.6496e-07, 0.0000e+00, 1.2107e-07, ..., 0.0000e+00, |
| -2.0396e-07, 0.0000e+00], |
| [-1.7926e-05, 0.0000e+00, 1.6112e-07, ..., 0.0000e+00, |
| 2.8871e-08, 0.0000e+00], |
| ..., |
| [ 1.2584e-05, 0.0000e+00, 1.0263e-06, ..., 0.0000e+00, |
| 1.8226e-06, 4.8429e-08], |
| [ 4.0829e-06, 0.0000e+00, 1.2945e-07, ..., 0.0000e+00, |
| 7.9535e-07, 0.0000e+00], |
| [ 3.6135e-07, 0.0000e+00, -3.2820e-06, ..., 0.0000e+00, |
| -2.6561e-06, -5.7742e-08]], device='cuda:0') |
| Epoch 187, bias, value: tensor([-0.0055, 0.0221, 0.0024, -0.0041, 0.0121, 0.0050, 0.0241, -0.0106, |
| 0.0072, -0.0053], device='cuda:0'), grad: tensor([ 2.3600e-06, 1.3169e-06, -2.9624e-05, 9.9465e-07, 1.6585e-05, |
| 8.2105e-06, -9.0748e-06, 3.7760e-05, 1.7405e-05, -4.5925e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 186, time 217.68, cls_loss 0.0023 cls_loss_mapping 0.0040 cls_loss_causal 0.5198 re_mapping 0.0063 re_causal 0.0189 /// teacc 98.92 lr 0.00010000 |
| Epoch 188, weight, value: tensor([[-0.0553, -0.0941, -0.0582, ..., -0.1067, -0.0162, -0.0323], |
| [-0.0504, 0.0444, -0.1573, ..., 0.0578, 0.0676, -0.1281], |
| [ 0.0593, -0.0071, -0.0552, ..., -0.0579, -0.0369, 0.0134], |
| ..., |
| [ 0.0565, -0.0558, -0.0124, ..., 0.0207, -0.0390, 0.0195], |
| [-0.0602, -0.0324, -0.0787, ..., -0.0209, -0.0323, -0.0639], |
| [-0.1407, -0.1899, 0.0281, ..., -0.0607, 0.0369, 0.0476]], |
| device='cuda:0'), grad: tensor([[ 5.5507e-07, 0.0000e+00, 3.4925e-07, ..., 0.0000e+00, |
| 2.2352e-08, 0.0000e+00], |
| [ 7.9349e-07, 0.0000e+00, 6.1467e-08, ..., 0.0000e+00, |
| 5.3085e-08, 0.0000e+00], |
| [-1.7434e-05, 0.0000e+00, 8.5030e-07, ..., 0.0000e+00, |
| 1.3225e-07, 0.0000e+00], |
| ..., |
| [-4.6659e-07, 0.0000e+00, 5.4948e-08, ..., 0.0000e+00, |
| 9.3132e-10, 0.0000e+00], |
| [ 1.7984e-06, 2.7940e-09, 1.5832e-07, ..., 0.0000e+00, |
| 8.5682e-08, 0.0000e+00], |
| [ 2.0638e-06, 0.0000e+00, 5.6997e-06, ..., 0.0000e+00, |
| -2.4773e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 188, bias, value: tensor([-0.0055, 0.0220, 0.0032, -0.0042, 0.0124, 0.0048, 0.0240, -0.0109, |
| 0.0081, -0.0052], device='cuda:0'), grad: tensor([ 1.1846e-06, 1.5805e-06, -4.0084e-05, 3.7253e-05, -1.7092e-05, |
| 2.5891e-07, 1.2200e-07, -7.5344e-07, 3.8780e-06, 1.3664e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 187, time 217.32, cls_loss 0.0020 cls_loss_mapping 0.0034 cls_loss_causal 0.5193 re_mapping 0.0064 re_causal 0.0192 /// teacc 98.91 lr 0.00010000 |
| Epoch 189, weight, value: tensor([[-0.0554, -0.0953, -0.0581, ..., -0.1068, -0.0164, -0.0323], |
| [-0.0505, 0.0446, -0.1574, ..., 0.0593, 0.0681, -0.1284], |
| [ 0.0592, -0.0063, -0.0565, ..., -0.0582, -0.0371, 0.0127], |
| ..., |
| [ 0.0565, -0.0565, -0.0119, ..., 0.0196, -0.0394, 0.0195], |
| [-0.0607, -0.0328, -0.0795, ..., -0.0212, -0.0324, -0.0647], |
| [-0.1407, -0.1907, 0.0285, ..., -0.0607, 0.0370, 0.0476]], |
| device='cuda:0'), grad: tensor([[ 1.5339e-06, 9.2201e-08, 4.7497e-08, ..., 1.8626e-09, |
| 1.5553e-07, 3.7253e-09], |
| [-5.2340e-07, -3.1143e-06, 3.9861e-07, ..., 4.0978e-08, |
| -5.1856e-06, 2.0489e-08], |
| [-2.5079e-05, 6.6031e-07, 5.6811e-08, ..., 3.7253e-09, |
| 1.0421e-06, 8.3819e-09], |
| ..., |
| [ 3.1106e-06, 1.7853e-06, 7.7952e-07, ..., 3.7253e-09, |
| 2.9616e-06, 2.7940e-08], |
| [ 1.5395e-06, 1.9558e-07, 7.8231e-08, ..., 0.0000e+00, |
| 3.1292e-07, 4.9360e-08], |
| [ 1.0505e-06, 1.3970e-07, 2.2426e-06, ..., 1.1828e-07, |
| 2.2911e-07, 6.5193e-08]], device='cuda:0') |
| Epoch 189, bias, value: tensor([-0.0055, 0.0223, 0.0030, -0.0042, 0.0123, 0.0047, 0.0245, -0.0111, |
| 0.0076, -0.0052], device='cuda:0'), grad: tensor([ 1.1865e-06, -8.8587e-06, -4.2409e-05, -5.9485e-05, -9.2685e-06, |
| 9.1076e-05, 2.1942e-06, 1.2279e-05, 3.5465e-06, 9.6038e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 188, time 217.87, cls_loss 0.0019 cls_loss_mapping 0.0027 cls_loss_causal 0.5406 re_mapping 0.0064 re_causal 0.0199 /// teacc 98.99 lr 0.00010000 |
| Epoch 190, weight, value: tensor([[-0.0557, -0.0962, -0.0581, ..., -0.1073, -0.0163, -0.0323], |
| [-0.0505, 0.0448, -0.1578, ..., 0.0610, 0.0685, -0.1292], |
| [ 0.0596, -0.0063, -0.0563, ..., -0.0582, -0.0371, 0.0127], |
| ..., |
| [ 0.0565, -0.0566, -0.0121, ..., 0.0180, -0.0397, 0.0198], |
| [-0.0611, -0.0326, -0.0796, ..., -0.0214, -0.0332, -0.0649], |
| [-0.1408, -0.1916, 0.0284, ..., -0.0611, 0.0370, 0.0476]], |
| device='cuda:0'), grad: tensor([[ 4.8429e-08, 3.4273e-07, 1.2852e-07, ..., 2.7940e-09, |
| 1.7975e-07, 0.0000e+00], |
| [-1.0589e-06, -1.9640e-05, 2.3283e-07, ..., 6.5193e-09, |
| -4.0047e-07, 0.0000e+00], |
| [ 2.2165e-07, 1.2331e-05, 1.1455e-07, ..., 3.7253e-09, |
| 3.2783e-07, 0.0000e+00], |
| ..., |
| [ 2.0582e-07, 7.9721e-07, 2.6450e-07, ..., 5.1223e-08, |
| 1.2740e-06, 0.0000e+00], |
| [ 8.1025e-08, 1.8906e-07, 2.5053e-07, ..., 9.3132e-10, |
| -2.5108e-06, 0.0000e+00], |
| [-5.1223e-08, 2.9802e-08, 2.7698e-06, ..., 4.6566e-09, |
| -2.4009e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 190, bias, value: tensor([-0.0052, 0.0227, 0.0035, -0.0040, 0.0123, 0.0033, 0.0253, -0.0114, |
| 0.0076, -0.0052], device='cuda:0'), grad: tensor([ 2.0191e-06, -7.1228e-05, 4.4823e-05, 4.6268e-06, 6.1393e-06, |
| 1.7136e-05, 7.8678e-06, 8.6725e-06, -1.5676e-05, -4.4778e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 189, time 217.83, cls_loss 0.0023 cls_loss_mapping 0.0035 cls_loss_causal 0.5231 re_mapping 0.0063 re_causal 0.0190 /// teacc 98.96 lr 0.00010000 |
| Epoch 191, weight, value: tensor([[-0.0559, -0.0985, -0.0588, ..., -0.1076, -0.0170, -0.0324], |
| [-0.0501, 0.0477, -0.1580, ..., 0.0612, 0.0687, -0.1306], |
| [ 0.0588, -0.0090, -0.0593, ..., -0.0583, -0.0379, 0.0126], |
| ..., |
| [ 0.0566, -0.0575, -0.0122, ..., 0.0179, -0.0396, 0.0192], |
| [-0.0611, -0.0319, -0.0797, ..., -0.0215, -0.0339, -0.0655], |
| [-0.1412, -0.1933, 0.0307, ..., -0.0612, 0.0371, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 1.9372e-07, 0.0000e+00, 7.4506e-08, ..., 0.0000e+00, |
| 1.6019e-07, 1.6578e-07], |
| [ 4.8056e-07, 0.0000e+00, 4.1071e-07, ..., 0.0000e+00, |
| -8.9221e-07, 2.7940e-09], |
| [ 1.8701e-06, 0.0000e+00, 1.8813e-07, ..., 0.0000e+00, |
| 1.2945e-07, 1.6764e-08], |
| ..., |
| [ 4.8243e-06, 0.0000e+00, 7.8976e-07, ..., 0.0000e+00, |
| 8.7265e-07, 0.0000e+00], |
| [ 7.1060e-07, 0.0000e+00, 3.9302e-07, ..., 0.0000e+00, |
| -4.9919e-07, 1.6484e-07], |
| [ 1.2619e-06, 0.0000e+00, 1.7649e-06, ..., 0.0000e+00, |
| 4.1351e-07, 7.4506e-09]], device='cuda:0') |
| Epoch 191, bias, value: tensor([-0.0056, 0.0231, 0.0017, -0.0038, 0.0125, 0.0014, 0.0254, -0.0113, |
| 0.0074, -0.0046], device='cuda:0'), grad: tensor([ 2.2724e-06, -1.3318e-07, 2.1532e-06, -1.0058e-05, -4.9174e-06, |
| -1.0151e-07, -4.2915e-06, 8.5682e-06, -2.0396e-07, 6.6534e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| ---------------------saving model at epoch 190---------------------------------------------------- |
| epoch 190, time 218.54, cls_loss 0.0026 cls_loss_mapping 0.0034 cls_loss_causal 0.5075 re_mapping 0.0065 re_causal 0.0187 /// teacc 99.13 lr 0.00010000 |
| Epoch 192, weight, value: tensor([[-0.0564, -0.1009, -0.0593, ..., -0.1078, -0.0182, -0.0324], |
| [-0.0500, 0.0478, -0.1583, ..., 0.0612, 0.0689, -0.1308], |
| [ 0.0549, -0.0103, -0.0633, ..., -0.0583, -0.0407, 0.0126], |
| ..., |
| [ 0.0572, -0.0554, -0.0125, ..., 0.0179, -0.0394, 0.0192], |
| [-0.0620, -0.0329, -0.0800, ..., -0.0216, -0.0335, -0.0655], |
| [-0.1417, -0.1968, 0.0305, ..., -0.0612, 0.0378, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 1.4715e-07, 0.0000e+00, 1.6764e-08, ..., 0.0000e+00, |
| 5.8673e-08, 0.0000e+00], |
| [-9.4064e-08, 9.3132e-10, 5.5879e-09, ..., 0.0000e+00, |
| -3.1013e-06, 0.0000e+00], |
| [ 6.9570e-07, 9.3132e-10, 1.8626e-09, ..., 0.0000e+00, |
| 8.5682e-07, 0.0000e+00], |
| ..., |
| [ 2.1327e-07, -1.7695e-08, 7.4506e-09, ..., 0.0000e+00, |
| 1.7565e-06, 0.0000e+00], |
| [-5.7742e-08, 1.8626e-09, 1.6287e-05, ..., 0.0000e+00, |
| 4.7088e-06, 0.0000e+00], |
| [ 8.6613e-08, 1.8626e-09, -1.7077e-05, ..., 0.0000e+00, |
| -5.1893e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 192, bias, value: tensor([-0.0063, 0.0232, -0.0031, -0.0041, 0.0126, 0.0044, 0.0250, -0.0107, |
| 0.0071, -0.0044], device='cuda:0'), grad: tensor([-8.1584e-07, -7.7188e-06, 2.5351e-06, -2.5053e-06, 2.0508e-06, |
| 2.9467e-06, 2.2277e-06, 5.1036e-06, 3.5316e-05, -3.9220e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 191, time 217.78, cls_loss 0.0013 cls_loss_mapping 0.0023 cls_loss_causal 0.4607 re_mapping 0.0063 re_causal 0.0180 /// teacc 98.80 lr 0.00010000 |
| Epoch 193, weight, value: tensor([[-0.0561, -0.1010, -0.0593, ..., -0.1079, -0.0184, -0.0324], |
| [-0.0504, 0.0478, -0.1585, ..., 0.0612, 0.0689, -0.1308], |
| [ 0.0550, -0.0103, -0.0634, ..., -0.0583, -0.0411, 0.0126], |
| ..., |
| [ 0.0574, -0.0554, -0.0136, ..., 0.0179, -0.0394, 0.0192], |
| [-0.0623, -0.0326, -0.0802, ..., -0.0217, -0.0336, -0.0656], |
| [-0.1419, -0.1969, 0.0305, ..., -0.0612, 0.0380, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 6.5193e-08, 0.0000e+00, 6.2399e-08, ..., 0.0000e+00, |
| 3.9116e-08, 0.0000e+00], |
| [ 1.1455e-07, 0.0000e+00, 1.1548e-07, ..., 9.3132e-10, |
| -2.7940e-07, 0.0000e+00], |
| [-2.2277e-06, 0.0000e+00, 5.7742e-08, ..., 0.0000e+00, |
| 3.0734e-08, 0.0000e+00], |
| ..., |
| [ 4.8578e-06, -9.3132e-10, 1.5274e-07, ..., -3.7253e-09, |
| 2.7101e-07, 0.0000e+00], |
| [ 1.0421e-06, 0.0000e+00, 9.4716e-07, ..., 0.0000e+00, |
| 6.7335e-07, 0.0000e+00], |
| [-2.3469e-06, 0.0000e+00, -8.0541e-06, ..., 9.3132e-10, |
| -6.3926e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 193, bias, value: tensor([-0.0063, 0.0229, -0.0032, -0.0041, 0.0125, 0.0045, 0.0251, -0.0106, |
| 0.0069, -0.0043], device='cuda:0'), grad: tensor([ 1.2601e-06, -1.0058e-07, -2.3302e-06, 4.3511e-05, -9.2667e-07, |
| 1.4929e-06, 2.5891e-07, 7.1712e-06, 5.6811e-06, -5.6058e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 192, time 217.55, cls_loss 0.0022 cls_loss_mapping 0.0027 cls_loss_causal 0.4963 re_mapping 0.0064 re_causal 0.0185 /// teacc 99.02 lr 0.00010000 |
| Epoch 194, weight, value: tensor([[-0.0559, -0.1018, -0.0595, ..., -0.1079, -0.0187, -0.0324], |
| [-0.0507, 0.0484, -0.1586, ..., 0.0612, 0.0690, -0.1311], |
| [ 0.0552, -0.0105, -0.0633, ..., -0.0583, -0.0413, 0.0129], |
| ..., |
| [ 0.0573, -0.0561, -0.0134, ..., 0.0180, -0.0395, 0.0189], |
| [-0.0629, -0.0313, -0.0803, ..., -0.0219, -0.0337, -0.0657], |
| [-0.1417, -0.1972, 0.0308, ..., -0.0612, 0.0386, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 6.6124e-08, 0.0000e+00, -1.5749e-06, ..., 0.0000e+00, |
| 3.7253e-08, 2.7940e-09], |
| [ 4.4666e-06, 0.0000e+00, 4.6566e-09, ..., 0.0000e+00, |
| 9.4716e-07, 4.6566e-08], |
| [-1.3016e-05, 0.0000e+00, 1.2107e-08, ..., 0.0000e+00, |
| -5.3458e-06, 1.8626e-08], |
| ..., |
| [-8.2105e-06, 0.0000e+00, 9.3132e-09, ..., 0.0000e+00, |
| -1.1446e-06, -2.3935e-07], |
| [ 1.4435e-07, 0.0000e+00, 8.3819e-08, ..., 0.0000e+00, |
| 1.4994e-07, 8.3819e-09], |
| [ 1.2061e-06, 0.0000e+00, 4.7218e-07, ..., 0.0000e+00, |
| -1.2759e-07, 5.5879e-08]], device='cuda:0') |
| Epoch 194, bias, value: tensor([-0.0071, 0.0228, -0.0029, -0.0041, 0.0121, 0.0047, 0.0248, -0.0108, |
| 0.0068, -0.0035], device='cuda:0'), grad: tensor([-8.2403e-06, 1.9744e-05, -8.9824e-05, 6.2846e-06, 4.1910e-06, |
| 1.0291e-06, 9.0480e-05, -2.6152e-05, 2.3749e-07, 2.1402e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 193, time 217.64, cls_loss 0.0020 cls_loss_mapping 0.0040 cls_loss_causal 0.5150 re_mapping 0.0060 re_causal 0.0186 /// teacc 98.87 lr 0.00010000 |
| Epoch 195, weight, value: tensor([[-0.0560, -0.1023, -0.0593, ..., -0.1092, -0.0187, -0.0324], |
| [-0.0494, 0.0485, -0.1588, ..., 0.0612, 0.0702, -0.1329], |
| [ 0.0555, -0.0100, -0.0633, ..., -0.0587, -0.0413, 0.0128], |
| ..., |
| [ 0.0564, -0.0569, -0.0131, ..., 0.0187, -0.0407, 0.0198], |
| [-0.0635, -0.0286, -0.0805, ..., -0.0237, -0.0343, -0.0658], |
| [-0.1417, -0.1975, 0.0304, ..., -0.0614, 0.0388, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 3.0734e-08, 0.0000e+00, 4.4238e-08, ..., 1.1176e-08, |
| 1.5907e-06, 0.0000e+00], |
| [ 5.3504e-07, 9.3132e-10, 1.5181e-07, ..., 2.6077e-08, |
| -6.4790e-05, 0.0000e+00], |
| [ 1.3132e-07, 0.0000e+00, 4.2375e-08, ..., 1.0710e-08, |
| 3.1441e-05, 0.0000e+00], |
| ..., |
| [-1.4473e-06, -1.8626e-09, 2.0768e-07, ..., 5.3551e-08, |
| 1.1120e-06, 0.0000e+00], |
| [ 9.9652e-08, 0.0000e+00, 5.8208e-08, ..., 5.1223e-09, |
| 2.7835e-05, 0.0000e+00], |
| [ 2.9476e-07, 0.0000e+00, 1.7919e-06, ..., 3.1339e-07, |
| 8.3353e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 195, bias, value: tensor([-0.0069, 0.0242, -0.0026, -0.0042, 0.0122, 0.0045, 0.0248, -0.0119, |
| 0.0071, -0.0037], device='cuda:0'), grad: tensor([ 3.6284e-06, -1.4067e-04, 6.9082e-05, 3.7532e-07, -6.2436e-06, |
| 1.4976e-06, 5.6848e-06, 7.6927e-07, 6.1214e-05, 4.6268e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 194, time 217.62, cls_loss 0.0022 cls_loss_mapping 0.0035 cls_loss_causal 0.5276 re_mapping 0.0060 re_causal 0.0191 /// teacc 98.90 lr 0.00010000 |
| Epoch 196, weight, value: tensor([[-0.0561, -0.1033, -0.0594, ..., -0.1103, -0.0213, -0.0324], |
| [-0.0495, 0.0485, -0.1590, ..., 0.0612, 0.0703, -0.1334], |
| [ 0.0563, -0.0076, -0.0632, ..., -0.0588, -0.0417, 0.0128], |
| ..., |
| [ 0.0562, -0.0595, -0.0136, ..., 0.0187, -0.0407, 0.0200], |
| [-0.0639, -0.0286, -0.0806, ..., -0.0242, -0.0348, -0.0658], |
| [-0.1419, -0.1980, 0.0303, ..., -0.0613, 0.0401, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 4.9360e-08, 0.0000e+00, 3.4925e-08, ..., 8.3819e-09, |
| 1.4529e-07, 0.0000e+00], |
| [ 3.1758e-07, 0.0000e+00, 9.6858e-08, ..., 1.7695e-08, |
| -3.7719e-08, 0.0000e+00], |
| [-5.3868e-06, 0.0000e+00, -8.4145e-07, ..., 1.8626e-08, |
| 1.7695e-08, 0.0000e+00], |
| ..., |
| [-4.7218e-07, 0.0000e+00, 3.5018e-07, ..., 1.0291e-07, |
| 3.1199e-08, 0.0000e+00], |
| [ 6.6124e-08, 0.0000e+00, -3.3341e-07, ..., 1.8626e-09, |
| -3.5390e-08, 0.0000e+00], |
| [ 5.2154e-08, 0.0000e+00, 5.9092e-07, ..., 1.1642e-08, |
| -2.6450e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 196, bias, value: tensor([-0.0090, 0.0242, -0.0018, -0.0041, 0.0124, 0.0045, 0.0242, -0.0122, |
| 0.0069, -0.0023], device='cuda:0'), grad: tensor([ 8.5980e-06, 1.0505e-06, -6.4299e-06, 1.0617e-06, -1.9446e-06, |
| 8.2701e-06, -2.0280e-05, -5.0105e-07, 8.1286e-06, 2.0191e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 195, time 218.00, cls_loss 0.0015 cls_loss_mapping 0.0026 cls_loss_causal 0.5352 re_mapping 0.0064 re_causal 0.0190 /// teacc 98.94 lr 0.00010000 |
| Epoch 197, weight, value: tensor([[-0.0559, -0.1038, -0.0593, ..., -0.1106, -0.0215, -0.0324], |
| [-0.0495, 0.0485, -0.1592, ..., 0.0611, 0.0705, -0.1337], |
| [ 0.0567, -0.0072, -0.0629, ..., -0.0589, -0.0419, 0.0133], |
| ..., |
| [ 0.0562, -0.0598, -0.0135, ..., 0.0187, -0.0408, 0.0193], |
| [-0.0646, -0.0285, -0.0807, ..., -0.0245, -0.0362, -0.0660], |
| [-0.1419, -0.1988, 0.0303, ..., -0.0610, 0.0402, 0.0478]], |
| device='cuda:0'), grad: tensor([[-1.1688e-07, 0.0000e+00, 3.8883e-07, ..., 1.8626e-09, |
| 6.7987e-08, 0.0000e+00], |
| [ 3.2177e-07, 0.0000e+00, 2.2305e-07, ..., 9.7789e-09, |
| -8.3353e-08, 0.0000e+00], |
| [ 7.3528e-07, 0.0000e+00, 1.4482e-07, ..., 2.3283e-09, |
| 1.9558e-07, 0.0000e+00], |
| ..., |
| [ 8.3167e-07, 0.0000e+00, 7.4971e-07, ..., 4.7963e-08, |
| 1.1129e-07, 0.0000e+00], |
| [ 5.8115e-07, 0.0000e+00, 1.7490e-06, ..., 4.6566e-10, |
| 5.6811e-08, 0.0000e+00], |
| [ 1.5358e-06, 0.0000e+00, 4.4517e-06, ..., 1.3970e-08, |
| 1.0617e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 197, bias, value: tensor([-0.0087, 0.0243, -0.0016, -0.0041, 0.0123, 0.0043, 0.0238, -0.0123, |
| 0.0063, -0.0023], device='cuda:0'), grad: tensor([-1.9390e-06, 4.6426e-07, 2.5649e-06, 7.7128e-05, 1.9874e-06, |
| -9.3937e-05, -6.4820e-06, 2.4643e-06, 4.3772e-06, 1.3426e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 196, time 217.69, cls_loss 0.0016 cls_loss_mapping 0.0029 cls_loss_causal 0.5102 re_mapping 0.0061 re_causal 0.0183 /// teacc 98.95 lr 0.00010000 |
| Epoch 198, weight, value: tensor([[-0.0553, -0.1042, -0.0594, ..., -0.1115, -0.0203, -0.0324], |
| [-0.0496, 0.0486, -0.1594, ..., 0.0614, 0.0704, -0.1337], |
| [ 0.0570, -0.0072, -0.0629, ..., -0.0589, -0.0423, 0.0133], |
| ..., |
| [ 0.0562, -0.0599, -0.0138, ..., 0.0184, -0.0408, 0.0193], |
| [-0.0661, -0.0282, -0.0810, ..., -0.0248, -0.0366, -0.0660], |
| [-0.1421, -0.1994, 0.0307, ..., -0.0607, 0.0405, 0.0478]], |
| device='cuda:0'), grad: tensor([[-4.8168e-06, 0.0000e+00, 1.0710e-08, ..., 0.0000e+00, |
| -5.8264e-06, 0.0000e+00], |
| [ 2.8554e-06, 9.3132e-10, 8.2888e-08, ..., 4.6566e-10, |
| 1.5646e-06, 0.0000e+00], |
| [ 3.4273e-06, -4.1910e-09, 1.2573e-08, ..., 0.0000e+00, |
| 6.5286e-07, 0.0000e+00], |
| ..., |
| [-4.3064e-06, 2.7940e-09, 4.4703e-08, ..., 9.3132e-10, |
| 3.6322e-07, 0.0000e+00], |
| [ 4.5309e-07, 0.0000e+00, 1.7229e-08, ..., 0.0000e+00, |
| 1.2247e-07, 4.6566e-10], |
| [ 2.3469e-07, 0.0000e+00, 2.4401e-07, ..., 2.3283e-09, |
| 2.3423e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 198, bias, value: tensor([-0.0077, 0.0241, -0.0015, -0.0044, 0.0123, 0.0042, 0.0239, -0.0124, |
| 0.0050, -0.0023], device='cuda:0'), grad: tensor([-3.3975e-05, 1.3359e-05, 7.4357e-06, 1.5944e-06, 3.2457e-07, |
| -1.4855e-07, 1.3851e-05, -5.1409e-06, 9.0431e-07, 1.8040e-06], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 197, time 217.76, cls_loss 0.0023 cls_loss_mapping 0.0036 cls_loss_causal 0.5214 re_mapping 0.0064 re_causal 0.0179 /// teacc 98.89 lr 0.00010000 |
| Epoch 199, weight, value: tensor([[-0.0551, -0.1055, -0.0586, ..., -0.1120, -0.0202, -0.0324], |
| [-0.0494, 0.0486, -0.1595, ..., 0.0612, 0.0706, -0.1345], |
| [ 0.0567, -0.0073, -0.0626, ..., -0.0592, -0.0431, 0.0136], |
| ..., |
| [ 0.0563, -0.0597, -0.0141, ..., 0.0190, -0.0409, 0.0195], |
| [-0.0657, -0.0283, -0.0811, ..., -0.0251, -0.0361, -0.0663], |
| [-0.1426, -0.2014, 0.0302, ..., -0.0626, 0.0406, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 7.3668e-07, 9.3132e-10, 1.8813e-07, ..., 0.0000e+00, |
| 2.0210e-07, 1.6764e-08], |
| [ 3.2410e-06, 2.5146e-08, 7.4226e-07, ..., 9.3132e-10, |
| 2.8964e-07, 7.4506e-09], |
| [ 4.4592e-06, 1.7695e-08, 9.4436e-07, ..., 9.3132e-10, |
| 7.5996e-07, 1.1176e-08], |
| ..., |
| [-2.0385e-05, -6.7055e-08, 1.2452e-06, ..., -1.2107e-08, |
| -4.1537e-06, 2.3842e-07], |
| [-1.3702e-05, 9.3132e-09, -6.6943e-06, ..., 0.0000e+00, |
| -1.0058e-05, 9.3132e-09], |
| [ 1.7971e-05, 7.4506e-09, 8.9128e-07, ..., 0.0000e+00, |
| 1.1697e-05, -2.9430e-07]], device='cuda:0') |
| Epoch 199, bias, value: tensor([-0.0066, 0.0242, -0.0018, -0.0045, 0.0124, 0.0046, 0.0243, -0.0122, |
| 0.0054, -0.0036], device='cuda:0'), grad: tensor([ 2.0191e-06, 9.6858e-06, 1.3098e-05, 1.1854e-05, 1.3299e-05, |
| 8.8010e-07, -1.8626e-08, -5.0873e-05, -9.9599e-05, 9.9659e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 198, time 217.76, cls_loss 0.0025 cls_loss_mapping 0.0038 cls_loss_causal 0.5459 re_mapping 0.0061 re_causal 0.0182 /// teacc 98.90 lr 0.00010000 |
| Epoch 200, weight, value: tensor([[-0.0557, -0.1071, -0.0590, ..., -0.1147, -0.0204, -0.0324], |
| [-0.0494, 0.0491, -0.1600, ..., 0.0614, 0.0709, -0.1355], |
| [ 0.0565, -0.0077, -0.0625, ..., -0.0595, -0.0434, 0.0117], |
| ..., |
| [ 0.0565, -0.0596, -0.0155, ..., 0.0187, -0.0410, 0.0189], |
| [-0.0661, -0.0266, -0.0825, ..., -0.0281, -0.0365, -0.0693], |
| [-0.1430, -0.2037, 0.0285, ..., -0.0653, 0.0397, 0.0478]], |
| device='cuda:0'), grad: tensor([[ 5.6811e-08, 9.3132e-10, 3.4459e-08, ..., 1.4622e-07, |
| 6.3330e-07, 2.7940e-09], |
| [ 6.2399e-07, 9.3132e-09, 1.8626e-08, ..., -4.4346e-05, |
| -5.9754e-05, 2.8871e-08], |
| [-1.0710e-06, 2.7940e-09, 1.8626e-09, ..., 1.0431e-07, |
| 2.7847e-07, 7.4506e-09], |
| ..., |
| [ 5.6531e-07, 7.4506e-09, 2.7940e-08, ..., 3.6180e-05, |
| 4.7743e-05, 2.3283e-08], |
| [ 2.6356e-07, 2.0489e-08, 1.4808e-07, ..., 2.7940e-09, |
| 2.4121e-07, 5.5879e-08], |
| [ 2.1979e-07, 1.6764e-08, 1.3784e-07, ..., 4.5113e-06, |
| 5.9046e-06, 5.1223e-08]], device='cuda:0') |
| Epoch 200, bias, value: tensor([-0.0068, 0.0244, -0.0021, -0.0049, 0.0147, 0.0054, 0.0251, -0.0122, |
| 0.0047, -0.0056], device='cuda:0'), grad: tensor([ 1.7108e-06, -2.0659e-04, -7.8697e-07, -6.5029e-05, 1.6347e-05, |
| 6.3658e-05, 5.0478e-07, 1.6820e-04, 1.1297e-06, 2.1100e-05], |
| device='cuda:0') |
| 100 |
| 0.0001 |
| changing lr |
| epoch 199, time 217.80, cls_loss 0.0022 cls_loss_mapping 0.0039 cls_loss_causal 0.5363 re_mapping 0.0064 re_causal 0.0190 /// teacc 98.85 lr 0.00001000 |
| Epoch 201, weight, value: tensor([[-0.0558, -0.1075, -0.0597, ..., -0.1150, -0.0204, -0.0324], |
| [-0.0494, 0.0489, -0.1612, ..., 0.0615, 0.0709, -0.1365], |
| [ 0.0563, -0.0076, -0.0633, ..., -0.0596, -0.0440, 0.0089], |
| ..., |
| [ 0.0566, -0.0597, -0.0167, ..., 0.0187, -0.0411, 0.0182], |
| [-0.0664, -0.0264, -0.0824, ..., -0.0285, -0.0351, -0.0708], |
| [-0.1432, -0.2039, 0.0286, ..., -0.0658, 0.0400, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.5739e-07, 0.0000e+00, 0.0000e+00, ..., 0.0000e+00, |
| 9.0338e-08, 0.0000e+00], |
| [ 4.1984e-06, 0.0000e+00, 3.4180e-07, ..., 0.0000e+00, |
| 1.9372e-06, 0.0000e+00], |
| [ 2.9579e-06, 0.0000e+00, 1.6764e-08, ..., 0.0000e+00, |
| 2.7195e-06, 0.0000e+00], |
| ..., |
| [-4.7237e-05, 0.0000e+00, 3.6992e-06, ..., 0.0000e+00, |
| -2.3842e-05, -9.3132e-10], |
| [ 5.6103e-06, 0.0000e+00, 5.3570e-06, ..., 0.0000e+00, |
| 3.3602e-06, 9.3132e-10], |
| [ 3.3200e-05, 0.0000e+00, -2.1994e-05, ..., 0.0000e+00, |
| 1.4156e-05, 0.0000e+00]], device='cuda:0') |
| Epoch 201, bias, value: tensor([-0.0068, 0.0244, -0.0026, -0.0052, 0.0149, 0.0056, 0.0250, -0.0123, |
| 0.0057, -0.0054], device='cuda:0'), grad: tensor([ 1.4668e-06, 3.3379e-05, 4.2379e-05, 1.5264e-06, 3.7491e-05, |
| 9.4026e-06, 9.5926e-08, -4.4703e-04, 7.0095e-05, 2.5082e-04], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 200, time 217.27, cls_loss 0.0020 cls_loss_mapping 0.0034 cls_loss_causal 0.5173 re_mapping 0.0064 re_causal 0.0187 /// teacc 98.95 lr 0.00001000 |
| Epoch 202, weight, value: tensor([[-0.0558, -0.1076, -0.0598, ..., -0.1151, -0.0204, -0.0325], |
| [-0.0495, 0.0490, -0.1613, ..., 0.0615, 0.0709, -0.1365], |
| [ 0.0563, -0.0076, -0.0633, ..., -0.0596, -0.0440, 0.0089], |
| ..., |
| [ 0.0566, -0.0597, -0.0167, ..., 0.0187, -0.0411, 0.0181], |
| [-0.0664, -0.0263, -0.0824, ..., -0.0285, -0.0350, -0.0709], |
| [-0.1433, -0.2039, 0.0287, ..., -0.0659, 0.0400, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 7.7300e-08, 0.0000e+00, 0.0000e+00, ..., 0.0000e+00, |
| 1.7695e-08, 0.0000e+00], |
| [ 1.8533e-07, 0.0000e+00, 4.6566e-09, ..., 0.0000e+00, |
| -4.2841e-08, 0.0000e+00], |
| [-4.6752e-07, 0.0000e+00, -1.2107e-08, ..., 0.0000e+00, |
| 1.9558e-08, 0.0000e+00], |
| ..., |
| [ 1.5115e-06, 0.0000e+00, 3.7253e-09, ..., 0.0000e+00, |
| 3.6787e-07, 0.0000e+00], |
| [ 7.6368e-08, 0.0000e+00, 3.7253e-09, ..., 0.0000e+00, |
| 4.6566e-08, 0.0000e+00], |
| [ 1.4808e-07, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| -7.7114e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 202, bias, value: tensor([-0.0068, 0.0244, -0.0026, -0.0051, 0.0148, 0.0056, 0.0250, -0.0123, |
| 0.0058, -0.0054], device='cuda:0'), grad: tensor([-2.4475e-06, 6.0629e-07, -1.9027e-06, -1.8049e-06, 1.6913e-06, |
| 2.6543e-07, -1.1642e-07, 2.5313e-06, 5.5693e-07, 6.2771e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 201, time 217.37, cls_loss 0.0018 cls_loss_mapping 0.0024 cls_loss_causal 0.5187 re_mapping 0.0058 re_causal 0.0183 /// teacc 98.99 lr 0.00001000 |
| Epoch 203, weight, value: tensor([[-0.0558, -0.1076, -0.0598, ..., -0.1151, -0.0204, -0.0325], |
| [-0.0496, 0.0490, -0.1613, ..., 0.0615, 0.0709, -0.1366], |
| [ 0.0563, -0.0076, -0.0633, ..., -0.0596, -0.0441, 0.0089], |
| ..., |
| [ 0.0567, -0.0597, -0.0168, ..., 0.0187, -0.0411, 0.0181], |
| [-0.0664, -0.0263, -0.0824, ..., -0.0285, -0.0350, -0.0709], |
| [-0.1433, -0.2040, 0.0286, ..., -0.0659, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.7323e-07, 9.3132e-10, 6.5193e-09, ..., 0.0000e+00, |
| 1.4901e-08, 0.0000e+00], |
| [ 4.9453e-07, -1.0338e-07, 7.1153e-07, ..., 0.0000e+00, |
| -8.5682e-08, 0.0000e+00], |
| [-5.2713e-06, 6.7987e-08, 3.0734e-08, ..., 0.0000e+00, |
| 5.8673e-08, 0.0000e+00], |
| ..., |
| [ 6.1467e-07, 2.0489e-08, 9.0525e-07, ..., 0.0000e+00, |
| 7.7765e-07, 0.0000e+00], |
| [ 3.8445e-06, 9.3132e-10, 1.1176e-08, ..., 0.0000e+00, |
| 3.2596e-08, 0.0000e+00], |
| [-4.9546e-07, 0.0000e+00, -9.7603e-07, ..., 0.0000e+00, |
| -1.0189e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 203, bias, value: tensor([-0.0068, 0.0242, -0.0026, -0.0051, 0.0148, 0.0055, 0.0249, -0.0122, |
| 0.0058, -0.0054], device='cuda:0'), grad: tensor([-4.6939e-07, 1.6829e-06, -1.3433e-05, 6.0815e-07, 1.9409e-06, |
| 5.5321e-07, 5.7556e-07, 1.0334e-05, 9.3132e-06, -1.1131e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 202, time 217.21, cls_loss 0.0014 cls_loss_mapping 0.0016 cls_loss_causal 0.5039 re_mapping 0.0056 re_causal 0.0177 /// teacc 99.04 lr 0.00001000 |
| Epoch 204, weight, value: tensor([[-0.0558, -0.1077, -0.0599, ..., -0.1151, -0.0204, -0.0325], |
| [-0.0496, 0.0490, -0.1613, ..., 0.0615, 0.0709, -0.1367], |
| [ 0.0564, -0.0076, -0.0633, ..., -0.0596, -0.0441, 0.0089], |
| ..., |
| [ 0.0567, -0.0597, -0.0169, ..., 0.0187, -0.0411, 0.0181], |
| [-0.0665, -0.0264, -0.0824, ..., -0.0286, -0.0350, -0.0709], |
| [-0.1433, -0.2040, 0.0286, ..., -0.0659, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.8626e-09, 0.0000e+00, 5.1223e-08, ..., 9.3132e-10, |
| 2.7940e-08, 9.3132e-10], |
| [ 4.9360e-08, 0.0000e+00, 3.1907e-06, ..., 9.3132e-09, |
| 7.2643e-08, 0.0000e+00], |
| [-1.9558e-08, 0.0000e+00, 1.0431e-07, ..., 1.8626e-09, |
| 3.5390e-08, 0.0000e+00], |
| ..., |
| [-1.9558e-08, 0.0000e+00, 2.5444e-06, ..., 4.7497e-08, |
| 2.9523e-07, 0.0000e+00], |
| [ 9.3132e-10, 0.0000e+00, 1.5087e-07, ..., 0.0000e+00, |
| 1.4342e-07, 1.8626e-09], |
| [ 2.7940e-09, 0.0000e+00, 6.3144e-06, ..., 5.5879e-09, |
| 1.6391e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 204, bias, value: tensor([-0.0068, 0.0242, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0121, |
| 0.0058, -0.0054], device='cuda:0'), grad: tensor([ 2.1327e-07, 1.0818e-05, 3.7812e-07, 1.4342e-07, -3.7342e-05, |
| 6.9756e-07, -8.4843e-07, 7.5474e-06, 9.1828e-07, 1.7449e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 203, time 217.32, cls_loss 0.0016 cls_loss_mapping 0.0016 cls_loss_causal 0.4828 re_mapping 0.0055 re_causal 0.0171 /// teacc 99.02 lr 0.00001000 |
| Epoch 205, weight, value: tensor([[-0.0558, -0.1078, -0.0599, ..., -0.1151, -0.0205, -0.0325], |
| [-0.0497, 0.0490, -0.1614, ..., 0.0616, 0.0709, -0.1367], |
| [ 0.0564, -0.0076, -0.0633, ..., -0.0596, -0.0441, 0.0088], |
| ..., |
| [ 0.0568, -0.0597, -0.0169, ..., 0.0186, -0.0412, 0.0181], |
| [-0.0665, -0.0264, -0.0824, ..., -0.0286, -0.0351, -0.0709], |
| [-0.1433, -0.2040, 0.0286, ..., -0.0659, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.0338e-07, 0.0000e+00, 1.4994e-07, ..., 0.0000e+00, |
| 5.2154e-08, 0.0000e+00], |
| [ 2.0210e-07, 0.0000e+00, 3.5390e-08, ..., 9.3132e-10, |
| 1.6391e-07, 0.0000e+00], |
| [-5.2713e-07, 0.0000e+00, 3.6322e-08, ..., 0.0000e+00, |
| 1.4994e-07, 0.0000e+00], |
| ..., |
| [ 1.8626e-07, 0.0000e+00, 9.1270e-08, ..., 9.3132e-10, |
| 1.8720e-07, 0.0000e+00], |
| [ 4.9919e-07, 0.0000e+00, 1.4333e-06, ..., 0.0000e+00, |
| 1.2573e-07, 0.0000e+00], |
| [ 2.1979e-07, 0.0000e+00, -1.7742e-06, ..., 3.7253e-09, |
| -1.3188e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 205, bias, value: tensor([-0.0068, 0.0241, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0121, |
| 0.0058, -0.0054], device='cuda:0'), grad: tensor([-2.7567e-07, 1.3702e-05, 5.9698e-07, 2.7511e-06, 1.5814e-06, |
| -2.4289e-06, 1.5087e-07, 1.3495e-06, -1.0237e-05, -7.2308e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 204, time 217.47, cls_loss 0.0019 cls_loss_mapping 0.0018 cls_loss_causal 0.4965 re_mapping 0.0054 re_causal 0.0175 /// teacc 99.00 lr 0.00001000 |
| Epoch 206, weight, value: tensor([[-0.0558, -0.1078, -0.0599, ..., -0.1152, -0.0205, -0.0325], |
| [-0.0499, 0.0491, -0.1614, ..., 0.0618, 0.0710, -0.1369], |
| [ 0.0564, -0.0076, -0.0633, ..., -0.0596, -0.0441, 0.0088], |
| ..., |
| [ 0.0569, -0.0597, -0.0169, ..., 0.0184, -0.0412, 0.0182], |
| [-0.0665, -0.0264, -0.0823, ..., -0.0287, -0.0351, -0.0709], |
| [-0.1433, -0.2040, 0.0285, ..., -0.0660, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.4435e-07, 0.0000e+00, -2.6543e-07, ..., 0.0000e+00, |
| 5.4948e-08, 0.0000e+00], |
| [ 2.2724e-07, 0.0000e+00, 8.1025e-08, ..., 0.0000e+00, |
| -5.5879e-08, 0.0000e+00], |
| [-7.2550e-07, 0.0000e+00, 1.7695e-08, ..., 0.0000e+00, |
| 1.7695e-08, 0.0000e+00], |
| ..., |
| [ 3.0734e-08, 0.0000e+00, 3.8184e-08, ..., 0.0000e+00, |
| 4.6566e-08, 0.0000e+00], |
| [ 1.6149e-06, 0.0000e+00, 7.2643e-08, ..., 0.0000e+00, |
| 3.7253e-08, 0.0000e+00], |
| [ 9.3132e-08, 0.0000e+00, 6.8545e-07, ..., 0.0000e+00, |
| -6.2399e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 206, bias, value: tensor([-0.0068, 0.0240, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0120, |
| 0.0059, -0.0054], device='cuda:0'), grad: tensor([-8.0373e-07, 6.0908e-07, -6.3889e-07, -2.4177e-06, 5.5283e-06, |
| 5.7369e-07, -7.8827e-06, -1.6950e-07, 3.4273e-06, 1.7742e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 205, time 217.63, cls_loss 0.0015 cls_loss_mapping 0.0017 cls_loss_causal 0.5079 re_mapping 0.0053 re_causal 0.0177 /// teacc 99.02 lr 0.00001000 |
| Epoch 207, weight, value: tensor([[-0.0558, -0.1079, -0.0599, ..., -0.1153, -0.0205, -0.0325], |
| [-0.0500, 0.0491, -0.1614, ..., 0.0618, 0.0710, -0.1369], |
| [ 0.0564, -0.0076, -0.0634, ..., -0.0596, -0.0441, 0.0088], |
| ..., |
| [ 0.0569, -0.0597, -0.0170, ..., 0.0184, -0.0412, 0.0181], |
| [-0.0665, -0.0264, -0.0823, ..., -0.0288, -0.0351, -0.0709], |
| [-0.1434, -0.2041, 0.0284, ..., -0.0660, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 9.0338e-08, 0.0000e+00, 7.4506e-09, ..., 0.0000e+00, |
| 3.1758e-07, 0.0000e+00], |
| [ 3.8277e-07, 0.0000e+00, 5.4948e-08, ..., 0.0000e+00, |
| -1.8235e-06, 0.0000e+00], |
| [ 9.8720e-07, 0.0000e+00, -7.4506e-09, ..., 0.0000e+00, |
| 1.7416e-07, 0.0000e+00], |
| ..., |
| [-2.5183e-06, 0.0000e+00, 2.4308e-07, ..., 0.0000e+00, |
| 7.6741e-07, 0.0000e+00], |
| [ 8.7544e-08, 0.0000e+00, 7.4506e-09, ..., 0.0000e+00, |
| 1.0710e-07, 0.0000e+00], |
| [ 1.8813e-07, 0.0000e+00, 2.6729e-07, ..., 1.8626e-09, |
| 8.2888e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 207, bias, value: tensor([-0.0068, 0.0239, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0059, -0.0054], device='cuda:0'), grad: tensor([-2.4647e-05, -2.9095e-06, 3.4720e-06, 1.2005e-06, -3.2783e-07, |
| 1.5674e-06, 2.0713e-05, -2.5723e-06, 2.7046e-06, 8.3167e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 206, time 217.72, cls_loss 0.0016 cls_loss_mapping 0.0014 cls_loss_causal 0.5105 re_mapping 0.0053 re_causal 0.0176 /// teacc 99.01 lr 0.00001000 |
| Epoch 208, weight, value: tensor([[-0.0559, -0.1079, -0.0599, ..., -0.1154, -0.0205, -0.0325], |
| [-0.0500, 0.0492, -0.1614, ..., 0.0620, 0.0711, -0.1369], |
| [ 0.0564, -0.0076, -0.0634, ..., -0.0597, -0.0443, 0.0088], |
| ..., |
| [ 0.0569, -0.0597, -0.0170, ..., 0.0183, -0.0412, 0.0181], |
| [-0.0665, -0.0264, -0.0823, ..., -0.0288, -0.0352, -0.0709], |
| [-0.1434, -0.2041, 0.0284, ..., -0.0660, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 2.1420e-08, 0.0000e+00, -1.3039e-08, ..., 2.7940e-09, |
| 5.8673e-08, 0.0000e+00], |
| [ 8.0932e-07, -1.1176e-08, 1.8626e-09, ..., 1.2573e-07, |
| -8.9407e-07, 0.0000e+00], |
| [ 6.7055e-08, 1.8626e-09, 9.3132e-10, ..., 5.5879e-09, |
| 7.6368e-08, 0.0000e+00], |
| ..., |
| [-2.0042e-06, 7.4506e-09, 2.7940e-09, ..., -3.2596e-07, |
| 3.5763e-07, 0.0000e+00], |
| [ 1.3039e-08, 0.0000e+00, 1.7695e-08, ..., 9.3132e-10, |
| 6.9849e-08, 0.0000e+00], |
| [ 9.5740e-07, 0.0000e+00, 4.6566e-09, ..., 1.7416e-07, |
| 1.1176e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 208, bias, value: tensor([-0.0067, 0.0239, -0.0027, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0059, -0.0055], device='cuda:0'), grad: tensor([ 4.1630e-07, -6.7987e-07, 1.2973e-06, 3.9116e-07, 1.0571e-06, |
| 3.4366e-07, -2.8312e-07, -3.2913e-06, -1.4212e-06, 2.1700e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 207, time 217.40, cls_loss 0.0015 cls_loss_mapping 0.0014 cls_loss_causal 0.4776 re_mapping 0.0052 re_causal 0.0165 /// teacc 99.04 lr 0.00001000 |
| Epoch 209, weight, value: tensor([[-0.0559, -0.1080, -0.0599, ..., -0.1154, -0.0205, -0.0325], |
| [-0.0501, 0.0492, -0.1615, ..., 0.0620, 0.0711, -0.1369], |
| [ 0.0564, -0.0076, -0.0634, ..., -0.0597, -0.0443, 0.0088], |
| ..., |
| [ 0.0570, -0.0597, -0.0170, ..., 0.0183, -0.0412, 0.0181], |
| [-0.0665, -0.0265, -0.0823, ..., -0.0289, -0.0352, -0.0709], |
| [-0.1434, -0.2041, 0.0284, ..., -0.0661, 0.0401, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 3.7253e-08, 9.3132e-10, 6.7055e-08, ..., 0.0000e+00, |
| 8.1956e-08, 0.0000e+00], |
| [ 9.4716e-07, -3.9116e-08, 1.1083e-07, ..., 1.1176e-08, |
| -1.5832e-08, 0.0000e+00], |
| [ 2.2072e-07, 1.7695e-08, 2.9802e-08, ..., 9.3132e-10, |
| 1.3690e-07, 0.0000e+00], |
| ..., |
| [-1.1697e-06, 8.3819e-09, 7.5437e-08, ..., 3.7253e-09, |
| 1.8068e-07, 0.0000e+00], |
| [ 6.8918e-08, 0.0000e+00, 2.4680e-07, ..., 9.3132e-10, |
| 6.1560e-07, 0.0000e+00], |
| [ 4.3400e-07, 0.0000e+00, 1.0245e-07, ..., 7.4506e-09, |
| -1.7835e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 209, bias, value: tensor([-0.0068, 0.0239, -0.0027, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 4.2096e-07, 1.8552e-06, 1.0254e-06, 2.9095e-06, 1.0394e-06, |
| -2.1867e-06, -9.6019e-07, -1.7770e-06, 1.6447e-06, -3.9972e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 208, time 217.22, cls_loss 0.0015 cls_loss_mapping 0.0013 cls_loss_causal 0.4651 re_mapping 0.0050 re_causal 0.0163 /// teacc 99.05 lr 0.00001000 |
| Epoch 210, weight, value: tensor([[-0.0559, -0.1080, -0.0600, ..., -0.1156, -0.0205, -0.0325], |
| [-0.0501, 0.0492, -0.1615, ..., 0.0621, 0.0711, -0.1370], |
| [ 0.0564, -0.0076, -0.0634, ..., -0.0597, -0.0444, 0.0088], |
| ..., |
| [ 0.0570, -0.0597, -0.0170, ..., 0.0182, -0.0413, 0.0181], |
| [-0.0666, -0.0264, -0.0823, ..., -0.0289, -0.0352, -0.0709], |
| [-0.1434, -0.2041, 0.0284, ..., -0.0660, 0.0402, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 3.1237e-06, 0.0000e+00, 8.3819e-09, ..., 0.0000e+00, |
| 7.4506e-09, 0.0000e+00], |
| [ 6.9197e-07, 0.0000e+00, 5.2154e-08, ..., 0.0000e+00, |
| -3.7253e-08, 0.0000e+00], |
| [-9.0972e-06, 0.0000e+00, 2.0489e-08, ..., 0.0000e+00, |
| 1.8626e-08, 0.0000e+00], |
| ..., |
| [-3.3807e-07, 0.0000e+00, 2.0489e-08, ..., 0.0000e+00, |
| -8.4750e-08, 0.0000e+00], |
| [ 2.5462e-06, 0.0000e+00, -3.8091e-07, ..., 0.0000e+00, |
| 1.1176e-08, 0.0000e+00], |
| [ 1.1455e-07, 0.0000e+00, 2.3376e-07, ..., 0.0000e+00, |
| -1.2107e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 210, bias, value: tensor([-0.0068, 0.0239, -0.0027, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0061, -0.0055], device='cuda:0'), grad: tensor([ 1.0259e-05, 1.9968e-06, -3.1561e-05, 6.5044e-06, 1.6205e-07, |
| 7.1432e-07, 4.2766e-06, -6.2864e-07, 6.6943e-06, 1.5358e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 209, time 217.24, cls_loss 0.0013 cls_loss_mapping 0.0013 cls_loss_causal 0.5117 re_mapping 0.0050 re_causal 0.0171 /// teacc 99.05 lr 0.00001000 |
| Epoch 211, weight, value: tensor([[-0.0559, -0.1080, -0.0600, ..., -0.1156, -0.0205, -0.0325], |
| [-0.0502, 0.0492, -0.1615, ..., 0.0621, 0.0712, -0.1370], |
| [ 0.0564, -0.0076, -0.0634, ..., -0.0597, -0.0444, 0.0088], |
| ..., |
| [ 0.0571, -0.0597, -0.0170, ..., 0.0181, -0.0413, 0.0181], |
| [-0.0666, -0.0265, -0.0823, ..., -0.0290, -0.0353, -0.0709], |
| [-0.1435, -0.2041, 0.0283, ..., -0.0659, 0.0402, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 5.5879e-09, 0.0000e+00, 2.0489e-08, ..., 0.0000e+00, |
| 5.3085e-08, 0.0000e+00], |
| [ 2.4680e-07, 0.0000e+00, 1.0058e-07, ..., 0.0000e+00, |
| -2.2259e-07, 0.0000e+00], |
| [-8.8476e-08, 0.0000e+00, 1.2107e-08, ..., 0.0000e+00, |
| 1.3970e-08, 0.0000e+00], |
| ..., |
| [-3.6787e-07, 0.0000e+00, 6.8918e-08, ..., 0.0000e+00, |
| 1.4622e-07, 0.0000e+00], |
| [ 4.6566e-09, 0.0000e+00, 4.2934e-07, ..., 0.0000e+00, |
| 1.3970e-07, 0.0000e+00], |
| [ 8.4750e-08, 0.0000e+00, 5.0571e-07, ..., 0.0000e+00, |
| 5.5879e-09, 0.0000e+00]], device='cuda:0') |
| Epoch 211, bias, value: tensor([-0.0068, 0.0238, -0.0027, -0.0051, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 1.2852e-07, 4.2189e-07, -2.7940e-08, 9.0338e-07, -1.6801e-06, |
| -7.8510e-07, -7.4785e-07, -3.5949e-07, 3.5483e-07, 1.7742e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 210, time 217.42, cls_loss 0.0013 cls_loss_mapping 0.0014 cls_loss_causal 0.4885 re_mapping 0.0050 re_causal 0.0169 /// teacc 99.05 lr 0.00001000 |
| Epoch 212, weight, value: tensor([[-0.0559, -0.1081, -0.0600, ..., -0.1157, -0.0205, -0.0325], |
| [-0.0502, 0.0492, -0.1615, ..., 0.0624, 0.0713, -0.1371], |
| [ 0.0565, -0.0076, -0.0634, ..., -0.0597, -0.0444, 0.0089], |
| ..., |
| [ 0.0571, -0.0597, -0.0170, ..., 0.0179, -0.0414, 0.0180], |
| [-0.0667, -0.0265, -0.0823, ..., -0.0291, -0.0353, -0.0709], |
| [-0.1435, -0.2042, 0.0283, ..., -0.0659, 0.0402, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.9092e-07, 0.0000e+00, 2.9150e-07, ..., 0.0000e+00, |
| 1.5646e-07, 5.5879e-09], |
| [ 8.0280e-07, 0.0000e+00, 6.0908e-07, ..., 2.7940e-08, |
| -8.4098e-07, 0.0000e+00], |
| [-1.1511e-06, 0.0000e+00, -1.4622e-07, ..., 9.3132e-10, |
| 1.0803e-07, 0.0000e+00], |
| ..., |
| [ 1.0980e-06, 0.0000e+00, 1.1213e-06, ..., 4.6566e-08, |
| 3.0268e-07, 3.7253e-09], |
| [ 6.6124e-08, 0.0000e+00, 2.2352e-08, ..., 0.0000e+00, |
| 1.0123e-06, 0.0000e+00], |
| [ 4.7591e-07, 0.0000e+00, 2.4363e-06, ..., 2.4214e-08, |
| -1.5665e-06, -2.9802e-08]], device='cuda:0') |
| Epoch 212, bias, value: tensor([-0.0067, 0.0239, -0.0027, -0.0051, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 1.0114e-06, 7.6648e-07, -1.4771e-06, 8.3260e-07, -3.3956e-06, |
| -4.9137e-06, 5.3737e-07, 3.7849e-06, 3.8408e-06, -9.9372e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 211, time 217.28, cls_loss 0.0011 cls_loss_mapping 0.0009 cls_loss_causal 0.4808 re_mapping 0.0049 re_causal 0.0169 /// teacc 99.03 lr 0.00001000 |
| Epoch 213, weight, value: tensor([[-0.0559, -0.1081, -0.0600, ..., -0.1158, -0.0205, -0.0325], |
| [-0.0502, 0.0493, -0.1616, ..., 0.0624, 0.0713, -0.1371], |
| [ 0.0565, -0.0076, -0.0634, ..., -0.0597, -0.0444, 0.0089], |
| ..., |
| [ 0.0571, -0.0598, -0.0171, ..., 0.0179, -0.0414, 0.0180], |
| [-0.0667, -0.0265, -0.0823, ..., -0.0292, -0.0354, -0.0709], |
| [-0.1435, -0.2042, 0.0283, ..., -0.0659, 0.0402, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.7695e-08, 0.0000e+00, 5.5879e-09, ..., 0.0000e+00, |
| 6.5193e-08, 0.0000e+00], |
| [ 4.4517e-07, -9.3132e-10, 5.3085e-08, ..., 0.0000e+00, |
| -5.6997e-07, 0.0000e+00], |
| [ 1.8161e-07, 0.0000e+00, 3.5390e-08, ..., 0.0000e+00, |
| 8.5682e-08, 0.0000e+00], |
| ..., |
| [-1.8710e-06, 9.3132e-10, -5.5879e-09, ..., 0.0000e+00, |
| 3.1292e-07, 0.0000e+00], |
| [-1.2387e-07, 0.0000e+00, 1.4901e-08, ..., 0.0000e+00, |
| -1.5646e-07, 0.0000e+00], |
| [ 2.3283e-07, 0.0000e+00, -4.0606e-07, ..., 0.0000e+00, |
| -2.2538e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 213, bias, value: tensor([-0.0067, 0.0239, -0.0027, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 2.2911e-07, -5.0385e-07, 3.4571e-06, 1.9055e-06, 5.1036e-06, |
| 3.6415e-07, 1.2387e-07, -8.7395e-06, -9.7603e-07, -9.7137e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 212, time 217.45, cls_loss 0.0011 cls_loss_mapping 0.0010 cls_loss_causal 0.5021 re_mapping 0.0049 re_causal 0.0171 /// teacc 99.06 lr 0.00001000 |
| Epoch 214, weight, value: tensor([[-0.0559, -0.1081, -0.0601, ..., -0.1159, -0.0205, -0.0325], |
| [-0.0502, 0.0493, -0.1616, ..., 0.0626, 0.0714, -0.1371], |
| [ 0.0565, -0.0076, -0.0634, ..., -0.0598, -0.0444, 0.0089], |
| ..., |
| [ 0.0571, -0.0598, -0.0171, ..., 0.0177, -0.0415, 0.0180], |
| [-0.0667, -0.0266, -0.0823, ..., -0.0293, -0.0354, -0.0709], |
| [-0.1435, -0.2042, 0.0283, ..., -0.0660, 0.0402, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 2.2352e-08, 0.0000e+00, 2.7008e-08, ..., 9.3132e-09, |
| 3.6322e-08, 0.0000e+00], |
| [ 6.1374e-07, 0.0000e+00, 1.4622e-07, ..., -4.5076e-06, |
| -7.3947e-06, 0.0000e+00], |
| [-3.1106e-07, 0.0000e+00, 5.2154e-08, ..., 3.7253e-09, |
| 2.5146e-08, 0.0000e+00], |
| ..., |
| [-6.5286e-07, 0.0000e+00, 7.6275e-07, ..., 2.5872e-06, |
| 4.2431e-06, 0.0000e+00], |
| [ 4.5635e-08, 0.0000e+00, 3.5390e-08, ..., 1.3039e-08, |
| 5.6811e-08, 0.0000e+00], |
| [ 8.7544e-08, 0.0000e+00, 1.4110e-06, ..., 3.1013e-07, |
| 4.1258e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 214, bias, value: tensor([-0.0067, 0.0239, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([-2.6915e-07, -1.7598e-05, -5.1875e-07, 3.6042e-07, -1.0915e-06, |
| -1.1921e-07, 2.6356e-07, 1.2495e-05, 3.7346e-07, 6.0759e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 213, time 217.46, cls_loss 0.0011 cls_loss_mapping 0.0012 cls_loss_causal 0.4840 re_mapping 0.0048 re_causal 0.0169 /// teacc 99.05 lr 0.00001000 |
| Epoch 215, weight, value: tensor([[-0.0559, -0.1082, -0.0601, ..., -0.1160, -0.0205, -0.0325], |
| [-0.0502, 0.0493, -0.1616, ..., 0.0627, 0.0714, -0.1371], |
| [ 0.0565, -0.0076, -0.0634, ..., -0.0598, -0.0445, 0.0088], |
| ..., |
| [ 0.0571, -0.0598, -0.0171, ..., 0.0176, -0.0415, 0.0180], |
| [-0.0668, -0.0267, -0.0823, ..., -0.0294, -0.0354, -0.0709], |
| [-0.1435, -0.2042, 0.0283, ..., -0.0658, 0.0403, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.0245e-08, 0.0000e+00, 2.3935e-07, ..., 0.0000e+00, |
| 6.5193e-09, 0.0000e+00], |
| [ 1.9092e-07, 0.0000e+00, 1.6950e-07, ..., 9.3132e-10, |
| -2.6729e-07, 0.0000e+00], |
| [-5.6811e-08, 0.0000e+00, 4.4703e-08, ..., 0.0000e+00, |
| 2.3283e-08, 0.0000e+00], |
| ..., |
| [-2.8871e-07, 0.0000e+00, 3.1013e-07, ..., 9.3132e-10, |
| 1.8999e-07, 0.0000e+00], |
| [ 8.9407e-08, 0.0000e+00, 1.7202e-06, ..., 0.0000e+00, |
| 2.7940e-08, 0.0000e+00], |
| [ 1.1176e-07, 0.0000e+00, 8.6520e-07, ..., 1.2107e-08, |
| 7.4506e-09, 0.0000e+00]], device='cuda:0') |
| Epoch 215, bias, value: tensor([-0.0067, 0.0239, -0.0026, -0.0051, 0.0149, 0.0055, 0.0248, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 6.1560e-07, 2.0023e-07, 8.5682e-08, 3.1114e-05, -1.6857e-06, |
| -4.1664e-05, 3.9414e-06, 5.8394e-07, 4.5672e-06, 2.3115e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 214, time 217.44, cls_loss 0.0013 cls_loss_mapping 0.0011 cls_loss_causal 0.4966 re_mapping 0.0049 re_causal 0.0167 /// teacc 99.03 lr 0.00001000 |
| Epoch 216, weight, value: tensor([[-0.0559, -0.1082, -0.0601, ..., -0.1161, -0.0206, -0.0325], |
| [-0.0502, 0.0493, -0.1616, ..., 0.0627, 0.0715, -0.1372], |
| [ 0.0566, -0.0076, -0.0634, ..., -0.0598, -0.0445, 0.0088], |
| ..., |
| [ 0.0571, -0.0598, -0.0172, ..., 0.0176, -0.0416, 0.0180], |
| [-0.0668, -0.0267, -0.0823, ..., -0.0295, -0.0355, -0.0709], |
| [-0.1435, -0.2042, 0.0282, ..., -0.0659, 0.0403, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 3.5390e-08, 0.0000e+00, -2.5146e-08, ..., 0.0000e+00, |
| 1.0431e-07, 0.0000e+00], |
| [-1.7695e-07, 0.0000e+00, 6.7055e-08, ..., 1.8626e-09, |
| -2.3693e-06, 0.0000e+00], |
| [-8.1025e-08, 0.0000e+00, 1.8626e-08, ..., 0.0000e+00, |
| 9.8627e-07, 0.0000e+00], |
| ..., |
| [ 4.2375e-07, 0.0000e+00, 2.5798e-07, ..., 9.3132e-10, |
| 1.1530e-06, 0.0000e+00], |
| [ 4.1910e-08, 0.0000e+00, 4.6566e-09, ..., 0.0000e+00, |
| 4.5728e-07, 0.0000e+00], |
| [ 4.0978e-08, 0.0000e+00, 1.0338e-07, ..., 4.6566e-09, |
| -1.3132e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 216, bias, value: tensor([-0.0067, 0.0239, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([-6.3516e-07, -5.3756e-06, 2.2668e-06, -4.6287e-07, 3.8091e-07, |
| 1.2405e-06, -2.3022e-06, 3.4161e-06, 1.4408e-06, 6.6124e-08], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 215, time 217.22, cls_loss 0.0013 cls_loss_mapping 0.0011 cls_loss_causal 0.4995 re_mapping 0.0049 re_causal 0.0170 /// teacc 99.06 lr 0.00001000 |
| Epoch 217, weight, value: tensor([[-0.0559, -0.1082, -0.0601, ..., -0.1162, -0.0206, -0.0325], |
| [-0.0502, 0.0494, -0.1617, ..., 0.0628, 0.0715, -0.1372], |
| [ 0.0566, -0.0076, -0.0634, ..., -0.0598, -0.0445, 0.0088], |
| ..., |
| [ 0.0571, -0.0598, -0.0173, ..., 0.0175, -0.0416, 0.0180], |
| [-0.0668, -0.0267, -0.0823, ..., -0.0296, -0.0355, -0.0709], |
| [-0.1436, -0.2042, 0.0281, ..., -0.0659, 0.0403, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 7.7300e-08, 0.0000e+00, 1.4901e-08, ..., 0.0000e+00, |
| 3.2410e-07, 0.0000e+00], |
| [ 5.3085e-08, 0.0000e+00, 2.4214e-08, ..., 0.0000e+00, |
| -1.2927e-06, 0.0000e+00], |
| [-7.5251e-07, 0.0000e+00, 2.7940e-08, ..., 0.0000e+00, |
| 1.0338e-07, 0.0000e+00], |
| ..., |
| [ 4.2375e-07, 0.0000e+00, 3.1665e-08, ..., 0.0000e+00, |
| 4.8615e-07, 0.0000e+00], |
| [ 1.0245e-07, 9.3132e-10, -1.0608e-06, ..., 1.8626e-09, |
| -7.7300e-08, 9.3132e-10], |
| [ 1.3970e-08, 0.0000e+00, 1.1688e-06, ..., 9.3132e-10, |
| 2.5239e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 217, bias, value: tensor([-0.0067, 0.0240, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 8.7358e-07, -2.3674e-06, -8.9686e-07, 1.2740e-06, 3.4831e-07, |
| 1.4622e-07, 2.3004e-07, 1.9260e-06, -7.6666e-06, 6.1281e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 216, time 217.54, cls_loss 0.0010 cls_loss_mapping 0.0011 cls_loss_causal 0.4542 re_mapping 0.0048 re_causal 0.0164 /// teacc 99.00 lr 0.00001000 |
| Epoch 218, weight, value: tensor([[-0.0559, -0.1083, -0.0601, ..., -0.1163, -0.0206, -0.0325], |
| [-0.0502, 0.0494, -0.1617, ..., 0.0628, 0.0715, -0.1372], |
| [ 0.0567, -0.0076, -0.0634, ..., -0.0598, -0.0446, 0.0088], |
| ..., |
| [ 0.0571, -0.0598, -0.0173, ..., 0.0175, -0.0416, 0.0180], |
| [-0.0668, -0.0267, -0.0823, ..., -0.0296, -0.0356, -0.0709], |
| [-0.1436, -0.2043, 0.0281, ..., -0.0658, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.4678e-06, 0.0000e+00, 9.8720e-08, ..., 4.6566e-09, |
| 2.3283e-08, 0.0000e+00], |
| [ 9.6858e-08, 0.0000e+00, 4.3772e-08, ..., 7.4506e-09, |
| -4.3120e-07, 0.0000e+00], |
| [-2.9597e-06, 0.0000e+00, -1.2666e-07, ..., 9.3132e-10, |
| 3.5390e-08, 0.0000e+00], |
| ..., |
| [ 5.0664e-07, 0.0000e+00, 2.9802e-08, ..., 3.7253e-09, |
| 2.6077e-07, 0.0000e+00], |
| [ 9.0711e-07, 0.0000e+00, 2.5984e-07, ..., 4.1910e-08, |
| 5.1223e-08, 0.0000e+00], |
| [ 1.7136e-07, 0.0000e+00, 3.9395e-07, ..., 1.2107e-08, |
| 4.7497e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 218, bias, value: tensor([-0.0067, 0.0239, -0.0026, -0.0051, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([ 9.0003e-06, -4.7963e-07, -1.5974e-05, 7.8753e-06, -3.7253e-07, |
| -8.4788e-06, 2.8033e-07, 1.5674e-06, 5.2229e-06, 1.3541e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 217, time 217.58, cls_loss 0.0011 cls_loss_mapping 0.0009 cls_loss_causal 0.4670 re_mapping 0.0047 re_causal 0.0166 /// teacc 99.02 lr 0.00001000 |
| Epoch 219, weight, value: tensor([[-0.0559, -0.1083, -0.0601, ..., -0.1163, -0.0206, -0.0325], |
| [-0.0503, 0.0494, -0.1617, ..., 0.0628, 0.0715, -0.1372], |
| [ 0.0567, -0.0076, -0.0633, ..., -0.0598, -0.0446, 0.0088], |
| ..., |
| [ 0.0571, -0.0598, -0.0174, ..., 0.0175, -0.0416, 0.0179], |
| [-0.0669, -0.0267, -0.0823, ..., -0.0297, -0.0356, -0.0709], |
| [-0.1437, -0.2043, 0.0280, ..., -0.0659, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.7788e-07, 0.0000e+00, 5.2620e-08, ..., 0.0000e+00, |
| 4.4238e-08, 4.6566e-10], |
| [ 4.7497e-07, 0.0000e+00, 4.1910e-09, ..., -2.3283e-09, |
| -7.1479e-07, 0.0000e+00], |
| [-6.0163e-07, 0.0000e+00, -1.8720e-07, ..., 0.0000e+00, |
| 1.3690e-07, 0.0000e+00], |
| ..., |
| [-1.0598e-06, 0.0000e+00, 2.5611e-08, ..., 4.6566e-10, |
| 6.7661e-07, 0.0000e+00], |
| [ 4.9919e-07, 0.0000e+00, 4.5169e-08, ..., 9.3132e-10, |
| 1.4110e-07, 9.3132e-10], |
| [ 2.2305e-07, 0.0000e+00, -2.7893e-07, ..., 4.6566e-10, |
| -1.4026e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 219, bias, value: tensor([-0.0067, 0.0239, -0.0026, -0.0050, 0.0149, 0.0055, 0.0248, -0.0119, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([-6.8732e-07, -2.9057e-07, -5.4715e-07, 5.2294e-07, 8.1211e-06, |
| 2.3982e-07, 1.4668e-07, 1.5218e-06, 1.2591e-06, -1.0327e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 218, time 217.43, cls_loss 0.0010 cls_loss_mapping 0.0007 cls_loss_causal 0.4783 re_mapping 0.0048 re_causal 0.0166 /// teacc 99.08 lr 0.00001000 |
| Epoch 220, weight, value: tensor([[-0.0560, -0.1083, -0.0601, ..., -0.1164, -0.0206, -0.0325], |
| [-0.0504, 0.0494, -0.1618, ..., 0.0628, 0.0715, -0.1372], |
| [ 0.0568, -0.0076, -0.0633, ..., -0.0598, -0.0446, 0.0088], |
| ..., |
| [ 0.0572, -0.0598, -0.0174, ..., 0.0175, -0.0416, 0.0179], |
| [-0.0669, -0.0267, -0.0823, ..., -0.0298, -0.0356, -0.0710], |
| [-0.1437, -0.2043, 0.0280, ..., -0.0660, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 8.3819e-09, 0.0000e+00, 6.5193e-09, ..., 0.0000e+00, |
| 1.6764e-08, 0.0000e+00], |
| [ 5.2620e-08, 0.0000e+00, 3.1199e-08, ..., 0.0000e+00, |
| -1.4389e-07, 0.0000e+00], |
| [ 3.2131e-08, 0.0000e+00, 3.7253e-09, ..., 0.0000e+00, |
| 2.0955e-08, 0.0000e+00], |
| ..., |
| [-6.4727e-08, 0.0000e+00, 1.7229e-08, ..., 4.6566e-10, |
| 1.2619e-07, 0.0000e+00], |
| [ 2.1886e-08, 0.0000e+00, 3.2131e-08, ..., 4.6566e-10, |
| 4.1444e-08, 0.0000e+00], |
| [ 2.6543e-08, 0.0000e+00, 1.3495e-06, ..., 4.6566e-10, |
| -1.6764e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 220, bias, value: tensor([-0.0067, 0.0239, -0.0025, -0.0050, 0.0149, 0.0055, 0.0249, -0.0119, |
| 0.0060, -0.0056], device='cuda:0'), grad: tensor([-1.6419e-06, -5.6811e-08, 1.2480e-06, 1.0319e-06, -7.6834e-08, |
| -3.3490e-06, -6.5193e-09, 2.6636e-07, 2.9337e-07, 2.3078e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 219, time 217.35, cls_loss 0.0012 cls_loss_mapping 0.0009 cls_loss_causal 0.4607 re_mapping 0.0048 re_causal 0.0166 /// teacc 99.08 lr 0.00001000 |
| Epoch 221, weight, value: tensor([[-0.0560, -0.1083, -0.0601, ..., -0.1164, -0.0206, -0.0325], |
| [-0.0505, 0.0495, -0.1618, ..., 0.0629, 0.0716, -0.1372], |
| [ 0.0568, -0.0076, -0.0633, ..., -0.0598, -0.0447, 0.0088], |
| ..., |
| [ 0.0573, -0.0598, -0.0175, ..., 0.0174, -0.0416, 0.0179], |
| [-0.0670, -0.0267, -0.0823, ..., -0.0299, -0.0357, -0.0710], |
| [-0.1438, -0.2043, 0.0280, ..., -0.0660, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 2.9802e-08, 0.0000e+00, 2.7940e-09, ..., 3.2596e-09, |
| 1.4901e-08, 3.2596e-09], |
| [ 1.0049e-06, 0.0000e+00, 8.6147e-08, ..., 2.0443e-07, |
| -2.0675e-07, 4.6566e-10], |
| [ 9.4529e-08, 0.0000e+00, 7.9162e-09, ..., 7.4506e-09, |
| 1.7695e-08, 9.3132e-10], |
| ..., |
| [-1.5199e-06, 0.0000e+00, 1.4901e-08, ..., -3.4412e-07, |
| 1.7369e-07, 0.0000e+00], |
| [ 1.4063e-07, 0.0000e+00, -3.4459e-08, ..., 1.8626e-09, |
| 2.4121e-07, 2.3283e-09], |
| [ 5.0943e-07, 0.0000e+00, -3.6694e-07, ..., 1.0151e-07, |
| -5.3551e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 221, bias, value: tensor([-0.0067, 0.0238, -0.0026, -0.0050, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0060, -0.0055], device='cuda:0'), grad: tensor([-1.2210e-06, 2.8107e-06, 3.2131e-07, -4.5495e-07, 3.1479e-06, |
| 8.6799e-07, -6.4448e-07, -4.1798e-06, 5.5414e-08, -7.0315e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 220, time 217.49, cls_loss 0.0009 cls_loss_mapping 0.0009 cls_loss_causal 0.4765 re_mapping 0.0048 re_causal 0.0169 /// teacc 99.09 lr 0.00001000 |
| Epoch 222, weight, value: tensor([[-0.0560, -0.1084, -0.0601, ..., -0.1165, -0.0206, -0.0325], |
| [-0.0505, 0.0495, -0.1618, ..., 0.0629, 0.0716, -0.1372], |
| [ 0.0568, -0.0076, -0.0633, ..., -0.0598, -0.0447, 0.0088], |
| ..., |
| [ 0.0573, -0.0598, -0.0175, ..., 0.0174, -0.0417, 0.0179], |
| [-0.0671, -0.0267, -0.0823, ..., -0.0300, -0.0357, -0.0710], |
| [-0.1438, -0.2043, 0.0280, ..., -0.0659, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 7.9162e-09, 0.0000e+00, 3.2596e-09, ..., 0.0000e+00, |
| 9.7789e-09, 0.0000e+00], |
| [ 2.6124e-07, 0.0000e+00, 2.7940e-09, ..., 0.0000e+00, |
| -7.3574e-08, 0.0000e+00], |
| [ 4.2375e-08, 0.0000e+00, 4.6566e-10, ..., 0.0000e+00, |
| 2.3749e-08, 0.0000e+00], |
| ..., |
| [-1.9651e-07, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| 1.0990e-07, 0.0000e+00], |
| [ 2.3283e-08, 0.0000e+00, 6.9384e-08, ..., 0.0000e+00, |
| -1.8021e-07, 0.0000e+00], |
| [ 8.6613e-08, 0.0000e+00, 1.6298e-08, ..., 0.0000e+00, |
| -4.9360e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 222, bias, value: tensor([-0.0066, 0.0238, -0.0025, -0.0050, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0060, -0.0056], device='cuda:0'), grad: tensor([-2.9802e-08, 9.0152e-07, 2.3050e-07, -9.6392e-08, 3.3760e-07, |
| 1.4082e-06, -9.2573e-07, -3.0873e-07, -1.6680e-06, 1.7788e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 221, time 217.43, cls_loss 0.0009 cls_loss_mapping 0.0009 cls_loss_causal 0.4679 re_mapping 0.0048 re_causal 0.0166 /// teacc 99.06 lr 0.00001000 |
| Epoch 223, weight, value: tensor([[-0.0560, -0.1084, -0.0601, ..., -0.1165, -0.0206, -0.0325], |
| [-0.0505, 0.0495, -0.1618, ..., 0.0629, 0.0717, -0.1372], |
| [ 0.0569, -0.0076, -0.0634, ..., -0.0598, -0.0448, 0.0088], |
| ..., |
| [ 0.0572, -0.0599, -0.0175, ..., 0.0174, -0.0417, 0.0179], |
| [-0.0672, -0.0268, -0.0823, ..., -0.0300, -0.0358, -0.0710], |
| [-0.1439, -0.2044, 0.0280, ..., -0.0659, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 7.8697e-08, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| -2.0489e-08, 0.0000e+00], |
| [ 2.1141e-07, 4.6566e-10, 3.9581e-08, ..., 0.0000e+00, |
| -2.4308e-07, 0.0000e+00], |
| [ 1.1809e-06, 4.6566e-10, 4.6566e-09, ..., 0.0000e+00, |
| 5.1223e-08, 0.0000e+00], |
| ..., |
| [ 1.0272e-06, -3.7253e-09, 2.8405e-08, ..., 0.0000e+00, |
| 1.8161e-07, 0.0000e+00], |
| [ 8.6613e-08, 0.0000e+00, 1.3970e-09, ..., 0.0000e+00, |
| 3.0734e-08, 0.0000e+00], |
| [ 2.5798e-07, 0.0000e+00, 1.3504e-07, ..., 0.0000e+00, |
| -1.3178e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 223, bias, value: tensor([-0.0066, 0.0238, -0.0025, -0.0050, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0059, -0.0056], device='cuda:0'), grad: tensor([-2.0564e-06, 1.2107e-08, 2.0638e-06, -3.6675e-06, -4.1304e-07, |
| 6.1328e-07, 4.9965e-07, 1.7984e-06, 2.7707e-07, 8.4564e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 222, time 217.28, cls_loss 0.0011 cls_loss_mapping 0.0008 cls_loss_causal 0.4832 re_mapping 0.0048 re_causal 0.0167 /// teacc 99.09 lr 0.00001000 |
| Epoch 224, weight, value: tensor([[-0.0560, -0.1084, -0.0601, ..., -0.1167, -0.0206, -0.0325], |
| [-0.0505, 0.0496, -0.1619, ..., 0.0629, 0.0717, -0.1372], |
| [ 0.0569, -0.0076, -0.0633, ..., -0.0599, -0.0448, 0.0088], |
| ..., |
| [ 0.0573, -0.0599, -0.0175, ..., 0.0174, -0.0417, 0.0179], |
| [-0.0673, -0.0268, -0.0823, ..., -0.0301, -0.0358, -0.0710], |
| [-0.1439, -0.2044, 0.0279, ..., -0.0658, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 2.8871e-08, 0.0000e+00, 2.3283e-09, ..., 0.0000e+00, |
| 3.8650e-08, 4.6566e-10], |
| [ 5.2154e-08, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| -1.2573e-07, 0.0000e+00], |
| [-4.5029e-07, 0.0000e+00, 3.2596e-09, ..., 0.0000e+00, |
| 2.1886e-08, 0.0000e+00], |
| ..., |
| [-5.6811e-08, 0.0000e+00, 1.7229e-08, ..., 0.0000e+00, |
| 9.7323e-08, 0.0000e+00], |
| [ 3.8091e-07, 0.0000e+00, -4.6566e-09, ..., 0.0000e+00, |
| 2.7474e-08, 5.5879e-09], |
| [ 5.4948e-08, 0.0000e+00, -7.4506e-09, ..., 0.0000e+00, |
| -7.8697e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 224, bias, value: tensor([-0.0066, 0.0238, -0.0025, -0.0051, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0059, -0.0056], device='cuda:0'), grad: tensor([-3.6601e-07, 4.7963e-08, -3.9814e-07, 2.4494e-07, 2.1666e-05, |
| 2.1085e-06, -2.4587e-05, 4.5914e-07, 1.1511e-06, -3.2643e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 223, time 217.27, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4846 re_mapping 0.0047 re_causal 0.0167 /// teacc 99.09 lr 0.00001000 |
| Epoch 225, weight, value: tensor([[-0.0560, -0.1085, -0.0601, ..., -0.1168, -0.0206, -0.0325], |
| [-0.0505, 0.0497, -0.1619, ..., 0.0629, 0.0718, -0.1372], |
| [ 0.0569, -0.0076, -0.0633, ..., -0.0599, -0.0448, 0.0088], |
| ..., |
| [ 0.0573, -0.0600, -0.0175, ..., 0.0174, -0.0418, 0.0179], |
| [-0.0674, -0.0268, -0.0823, ..., -0.0302, -0.0359, -0.0710], |
| [-0.1440, -0.2044, 0.0279, ..., -0.0659, 0.0404, 0.0480]], |
| device='cuda:0'), grad: tensor([[-1.1967e-07, 0.0000e+00, 1.3970e-08, ..., 1.8626e-09, |
| 2.7986e-07, 0.0000e+00], |
| [ 1.5181e-07, 0.0000e+00, 3.9814e-07, ..., 1.0710e-08, |
| -8.0392e-06, 0.0000e+00], |
| [-1.5661e-05, 0.0000e+00, 3.9581e-08, ..., 3.7253e-09, |
| 1.5087e-07, 0.0000e+00], |
| ..., |
| [ 1.5557e-05, 0.0000e+00, 1.5693e-07, ..., 3.4459e-08, |
| 1.1967e-06, 0.0000e+00], |
| [ 2.4354e-07, 0.0000e+00, 3.7719e-08, ..., 2.3283e-09, |
| 4.0270e-06, 0.0000e+00], |
| [ 2.9383e-07, 0.0000e+00, 5.9605e-07, ..., 9.7789e-09, |
| 2.2724e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 225, bias, value: tensor([-0.0066, 0.0239, -0.0025, -0.0051, 0.0149, 0.0055, 0.0249, -0.0118, |
| 0.0059, -0.0056], device='cuda:0'), grad: tensor([-8.4564e-06, -3.4094e-05, -1.8910e-05, -6.5891e-07, 7.0930e-06, |
| 8.5216e-07, 9.5554e-07, 2.4691e-05, 1.8150e-05, 1.0386e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 224, time 217.37, cls_loss 0.0009 cls_loss_mapping 0.0008 cls_loss_causal 0.4598 re_mapping 0.0046 re_causal 0.0162 /// teacc 99.07 lr 0.00001000 |
| Epoch 226, weight, value: tensor([[-0.0560, -0.1087, -0.0601, ..., -0.1169, -0.0206, -0.0325], |
| [-0.0505, 0.0498, -0.1619, ..., 0.0629, 0.0718, -0.1372], |
| [ 0.0570, -0.0076, -0.0633, ..., -0.0599, -0.0448, 0.0088], |
| ..., |
| [ 0.0573, -0.0600, -0.0176, ..., 0.0174, -0.0418, 0.0179], |
| [-0.0675, -0.0269, -0.0824, ..., -0.0303, -0.0360, -0.0710], |
| [-0.1440, -0.2045, 0.0278, ..., -0.0659, 0.0405, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.3970e-07, 7.9162e-09, 8.1025e-08, ..., 0.0000e+00, |
| -1.1036e-07, 0.0000e+00], |
| [ 1.7118e-06, 3.4319e-07, 3.0268e-08, ..., 0.0000e+00, |
| -2.6543e-08, 0.0000e+00], |
| [-3.2084e-07, 9.8441e-07, 8.7079e-08, ..., 0.0000e+00, |
| 1.5832e-08, 0.0000e+00], |
| ..., |
| [ 1.5059e-06, 2.4540e-07, 2.5099e-07, ..., 1.8626e-09, |
| 1.0058e-07, 0.0000e+00], |
| [-9.4846e-06, -4.2617e-06, -1.8366e-06, ..., 0.0000e+00, |
| -3.2596e-09, 4.6566e-10], |
| [ 1.4128e-06, 2.2352e-08, 9.1875e-07, ..., 0.0000e+00, |
| -1.8021e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 226, bias, value: tensor([-0.0066, 0.0239, -0.0025, -0.0051, 0.0148, 0.0055, 0.0249, -0.0118, |
| 0.0058, -0.0056], device='cuda:0'), grad: tensor([-1.4342e-06, 5.1707e-06, 7.7859e-06, 5.4725e-06, 1.0990e-06, |
| 9.1642e-06, 1.3642e-05, 4.9323e-06, -5.1469e-05, 5.7034e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 225, time 217.26, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4718 re_mapping 0.0046 re_causal 0.0162 /// teacc 99.06 lr 0.00001000 |
| Epoch 227, weight, value: tensor([[-0.0560, -0.1087, -0.0601, ..., -0.1169, -0.0207, -0.0325], |
| [-0.0505, 0.0499, -0.1620, ..., 0.0629, 0.0718, -0.1373], |
| [ 0.0570, -0.0077, -0.0633, ..., -0.0599, -0.0448, 0.0088], |
| ..., |
| [ 0.0573, -0.0600, -0.0176, ..., 0.0174, -0.0418, 0.0178], |
| [-0.0676, -0.0269, -0.0824, ..., -0.0303, -0.0360, -0.0710], |
| [-0.1441, -0.2045, 0.0278, ..., -0.0659, 0.0405, 0.0480]], |
| device='cuda:0'), grad: tensor([[ 1.0887e-06, 0.0000e+00, -4.2282e-07, ..., 4.6566e-10, |
| 1.7229e-08, 0.0000e+00], |
| [ 8.6501e-06, 4.6566e-10, 5.1223e-08, ..., 8.3819e-09, |
| -1.8300e-07, 0.0000e+00], |
| [ 1.9930e-07, -3.0268e-08, 8.7079e-08, ..., 4.6566e-10, |
| 2.9337e-08, 0.0000e+00], |
| ..., |
| [-1.4834e-05, 2.7940e-08, 4.0978e-08, ..., 1.8626e-09, |
| 1.8068e-07, 0.0000e+00], |
| [ 8.0559e-08, 4.6566e-10, -3.1665e-08, ..., 0.0000e+00, |
| 4.4238e-08, 0.0000e+00], |
| [ 7.6182e-07, 0.0000e+00, 2.2445e-07, ..., 5.1223e-09, |
| -2.0023e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 227, bias, value: tensor([-0.0066, 0.0239, -0.0024, -0.0050, 0.0148, 0.0055, 0.0250, -0.0118, |
| 0.0058, -0.0056], device='cuda:0'), grad: tensor([ 5.4482e-08, 1.9237e-05, 1.0636e-06, 3.3602e-06, 4.5709e-06, |
| 2.9020e-06, -2.7474e-08, -3.2455e-05, -4.2375e-08, 1.3420e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 226, time 217.33, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4768 re_mapping 0.0046 re_causal 0.0162 /// teacc 99.06 lr 0.00001000 |
| Epoch 228, weight, value: tensor([[-0.0560, -0.1089, -0.0601, ..., -0.1170, -0.0207, -0.0325], |
| [-0.0506, 0.0501, -0.1620, ..., 0.0629, 0.0718, -0.1373], |
| [ 0.0571, -0.0077, -0.0633, ..., -0.0599, -0.0449, 0.0088], |
| ..., |
| [ 0.0573, -0.0602, -0.0177, ..., 0.0174, -0.0418, 0.0178], |
| [-0.0677, -0.0269, -0.0824, ..., -0.0304, -0.0360, -0.0710], |
| [-0.1441, -0.2046, 0.0278, ..., -0.0660, 0.0406, 0.0481]], |
| device='cuda:0'), grad: tensor([[-1.5069e-06, 0.0000e+00, 1.3970e-09, ..., 9.3132e-10, |
| 4.1444e-08, 0.0000e+00], |
| [ 2.5202e-06, 0.0000e+00, 1.4435e-08, ..., 1.2424e-06, |
| -2.6915e-07, 0.0000e+00], |
| [-7.4320e-07, 0.0000e+00, -6.9849e-09, ..., 2.3283e-09, |
| 2.0489e-08, 0.0000e+00], |
| ..., |
| [-3.3397e-06, 0.0000e+00, -1.0617e-07, ..., -1.3513e-06, |
| 2.9290e-07, 0.0000e+00], |
| [ 4.2468e-07, 0.0000e+00, 1.5832e-08, ..., 0.0000e+00, |
| 4.4703e-08, 1.3970e-09], |
| [ 1.7658e-06, 0.0000e+00, 6.2399e-08, ..., 8.1956e-08, |
| -2.5518e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 228, bias, value: tensor([-0.0065, 0.0238, -0.0024, -0.0051, 0.0148, 0.0055, 0.0249, -0.0118, |
| 0.0058, -0.0056], device='cuda:0'), grad: tensor([-7.0482e-06, 5.2638e-06, -5.5600e-07, 6.2585e-07, 1.0468e-06, |
| 4.6752e-07, 1.5907e-06, -7.0408e-06, 1.1679e-06, 4.4964e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 227, time 217.31, cls_loss 0.0011 cls_loss_mapping 0.0008 cls_loss_causal 0.4716 re_mapping 0.0045 re_causal 0.0158 /// teacc 99.06 lr 0.00001000 |
| Epoch 229, weight, value: tensor([[-0.0560, -0.1089, -0.0602, ..., -0.1171, -0.0207, -0.0325], |
| [-0.0506, 0.0501, -0.1620, ..., 0.0629, 0.0719, -0.1373], |
| [ 0.0571, -0.0077, -0.0632, ..., -0.0600, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0178, ..., 0.0174, -0.0418, 0.0178], |
| [-0.0678, -0.0269, -0.0824, ..., -0.0305, -0.0360, -0.0710], |
| [-0.1442, -0.2046, 0.0277, ..., -0.0660, 0.0406, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 4.0978e-08, 0.0000e+00, -9.5740e-06, ..., 4.6566e-10, |
| 5.0291e-08, 0.0000e+00], |
| [ 4.0382e-06, 0.0000e+00, 7.0781e-08, ..., 0.0000e+00, |
| -1.7462e-07, 0.0000e+00], |
| [ 1.1083e-07, 0.0000e+00, 1.4435e-08, ..., 0.0000e+00, |
| 2.7008e-08, 0.0000e+00], |
| ..., |
| [-1.2569e-05, 0.0000e+00, 1.5600e-07, ..., 4.6566e-10, |
| 1.7276e-07, 0.0000e+00], |
| [ 1.5134e-07, 0.0000e+00, -1.1735e-07, ..., 1.3970e-09, |
| -7.8231e-08, 0.0000e+00], |
| [ 5.5693e-06, 0.0000e+00, 1.9604e-07, ..., 1.3970e-09, |
| -6.3051e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 229, bias, value: tensor([-0.0065, 0.0238, -0.0024, -0.0051, 0.0148, 0.0055, 0.0249, -0.0118, |
| 0.0058, -0.0056], device='cuda:0'), grad: tensor([-7.5281e-05, 1.2934e-05, 9.1121e-06, 9.0823e-06, 3.4235e-06, |
| 1.7181e-05, 4.0114e-05, -4.0084e-05, 1.9614e-06, 2.1502e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 228, time 217.25, cls_loss 0.0011 cls_loss_mapping 0.0010 cls_loss_causal 0.4693 re_mapping 0.0045 re_causal 0.0162 /// teacc 99.07 lr 0.00001000 |
| Epoch 230, weight, value: tensor([[-0.0560, -0.1090, -0.0601, ..., -0.1172, -0.0207, -0.0325], |
| [-0.0507, 0.0501, -0.1621, ..., 0.0629, 0.0719, -0.1373], |
| [ 0.0573, -0.0076, -0.0632, ..., -0.0600, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0178, ..., 0.0174, -0.0418, 0.0178], |
| [-0.0680, -0.0269, -0.0824, ..., -0.0306, -0.0361, -0.0710], |
| [-0.1442, -0.2046, 0.0277, ..., -0.0660, 0.0406, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 3.3528e-08, 0.0000e+00, 4.6566e-09, ..., 0.0000e+00, |
| 3.9581e-08, 0.0000e+00], |
| [ 7.5437e-08, 8.3819e-09, 4.2375e-08, ..., 4.6566e-10, |
| -4.6706e-07, 0.0000e+00], |
| [-6.9477e-07, -1.3970e-08, 8.3819e-09, ..., 0.0000e+00, |
| 3.7253e-08, 0.0000e+00], |
| ..., |
| [ 4.6426e-07, 5.1223e-09, 3.2596e-08, ..., 0.0000e+00, |
| 2.7241e-07, 0.0000e+00], |
| [ 7.3574e-08, 0.0000e+00, 4.1910e-09, ..., 0.0000e+00, |
| 8.9873e-08, 0.0000e+00], |
| [ 9.7789e-09, 4.6566e-10, 1.2629e-06, ..., 4.6566e-10, |
| -1.9791e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 230, bias, value: tensor([-0.0065, 0.0238, -0.0023, -0.0051, 0.0148, 0.0055, 0.0249, -0.0118, |
| 0.0057, -0.0056], device='cuda:0'), grad: tensor([-2.1653e-07, -7.5204e-07, -7.1526e-07, 1.3178e-07, -2.4363e-06, |
| 1.5181e-07, -3.0687e-07, 1.3579e-06, 2.9337e-07, 2.4978e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 229, time 217.25, cls_loss 0.0008 cls_loss_mapping 0.0007 cls_loss_causal 0.4313 re_mapping 0.0045 re_causal 0.0159 /// teacc 99.05 lr 0.00001000 |
| Epoch 231, weight, value: tensor([[-0.0560, -0.1090, -0.0601, ..., -0.1173, -0.0207, -0.0325], |
| [-0.0507, 0.0501, -0.1621, ..., 0.0629, 0.0719, -0.1374], |
| [ 0.0574, -0.0076, -0.0631, ..., -0.0600, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0180, ..., 0.0174, -0.0419, 0.0177], |
| [-0.0681, -0.0269, -0.0825, ..., -0.0306, -0.0362, -0.0710], |
| [-0.1443, -0.2047, 0.0276, ..., -0.0660, 0.0407, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 7.7020e-07, 0.0000e+00, 7.1852e-07, ..., 9.3132e-10, |
| 3.9022e-07, 0.0000e+00], |
| [ 2.0284e-06, 0.0000e+00, 2.3330e-07, ..., 9.3132e-10, |
| -5.2787e-06, 0.0000e+00], |
| [-4.7907e-06, 0.0000e+00, 2.8610e-06, ..., 0.0000e+00, |
| 5.4017e-07, 0.0000e+00], |
| ..., |
| [ 1.9036e-06, 0.0000e+00, 1.4855e-07, ..., 1.4435e-08, |
| -5.1735e-07, 0.0000e+00], |
| [-2.0768e-06, 0.0000e+00, -5.4352e-06, ..., 2.3283e-09, |
| 1.2089e-06, 0.0000e+00], |
| [ 2.7716e-06, 0.0000e+00, 7.6648e-07, ..., -1.5832e-08, |
| 8.6613e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 231, bias, value: tensor([-0.0065, 0.0237, -0.0022, -0.0051, 0.0148, 0.0055, 0.0248, -0.0118, |
| 0.0056, -0.0057], device='cuda:0'), grad: tensor([ 6.5342e-06, -1.7121e-05, 5.6028e-06, -6.1607e-07, 1.5572e-05, |
| 1.8459e-06, -5.3868e-06, 2.3749e-07, -1.7524e-05, 1.0848e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 230, time 217.45, cls_loss 0.0009 cls_loss_mapping 0.0008 cls_loss_causal 0.4844 re_mapping 0.0045 re_causal 0.0164 /// teacc 99.04 lr 0.00001000 |
| Epoch 232, weight, value: tensor([[-0.0560, -0.1090, -0.0601, ..., -0.1173, -0.0207, -0.0325], |
| [-0.0507, 0.0501, -0.1622, ..., 0.0629, 0.0720, -0.1374], |
| [ 0.0574, -0.0076, -0.0632, ..., -0.0600, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0181, ..., 0.0174, -0.0419, 0.0176], |
| [-0.0681, -0.0270, -0.0825, ..., -0.0307, -0.0363, -0.0710], |
| [-0.1443, -0.2047, 0.0276, ..., -0.0660, 0.0407, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 1.2573e-08, 1.8626e-09, 1.2107e-08, ..., 0.0000e+00, |
| 3.9581e-08, 0.0000e+00], |
| [-6.3656e-07, -1.3709e-06, 2.4214e-08, ..., 0.0000e+00, |
| -2.9728e-06, 0.0000e+00], |
| [ 1.9465e-07, 3.1060e-07, 9.3132e-10, ..., 0.0000e+00, |
| 7.3761e-07, -4.6566e-10], |
| ..., |
| [ 3.9767e-07, 9.4064e-07, 6.9849e-09, ..., 0.0000e+00, |
| 1.4007e-06, 4.6566e-10], |
| [-6.9384e-08, 1.3970e-09, 2.7940e-08, ..., 0.0000e+00, |
| 1.1036e-07, 0.0000e+00], |
| [ 1.3504e-07, 9.3132e-10, 1.5311e-06, ..., 0.0000e+00, |
| -6.1467e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 232, bias, value: tensor([-0.0064, 0.0238, -0.0022, -0.0050, 0.0148, 0.0055, 0.0248, -0.0118, |
| 0.0056, -0.0057], device='cuda:0'), grad: tensor([-2.6077e-07, -1.1854e-05, 3.8091e-06, 5.7835e-07, 1.8813e-06, |
| -2.9672e-06, 9.7416e-07, 6.3889e-06, -3.2075e-06, 4.6380e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 231, time 217.33, cls_loss 0.0009 cls_loss_mapping 0.0008 cls_loss_causal 0.4745 re_mapping 0.0045 re_causal 0.0162 /// teacc 99.05 lr 0.00001000 |
| Epoch 233, weight, value: tensor([[-0.0560, -0.1091, -0.0600, ..., -0.1174, -0.0207, -0.0325], |
| [-0.0508, 0.0502, -0.1622, ..., 0.0629, 0.0720, -0.1374], |
| [ 0.0575, -0.0076, -0.0631, ..., -0.0601, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0182, ..., 0.0174, -0.0419, 0.0176], |
| [-0.0682, -0.0271, -0.0825, ..., -0.0308, -0.0364, -0.0710], |
| [-0.1444, -0.2047, 0.0276, ..., -0.0660, 0.0407, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 1.4901e-08, 9.3132e-10, 3.2596e-09, ..., 0.0000e+00, |
| 2.4680e-08, 0.0000e+00], |
| [ 1.0934e-06, -2.8266e-07, 1.5832e-08, ..., 4.6566e-10, |
| -2.2259e-07, 0.0000e+00], |
| [-8.6986e-07, 1.7509e-07, 3.7253e-09, ..., 0.0000e+00, |
| 1.9185e-07, 0.0000e+00], |
| ..., |
| [-2.7614e-07, 9.8255e-08, 1.6764e-08, ..., 4.6566e-10, |
| 1.9511e-07, 0.0000e+00], |
| [ 1.4482e-07, 4.6566e-10, 1.3039e-08, ..., 0.0000e+00, |
| 1.3318e-07, 0.0000e+00], |
| [ 1.7695e-07, 0.0000e+00, -6.0536e-09, ..., 4.6566e-10, |
| -2.3143e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 233, bias, value: tensor([-0.0064, 0.0237, -0.0022, -0.0051, 0.0148, 0.0055, 0.0248, -0.0118, |
| 0.0056, -0.0057], device='cuda:0'), grad: tensor([-9.6206e-07, 1.1604e-06, -5.0291e-07, -6.1467e-07, 7.1200e-07, |
| 5.9977e-07, -1.1763e-06, -5.8673e-08, 6.4075e-07, 2.2445e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 232, time 217.63, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4912 re_mapping 0.0045 re_causal 0.0163 /// teacc 99.05 lr 0.00001000 |
| Epoch 234, weight, value: tensor([[-0.0560, -0.1091, -0.0600, ..., -0.1175, -0.0208, -0.0325], |
| [-0.0508, 0.0502, -0.1623, ..., 0.0629, 0.0720, -0.1374], |
| [ 0.0575, -0.0076, -0.0631, ..., -0.0601, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0182, ..., 0.0174, -0.0419, 0.0176], |
| [-0.0683, -0.0271, -0.0826, ..., -0.0308, -0.0365, -0.0710], |
| [-0.1445, -0.2047, 0.0275, ..., -0.0660, 0.0408, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 4.4703e-08, 7.4506e-09, 3.9116e-08, ..., 0.0000e+00, |
| 4.6566e-08, 1.1176e-08], |
| [ 1.8841e-06, 3.7998e-07, 2.8126e-07, ..., 0.0000e+00, |
| 2.3842e-07, 1.8626e-09], |
| [ 7.9721e-07, 1.8626e-07, 4.1910e-08, ..., 0.0000e+00, |
| 1.0245e-08, 9.3132e-10], |
| ..., |
| [-3.7961e-06, -8.1025e-07, 1.2480e-06, ..., 0.0000e+00, |
| 1.0896e-07, 0.0000e+00], |
| [ 4.4703e-08, 4.6566e-09, 1.4901e-08, ..., 0.0000e+00, |
| -5.3272e-07, 3.4459e-08], |
| [ 5.1316e-07, 9.3132e-08, 6.3293e-06, ..., 9.3132e-10, |
| 1.9558e-08, 9.3132e-10]], device='cuda:0') |
| Epoch 234, bias, value: tensor([-0.0064, 0.0237, -0.0021, -0.0051, 0.0148, 0.0056, 0.0248, -0.0117, |
| 0.0054, -0.0057], device='cuda:0'), grad: tensor([-1.9632e-06, 7.5102e-06, 2.1011e-06, 1.5208e-06, -1.8805e-05, |
| 1.8440e-06, -2.1793e-07, -5.1335e-06, -3.6955e-06, 1.6794e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 233, time 217.48, cls_loss 0.0009 cls_loss_mapping 0.0006 cls_loss_causal 0.4746 re_mapping 0.0044 re_causal 0.0160 /// teacc 99.04 lr 0.00001000 |
| Epoch 235, weight, value: tensor([[-0.0561, -0.1092, -0.0600, ..., -0.1175, -0.0208, -0.0325], |
| [-0.0509, 0.0502, -0.1624, ..., 0.0629, 0.0721, -0.1374], |
| [ 0.0577, -0.0076, -0.0630, ..., -0.0601, -0.0449, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0183, ..., 0.0174, -0.0419, 0.0176], |
| [-0.0684, -0.0271, -0.0826, ..., -0.0308, -0.0366, -0.0710], |
| [-0.1445, -0.2047, 0.0274, ..., -0.0661, 0.0408, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 4.4703e-08, 2.7940e-09, 6.5193e-09, ..., 0.0000e+00, |
| 3.9116e-08, 0.0000e+00], |
| [ 6.7521e-07, 1.0058e-07, 1.8626e-08, ..., 0.0000e+00, |
| -3.9116e-08, 0.0000e+00], |
| [ 8.5589e-07, 2.2538e-07, -8.3819e-09, ..., 0.0000e+00, |
| 1.3970e-08, 0.0000e+00], |
| ..., |
| [-2.7753e-07, 6.8918e-08, 2.5146e-08, ..., 0.0000e+00, |
| 7.7300e-08, 4.6566e-09], |
| [-4.3176e-06, -1.2023e-06, 1.2387e-07, ..., 0.0000e+00, |
| 7.9162e-08, 0.0000e+00], |
| [ 6.6776e-07, 7.4506e-09, 8.1025e-08, ..., 0.0000e+00, |
| -1.7136e-07, -5.5879e-09]], device='cuda:0') |
| Epoch 235, bias, value: tensor([-0.0064, 0.0237, -0.0021, -0.0051, 0.0148, 0.0056, 0.0248, -0.0117, |
| 0.0053, -0.0058], device='cuda:0'), grad: tensor([ 1.6764e-07, 3.3490e-06, 5.7146e-06, 7.9721e-07, 8.7172e-07, |
| 8.1435e-06, 8.4713e-06, -1.3318e-07, -2.8715e-05, 1.3076e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 234, time 217.43, cls_loss 0.0009 cls_loss_mapping 0.0007 cls_loss_causal 0.4849 re_mapping 0.0044 re_causal 0.0165 /// teacc 99.03 lr 0.00001000 |
| Epoch 236, weight, value: tensor([[-0.0560, -0.1092, -0.0600, ..., -0.1176, -0.0208, -0.0325], |
| [-0.0509, 0.0502, -0.1625, ..., 0.0630, 0.0721, -0.1374], |
| [ 0.0578, -0.0076, -0.0629, ..., -0.0601, -0.0450, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0184, ..., 0.0173, -0.0420, 0.0176], |
| [-0.0686, -0.0271, -0.0827, ..., -0.0309, -0.0368, -0.0710], |
| [-0.1446, -0.2048, 0.0273, ..., -0.0661, 0.0409, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 2.2352e-08, 9.3132e-10, -2.7940e-09, ..., 0.0000e+00, |
| 1.4901e-08, 0.0000e+00], |
| [-9.7696e-07, -1.7183e-06, 2.7008e-08, ..., 0.0000e+00, |
| -2.5313e-06, 0.0000e+00], |
| [ 6.7800e-07, 9.8720e-07, 6.5193e-09, ..., 0.0000e+00, |
| 1.3402e-06, 0.0000e+00], |
| ..., |
| [-1.2135e-06, 6.8825e-07, -7.0781e-08, ..., 0.0000e+00, |
| 1.0477e-06, 0.0000e+00], |
| [ 4.3493e-07, 9.3132e-10, 9.6858e-08, ..., 0.0000e+00, |
| 3.2596e-08, 0.0000e+00], |
| [ 1.0971e-06, 0.0000e+00, 1.8813e-07, ..., 9.3132e-10, |
| -5.9605e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 236, bias, value: tensor([-0.0063, 0.0237, -0.0020, -0.0051, 0.0148, 0.0056, 0.0248, -0.0118, |
| 0.0052, -0.0058], device='cuda:0'), grad: tensor([-2.3283e-07, -4.9025e-06, 2.6934e-06, -2.4587e-07, 5.1595e-07, |
| 5.3085e-08, 1.3411e-07, -2.5406e-06, 1.4957e-06, 3.0212e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 235, time 217.60, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4786 re_mapping 0.0045 re_causal 0.0162 /// teacc 99.04 lr 0.00001000 |
| Epoch 237, weight, value: tensor([[-0.0561, -0.1093, -0.0601, ..., -0.1177, -0.0209, -0.0325], |
| [-0.0509, 0.0502, -0.1626, ..., 0.0629, 0.0722, -0.1374], |
| [ 0.0578, -0.0077, -0.0629, ..., -0.0601, -0.0451, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0185, ..., 0.0173, -0.0420, 0.0175], |
| [-0.0687, -0.0271, -0.0827, ..., -0.0310, -0.0369, -0.0711], |
| [-0.1446, -0.2048, 0.0274, ..., -0.0661, 0.0410, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 2.9802e-08, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| 1.5832e-08, 0.0000e+00], |
| [ 1.8720e-07, 0.0000e+00, 1.3970e-08, ..., 4.6566e-09, |
| -2.3376e-07, 0.0000e+00], |
| [-2.7474e-07, 0.0000e+00, 1.8626e-09, ..., 0.0000e+00, |
| 5.1223e-08, 0.0000e+00], |
| ..., |
| [-4.1723e-07, 0.0000e+00, 3.7253e-09, ..., -2.3283e-08, |
| 2.1700e-07, 0.0000e+00], |
| [ 6.7055e-08, 0.0000e+00, 5.5879e-09, ..., 0.0000e+00, |
| 4.0047e-08, 0.0000e+00], |
| [ 7.6368e-08, 0.0000e+00, -1.1176e-08, ..., 3.7253e-09, |
| -1.0272e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 237, bias, value: tensor([-0.0063, 0.0237, -0.0020, -0.0051, 0.0147, 0.0057, 0.0249, -0.0118, |
| 0.0052, -0.0057], device='cuda:0'), grad: tensor([-4.1425e-06, 2.0675e-07, -2.4866e-07, 1.0775e-06, 3.2783e-06, |
| 3.0287e-06, 7.1060e-07, -1.2517e-06, 3.1572e-07, -2.9635e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 236, time 217.65, cls_loss 0.0009 cls_loss_mapping 0.0007 cls_loss_causal 0.4453 re_mapping 0.0043 re_causal 0.0157 /// teacc 99.08 lr 0.00001000 |
| Epoch 238, weight, value: tensor([[-0.0561, -0.1093, -0.0600, ..., -0.1178, -0.0210, -0.0325], |
| [-0.0509, 0.0502, -0.1626, ..., 0.0629, 0.0722, -0.1375], |
| [ 0.0579, -0.0077, -0.0629, ..., -0.0602, -0.0451, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0185, ..., 0.0174, -0.0420, 0.0175], |
| [-0.0688, -0.0271, -0.0828, ..., -0.0310, -0.0370, -0.0711], |
| [-0.1447, -0.2048, 0.0273, ..., -0.0661, 0.0411, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 1.4901e-08, 0.0000e+00, 1.6671e-07, ..., 9.3132e-10, |
| 1.0245e-08, 0.0000e+00], |
| [ 1.5553e-07, 0.0000e+00, 9.8720e-08, ..., 1.8626e-09, |
| 3.6322e-08, 0.0000e+00], |
| [-3.7253e-09, 0.0000e+00, 1.5739e-07, ..., 0.0000e+00, |
| 1.3970e-08, 0.0000e+00], |
| ..., |
| [-9.9652e-08, 0.0000e+00, 4.9081e-07, ..., 1.6671e-07, |
| 6.8732e-07, 0.0000e+00], |
| [ 3.7253e-08, 0.0000e+00, 8.0932e-07, ..., 1.8626e-09, |
| -3.3900e-07, 0.0000e+00], |
| [-6.0536e-08, 0.0000e+00, -8.2795e-07, ..., -2.1048e-07, |
| -5.6252e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 238, bias, value: tensor([-0.0063, 0.0237, -0.0020, -0.0051, 0.0147, 0.0057, 0.0249, -0.0118, |
| 0.0051, -0.0057], device='cuda:0'), grad: tensor([-6.0536e-06, 1.3392e-06, 4.1537e-07, 3.9041e-05, 8.0913e-06, |
| -4.2588e-05, 1.8114e-06, 5.6326e-06, 1.7956e-06, -9.5442e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 237, time 217.36, cls_loss 0.0008 cls_loss_mapping 0.0008 cls_loss_causal 0.4947 re_mapping 0.0045 re_causal 0.0168 /// teacc 99.07 lr 0.00001000 |
| Epoch 239, weight, value: tensor([[-0.0562, -0.1093, -0.0601, ..., -0.1178, -0.0210, -0.0325], |
| [-0.0509, 0.0503, -0.1627, ..., 0.0629, 0.0723, -0.1375], |
| [ 0.0579, -0.0077, -0.0629, ..., -0.0602, -0.0452, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0186, ..., 0.0174, -0.0421, 0.0175], |
| [-0.0689, -0.0271, -0.0828, ..., -0.0311, -0.0370, -0.0711], |
| [-0.1447, -0.2048, 0.0272, ..., -0.0661, 0.0411, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 7.6368e-08, 0.0000e+00, 8.6613e-08, ..., 0.0000e+00, |
| 3.5390e-08, 0.0000e+00], |
| [ 8.9407e-08, 0.0000e+00, 2.9802e-08, ..., 0.0000e+00, |
| -6.5193e-08, 0.0000e+00], |
| [-2.7940e-08, 0.0000e+00, 1.2107e-08, ..., 0.0000e+00, |
| 7.4506e-09, 0.0000e+00], |
| ..., |
| [-1.8626e-08, 0.0000e+00, 2.5146e-08, ..., 0.0000e+00, |
| 4.1910e-08, 0.0000e+00], |
| [ 1.5832e-08, 0.0000e+00, 3.7253e-09, ..., 0.0000e+00, |
| 1.2107e-08, 0.0000e+00], |
| [ 6.0536e-08, 0.0000e+00, -9.1176e-07, ..., 0.0000e+00, |
| -3.8650e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 239, bias, value: tensor([-0.0063, 0.0237, -0.0020, -0.0051, 0.0147, 0.0057, 0.0249, -0.0118, |
| 0.0051, -0.0057], device='cuda:0'), grad: tensor([ 2.4214e-06, 1.5926e-07, 8.4750e-07, -1.7695e-08, 4.8876e-06, |
| 6.4634e-07, -6.2585e-06, 1.3970e-08, 3.7253e-08, -2.7139e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 238, time 217.38, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4748 re_mapping 0.0045 re_causal 0.0163 /// teacc 99.08 lr 0.00001000 |
| Epoch 240, weight, value: tensor([[-0.0561, -0.1093, -0.0601, ..., -0.1179, -0.0210, -0.0325], |
| [-0.0509, 0.0503, -0.1627, ..., 0.0632, 0.0725, -0.1375], |
| [ 0.0580, -0.0077, -0.0629, ..., -0.0602, -0.0453, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0187, ..., 0.0171, -0.0423, 0.0175], |
| [-0.0690, -0.0271, -0.0828, ..., -0.0312, -0.0371, -0.0711], |
| [-0.1448, -0.2048, 0.0271, ..., -0.0661, 0.0412, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 9.2108e-07, 0.0000e+00, 1.4901e-08, ..., 2.0489e-08, |
| 6.0536e-08, 0.0000e+00], |
| [ 1.9372e-07, 0.0000e+00, 1.3690e-07, ..., 1.7229e-07, |
| -9.3132e-10, 0.0000e+00], |
| [-2.3954e-06, 0.0000e+00, 5.9605e-08, ..., 9.6858e-08, |
| 8.0094e-08, 0.0000e+00], |
| ..., |
| [-3.3453e-06, 0.0000e+00, -6.5006e-06, ..., -1.0706e-05, |
| 4.2841e-08, 0.0000e+00], |
| [ 4.1723e-07, 0.0000e+00, 2.4214e-08, ..., 5.5879e-09, |
| 6.9849e-08, 0.0000e+00], |
| [ 2.5146e-07, 0.0000e+00, 6.9197e-07, ..., 2.8126e-07, |
| -6.7987e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 240, bias, value: tensor([-0.0063, 0.0238, -0.0020, -0.0051, 0.0147, 0.0057, 0.0249, -0.0119, |
| 0.0050, -0.0058], device='cuda:0'), grad: tensor([ 2.0154e-06, 8.3912e-07, -5.2452e-06, 2.4885e-06, 1.5751e-05, |
| 4.3623e-06, -1.8626e-07, -2.3693e-05, 1.2135e-06, 2.4140e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 239, time 217.94, cls_loss 0.0009 cls_loss_mapping 0.0009 cls_loss_causal 0.4928 re_mapping 0.0045 re_causal 0.0165 /// teacc 99.09 lr 0.00001000 |
| Epoch 241, weight, value: tensor([[-0.0561, -0.1094, -0.0601, ..., -0.1179, -0.0210, -0.0325], |
| [-0.0510, 0.0503, -0.1628, ..., 0.0634, 0.0727, -0.1375], |
| [ 0.0581, -0.0077, -0.0628, ..., -0.0603, -0.0453, 0.0088], |
| ..., |
| [ 0.0574, -0.0602, -0.0187, ..., 0.0170, -0.0425, 0.0174], |
| [-0.0691, -0.0271, -0.0828, ..., -0.0312, -0.0372, -0.0711], |
| [-0.1448, -0.2048, 0.0270, ..., -0.0661, 0.0412, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 1.3318e-07, 0.0000e+00, 4.7870e-07, ..., 0.0000e+00, |
| 8.3167e-07, 0.0000e+00], |
| [ 3.0641e-07, 0.0000e+00, 3.7439e-07, ..., 0.0000e+00, |
| 4.0978e-08, 0.0000e+00], |
| [-3.0976e-06, 0.0000e+00, 1.9483e-06, ..., 0.0000e+00, |
| 3.1944e-07, 0.0000e+00], |
| ..., |
| [ 2.5146e-06, 0.0000e+00, 4.3958e-07, ..., 0.0000e+00, |
| 4.5635e-08, 0.0000e+00], |
| [-4.4424e-07, 0.0000e+00, -3.5241e-06, ..., 0.0000e+00, |
| 1.4156e-07, 0.0000e+00], |
| [ 4.0699e-07, 0.0000e+00, 5.8860e-07, ..., 0.0000e+00, |
| -8.3819e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 241, bias, value: tensor([-0.0063, 0.0239, -0.0019, -0.0052, 0.0148, 0.0057, 0.0249, -0.0119, |
| 0.0050, -0.0058], device='cuda:0'), grad: tensor([ 4.8317e-06, 1.8971e-06, -1.1455e-07, 4.8056e-07, -1.3271e-06, |
| 2.4453e-05, -2.7820e-05, 8.4937e-06, -1.3821e-05, 2.8722e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 240, time 218.12, cls_loss 0.0009 cls_loss_mapping 0.0008 cls_loss_causal 0.4641 re_mapping 0.0044 re_causal 0.0161 /// teacc 99.09 lr 0.00001000 |
| Epoch 242, weight, value: tensor([[-0.0562, -0.1094, -0.0602, ..., -0.1180, -0.0210, -0.0325], |
| [-0.0509, 0.0504, -0.1629, ..., 0.0634, 0.0727, -0.1375], |
| [ 0.0581, -0.0077, -0.0628, ..., -0.0603, -0.0454, 0.0088], |
| ..., |
| [ 0.0574, -0.0603, -0.0189, ..., 0.0170, -0.0425, 0.0174], |
| [-0.0692, -0.0271, -0.0829, ..., -0.0313, -0.0373, -0.0711], |
| [-0.1449, -0.2048, 0.0270, ..., -0.0661, 0.0413, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 1.5832e-08, 0.0000e+00, 5.5879e-09, ..., 0.0000e+00, |
| 1.8626e-08, 0.0000e+00], |
| [ 4.2003e-07, 0.0000e+00, 1.4435e-07, ..., 3.7253e-09, |
| -1.0338e-07, 0.0000e+00], |
| [-7.1246e-07, 0.0000e+00, -2.4214e-08, ..., 9.3132e-10, |
| 1.4901e-08, 0.0000e+00], |
| ..., |
| [-9.2201e-08, 0.0000e+00, -4.0978e-08, ..., -4.0047e-08, |
| 6.3889e-07, 0.0000e+00], |
| [ 4.1910e-08, 0.0000e+00, -2.9802e-08, ..., 0.0000e+00, |
| 7.0781e-08, 0.0000e+00], |
| [ 1.8161e-07, 0.0000e+00, 1.7136e-07, ..., 2.4214e-08, |
| -1.0300e-06, 0.0000e+00]], device='cuda:0') |
| Epoch 242, bias, value: tensor([-0.0063, 0.0239, -0.0019, -0.0051, 0.0147, 0.0057, 0.0248, -0.0120, |
| 0.0049, -0.0057], device='cuda:0'), grad: tensor([ 1.1269e-07, 1.0245e-06, -9.7975e-07, -6.3796e-07, 4.9639e-07, |
| 1.2433e-06, 1.0058e-07, 1.3551e-06, 6.0536e-08, -2.7735e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 241, time 218.15, cls_loss 0.0009 cls_loss_mapping 0.0007 cls_loss_causal 0.4478 re_mapping 0.0045 re_causal 0.0161 /// teacc 99.10 lr 0.00001000 |
| Epoch 243, weight, value: tensor([[-0.0562, -0.1094, -0.0603, ..., -0.1181, -0.0211, -0.0326], |
| [-0.0509, 0.0505, -0.1630, ..., 0.0634, 0.0728, -0.1376], |
| [ 0.0582, -0.0077, -0.0628, ..., -0.0603, -0.0455, 0.0087], |
| ..., |
| [ 0.0574, -0.0603, -0.0189, ..., 0.0170, -0.0426, 0.0174], |
| [-0.0694, -0.0271, -0.0830, ..., -0.0314, -0.0374, -0.0711], |
| [-0.1449, -0.2049, 0.0270, ..., -0.0661, 0.0415, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 4.8429e-08, 0.0000e+00, 1.4901e-08, ..., 0.0000e+00, |
| 1.2107e-08, 0.0000e+00], |
| [ 9.1922e-07, 0.0000e+00, 6.0536e-08, ..., 0.0000e+00, |
| -5.3085e-08, 0.0000e+00], |
| [-6.2399e-08, 0.0000e+00, 1.4901e-08, ..., 0.0000e+00, |
| -5.5879e-08, 0.0000e+00], |
| ..., |
| [-1.2908e-06, 0.0000e+00, 9.4995e-08, ..., 0.0000e+00, |
| 7.9162e-08, 0.0000e+00], |
| [ 6.7055e-08, 0.0000e+00, -3.7253e-09, ..., 0.0000e+00, |
| 6.5193e-09, 0.0000e+00], |
| [ 4.1910e-08, 0.0000e+00, 2.3022e-05, ..., 0.0000e+00, |
| -5.5879e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 243, bias, value: tensor([-0.0063, 0.0239, -0.0019, -0.0051, 0.0147, 0.0058, 0.0247, -0.0120, |
| 0.0048, -0.0057], device='cuda:0'), grad: tensor([ 1.2945e-07, 1.6727e-06, -3.2783e-07, 6.6124e-07, -5.6118e-05, |
| -1.3923e-06, 1.3961e-06, -1.9539e-06, 6.9849e-08, 5.5909e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 242, time 217.94, cls_loss 0.0008 cls_loss_mapping 0.0009 cls_loss_causal 0.4683 re_mapping 0.0046 re_causal 0.0163 /// teacc 99.09 lr 0.00001000 |
| Epoch 244, weight, value: tensor([[-0.0562, -0.1095, -0.0603, ..., -0.1181, -0.0211, -0.0326], |
| [-0.0509, 0.0508, -0.1631, ..., 0.0634, 0.0728, -0.1376], |
| [ 0.0583, -0.0077, -0.0628, ..., -0.0603, -0.0455, 0.0087], |
| ..., |
| [ 0.0574, -0.0605, -0.0191, ..., 0.0170, -0.0426, 0.0174], |
| [-0.0695, -0.0271, -0.0830, ..., -0.0314, -0.0374, -0.0711], |
| [-0.1450, -0.2049, 0.0268, ..., -0.0661, 0.0415, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 5.7742e-08, 0.0000e+00, 1.2107e-07, ..., 0.0000e+00, |
| 1.6298e-07, 0.0000e+00], |
| [ 3.0547e-07, 0.0000e+00, 6.8918e-08, ..., 0.0000e+00, |
| -1.2144e-06, 0.0000e+00], |
| [ 1.3877e-07, 0.0000e+00, 6.1933e-07, ..., 0.0000e+00, |
| 1.3905e-06, 0.0000e+00], |
| ..., |
| [-2.1011e-06, 0.0000e+00, 3.8184e-08, ..., 0.0000e+00, |
| 7.4320e-07, 6.5193e-09], |
| [-9.3132e-09, 0.0000e+00, 4.1164e-07, ..., 0.0000e+00, |
| 1.4901e-07, 9.3132e-10], |
| [ 9.5554e-07, 0.0000e+00, 2.2817e-07, ..., 0.0000e+00, |
| -3.0920e-07, -2.8871e-08]], device='cuda:0') |
| Epoch 244, bias, value: tensor([-0.0062, 0.0239, -0.0018, -0.0051, 0.0148, 0.0058, 0.0247, -0.0120, |
| 0.0048, -0.0058], device='cuda:0'), grad: tensor([ 1.6363e-06, -1.4044e-06, 9.6411e-06, 2.4773e-06, 6.1169e-06, |
| -5.4110e-07, -1.8746e-05, -3.8296e-06, 1.9372e-06, 2.7083e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 243, time 217.97, cls_loss 0.0008 cls_loss_mapping 0.0007 cls_loss_causal 0.4854 re_mapping 0.0045 re_causal 0.0166 /// teacc 99.10 lr 0.00001000 |
| Epoch 245, weight, value: tensor([[-0.0563, -0.1095, -0.0603, ..., -0.1181, -0.0211, -0.0326], |
| [-0.0509, 0.0509, -0.1632, ..., 0.0634, 0.0729, -0.1376], |
| [ 0.0584, -0.0077, -0.0628, ..., -0.0604, -0.0455, 0.0088], |
| ..., |
| [ 0.0574, -0.0606, -0.0191, ..., 0.0170, -0.0426, 0.0174], |
| [-0.0698, -0.0271, -0.0830, ..., -0.0314, -0.0375, -0.0711], |
| [-0.1450, -0.2049, 0.0267, ..., -0.0661, 0.0415, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 2.0582e-07, 0.0000e+00, 8.7544e-08, ..., 0.0000e+00, |
| 1.8720e-07, 0.0000e+00], |
| [ 2.5984e-07, 0.0000e+00, 1.6298e-07, ..., 0.0000e+00, |
| -5.3309e-06, 0.0000e+00], |
| [-8.6427e-06, 0.0000e+00, -4.4517e-07, ..., 0.0000e+00, |
| -3.7253e-09, 0.0000e+00], |
| ..., |
| [ 7.5400e-06, 0.0000e+00, 3.9302e-07, ..., 0.0000e+00, |
| 3.6508e-06, 0.0000e+00], |
| [ 4.4610e-07, 0.0000e+00, 3.0082e-07, ..., 0.0000e+00, |
| 1.5274e-07, 0.0000e+00], |
| [ 7.1712e-08, 0.0000e+00, 8.5495e-07, ..., 0.0000e+00, |
| 9.6578e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 245, bias, value: tensor([-0.0062, 0.0239, -0.0018, -0.0050, 0.0148, 0.0057, 0.0248, -0.0120, |
| 0.0047, -0.0058], device='cuda:0'), grad: tensor([-4.5542e-07, -1.5318e-05, -1.4447e-05, 3.1665e-07, -2.4922e-06, |
| 6.5193e-09, -4.9360e-08, 2.4512e-05, 1.8803e-06, 6.0573e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 244, time 217.95, cls_loss 0.0008 cls_loss_mapping 0.0008 cls_loss_causal 0.4894 re_mapping 0.0045 re_causal 0.0166 /// teacc 99.10 lr 0.00001000 |
| Epoch 246, weight, value: tensor([[-0.0563, -0.1096, -0.0603, ..., -0.1181, -0.0211, -0.0326], |
| [-0.0509, 0.0511, -0.1632, ..., 0.0634, 0.0730, -0.1376], |
| [ 0.0585, -0.0078, -0.0628, ..., -0.0604, -0.0455, 0.0088], |
| ..., |
| [ 0.0573, -0.0608, -0.0193, ..., 0.0169, -0.0427, 0.0173], |
| [-0.0701, -0.0271, -0.0830, ..., -0.0315, -0.0375, -0.0711], |
| [-0.1451, -0.2049, 0.0266, ..., -0.0662, 0.0416, 0.0481]], |
| device='cuda:0'), grad: tensor([[ 6.9849e-08, 0.0000e+00, -7.1712e-08, ..., 9.3132e-10, |
| 1.4901e-08, 8.3819e-09], |
| [ 1.7975e-07, 0.0000e+00, 6.1616e-06, ..., 9.1735e-07, |
| 1.8338e-06, 4.6566e-09], |
| [ 9.0338e-07, 0.0000e+00, 4.8429e-08, ..., 3.7253e-09, |
| 2.7008e-08, 8.3819e-08], |
| ..., |
| [-4.5914e-07, 0.0000e+00, 2.3935e-07, ..., 6.5193e-09, |
| 1.5926e-07, 2.1420e-08], |
| [ 3.5390e-08, 0.0000e+00, 7.2643e-08, ..., 0.0000e+00, |
| 1.1176e-08, 3.7253e-09], |
| [ 8.1956e-08, 0.0000e+00, 3.5651e-06, ..., 4.9639e-07, |
| 1.1204e-06, 7.4506e-09]], device='cuda:0') |
| Epoch 246, bias, value: tensor([-0.0062, 0.0240, -0.0017, -0.0050, 0.0148, 0.0057, 0.0247, -0.0121, |
| 0.0046, -0.0058], device='cuda:0'), grad: tensor([-3.7961e-06, 1.4223e-05, 1.6093e-06, -1.5646e-06, -2.2992e-05, |
| 1.1595e-06, 1.9930e-06, 2.4866e-07, 2.8033e-07, 8.8140e-06], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 245, time 217.97, cls_loss 0.0010 cls_loss_mapping 0.0007 cls_loss_causal 0.4716 re_mapping 0.0044 re_causal 0.0160 /// teacc 99.06 lr 0.00001000 |
| Epoch 247, weight, value: tensor([[-0.0564, -0.1097, -0.0602, ..., -0.1182, -0.0212, -0.0326], |
| [-0.0510, 0.0512, -0.1634, ..., 0.0634, 0.0731, -0.1376], |
| [ 0.0586, -0.0077, -0.0628, ..., -0.0604, -0.0456, 0.0087], |
| ..., |
| [ 0.0574, -0.0609, -0.0194, ..., 0.0169, -0.0428, 0.0172], |
| [-0.0703, -0.0271, -0.0831, ..., -0.0315, -0.0377, -0.0711], |
| [-0.1452, -0.2049, 0.0266, ..., -0.0662, 0.0417, 0.0482]], |
| device='cuda:0'), grad: tensor([[ 1.0245e-08, 0.0000e+00, 4.6566e-09, ..., 0.0000e+00, |
| 1.5944e-06, 1.8626e-09], |
| [ 1.2852e-07, 0.0000e+00, 8.3819e-08, ..., 0.0000e+00, |
| -1.0617e-07, 2.7940e-09], |
| [-7.4506e-09, 0.0000e+00, 4.0047e-08, ..., 0.0000e+00, |
| 2.2352e-08, 0.0000e+00], |
| ..., |
| [-6.7055e-08, 0.0000e+00, 7.7300e-08, ..., 5.5879e-09, |
| 7.9442e-07, 1.2293e-07], |
| [ 1.4901e-08, 0.0000e+00, 1.3970e-08, ..., 0.0000e+00, |
| 3.7253e-08, 0.0000e+00], |
| [ 3.4459e-08, 0.0000e+00, 3.1114e-05, ..., -2.2352e-08, |
| -2.0117e-06, -3.1479e-07]], device='cuda:0') |
| Epoch 247, bias, value: tensor([-0.0061, 0.0240, -0.0017, -0.0050, 0.0148, 0.0057, 0.0248, -0.0121, |
| 0.0044, -0.0059], device='cuda:0'), grad: tensor([ 1.9282e-05, 2.9895e-07, 1.5646e-07, -8.5682e-08, -9.7990e-05, |
| 3.8743e-06, -2.3216e-05, 3.5446e-06, 2.1420e-07, 9.3937e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 246, time 217.96, cls_loss 0.0010 cls_loss_mapping 0.0008 cls_loss_causal 0.4909 re_mapping 0.0044 re_causal 0.0165 /// teacc 99.06 lr 0.00001000 |
| Epoch 248, weight, value: tensor([[-0.0564, -0.1098, -0.0602, ..., -0.1182, -0.0213, -0.0326], |
| [-0.0509, 0.0514, -0.1635, ..., 0.0636, 0.0733, -0.1376], |
| [ 0.0586, -0.0078, -0.0628, ..., -0.0604, -0.0457, 0.0087], |
| ..., |
| [ 0.0573, -0.0610, -0.0195, ..., 0.0168, -0.0430, 0.0171], |
| [-0.0705, -0.0271, -0.0832, ..., -0.0316, -0.0378, -0.0711], |
| [-0.1452, -0.2049, 0.0264, ..., -0.0662, 0.0417, 0.0482]], |
| device='cuda:0'), grad: tensor([[ 7.4506e-08, 6.5193e-09, -9.6858e-08, ..., 0.0000e+00, |
| 1.1455e-07, 0.0000e+00], |
| [-1.0006e-05, -3.0398e-06, 2.2743e-06, ..., 0.0000e+00, |
| 2.7474e-07, 0.0000e+00], |
| [ 8.7768e-06, 2.6524e-06, 4.0047e-08, ..., 0.0000e+00, |
| 2.4717e-06, 0.0000e+00], |
| ..., |
| [ 1.4026e-06, 3.5856e-07, 9.4995e-08, ..., 0.0000e+00, |
| 3.8743e-07, 0.0000e+00], |
| [ 3.7160e-07, 1.8626e-09, 1.4693e-05, ..., 0.0000e+00, |
| 2.1145e-05, 0.0000e+00], |
| [ 9.4995e-08, 0.0000e+00, 4.7907e-06, ..., 0.0000e+00, |
| -6.0536e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 248, bias, value: tensor([-0.0061, 0.0241, -0.0017, -0.0049, 0.0148, 0.0056, 0.0248, -0.0122, |
| 0.0044, -0.0059], device='cuda:0'), grad: tensor([-1.3225e-07, -2.0951e-05, 3.2604e-05, -8.4937e-07, -1.3880e-05, |
| -1.9765e-04, 7.2956e-05, 5.0738e-06, 1.0872e-04, 1.3739e-05], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 247, time 217.99, cls_loss 0.0012 cls_loss_mapping 0.0010 cls_loss_causal 0.4750 re_mapping 0.0044 re_causal 0.0158 /// teacc 99.03 lr 0.00001000 |
| Epoch 249, weight, value: tensor([[-0.0564, -0.1098, -0.0603, ..., -0.1183, -0.0214, -0.0326], |
| [-0.0510, 0.0514, -0.1637, ..., 0.0636, 0.0735, -0.1377], |
| [ 0.0586, -0.0078, -0.0628, ..., -0.0605, -0.0458, 0.0086], |
| ..., |
| [ 0.0574, -0.0611, -0.0197, ..., 0.0167, -0.0431, 0.0171], |
| [-0.0706, -0.0272, -0.0832, ..., -0.0316, -0.0380, -0.0711], |
| [-0.1453, -0.2050, 0.0264, ..., -0.0664, 0.0419, 0.0482]], |
| device='cuda:0'), grad: tensor([[ 4.7777e-07, 9.3132e-10, 1.1176e-08, ..., 0.0000e+00, |
| 2.7008e-08, 0.0000e+00], |
| [ 3.9767e-07, -4.0699e-07, 1.5832e-08, ..., 0.0000e+00, |
| -8.4192e-07, 0.0000e+00], |
| [-1.7695e-06, 1.0245e-08, -1.1642e-07, ..., 0.0000e+00, |
| 3.4459e-08, 0.0000e+00], |
| ..., |
| [ 3.7067e-07, 3.8184e-07, 2.4214e-08, ..., 0.0000e+00, |
| 7.2829e-07, 0.0000e+00], |
| [ 3.8221e-06, 5.5879e-09, 1.2107e-08, ..., 0.0000e+00, |
| 4.0047e-08, 0.0000e+00], |
| [ 9.4995e-08, 0.0000e+00, 1.5832e-08, ..., 0.0000e+00, |
| -1.0151e-07, 0.0000e+00]], device='cuda:0') |
| Epoch 249, bias, value: tensor([-0.0061, 0.0242, -0.0017, -0.0048, 0.0148, 0.0056, 0.0247, -0.0123, |
| 0.0042, -0.0059], device='cuda:0'), grad: tensor([ 5.9325e-07, -1.1474e-06, -7.4543e-06, -6.0275e-06, 5.3085e-07, |
| 2.7977e-06, -6.1467e-08, 1.8887e-06, 9.1195e-06, -2.4028e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 248, time 217.94, cls_loss 0.0009 cls_loss_mapping 0.0008 cls_loss_causal 0.4825 re_mapping 0.0043 re_causal 0.0161 /// teacc 99.07 lr 0.00001000 |
| Epoch 250, weight, value: tensor([[-0.0565, -0.1099, -0.0603, ..., -0.1183, -0.0214, -0.0326], |
| [-0.0510, 0.0515, -0.1638, ..., 0.0636, 0.0735, -0.1377], |
| [ 0.0587, -0.0078, -0.0628, ..., -0.0605, -0.0459, 0.0086], |
| ..., |
| [ 0.0574, -0.0611, -0.0198, ..., 0.0167, -0.0432, 0.0171], |
| [-0.0707, -0.0272, -0.0833, ..., -0.0317, -0.0379, -0.0711], |
| [-0.1453, -0.2050, 0.0263, ..., -0.0664, 0.0419, 0.0482]], |
| device='cuda:0'), grad: tensor([[ 1.6857e-07, 0.0000e+00, 1.2107e-08, ..., 0.0000e+00, |
| 8.0094e-08, 0.0000e+00], |
| [ 3.0454e-07, 0.0000e+00, 7.5437e-08, ..., 0.0000e+00, |
| -2.2631e-07, 0.0000e+00], |
| [-6.8098e-06, 0.0000e+00, 7.0781e-08, ..., 0.0000e+00, |
| 3.0734e-08, 0.0000e+00], |
| ..., |
| [ 6.0499e-06, 0.0000e+00, 9.1270e-08, ..., 0.0000e+00, |
| 8.4750e-08, 0.0000e+00], |
| [ 1.1921e-07, 0.0000e+00, 3.7253e-09, ..., 0.0000e+00, |
| 2.7940e-08, 0.0000e+00], |
| [ 8.5682e-08, 0.0000e+00, 4.1910e-08, ..., 0.0000e+00, |
| 6.8918e-08, 0.0000e+00]], device='cuda:0') |
| Epoch 250, bias, value: tensor([-0.0060, 0.0242, -0.0017, -0.0049, 0.0149, 0.0056, 0.0246, -0.0122, |
| 0.0042, -0.0059], device='cuda:0'), grad: tensor([ 4.5355e-07, 3.3993e-07, -9.6262e-06, 1.1735e-07, 3.1199e-07, |
| 3.3528e-07, -9.0897e-07, 8.1509e-06, 2.8033e-07, 5.3365e-07], |
| device='cuda:0') |
| 100 |
| 1e-05 |
| changing lr |
| epoch 249, time 217.65, cls_loss 0.0009 cls_loss_mapping 0.0008 cls_loss_causal 0.4359 re_mapping 0.0043 re_causal 0.0156 /// teacc 99.02 lr 0.00001000 |
| ---------------------saving last model at epoch 249---------------------------------------------------- |
| /home/yuqian_fu |
| {'gpu': '0', 'svroot': '/data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/saved-digit/CA_multiple_14fa_all_ep250_lr1e-4_lr_schedulerStep0.8_bs32_lamCa_1_lamRe_1_cls1_adt2_EW2_100_rmTrue_rnTrue_str3_WithStyleAttackExp1_epoch250', 'svpath': '/data/work-gcp-europe-west4-a/yuqian_fu/datasets/SingleSourceDG/saved-digit/CA_multiple_14fa_all_ep250_lr1e-4_lr_schedulerStep0.8_bs32_lamCa_1_lamRe_1_cls1_adt2_EW2_100_rmTrue_rnTrue_str3_WithStyleAttackExp1_epoch250/14factor_last.csv', 'channels': 3, 'factor_num': 14, 'stride': 3, 'epoch': 'last', 'eval_mapping': True} |
| loading weight of last |
| randm: False |
| stride: 3 |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| loading weight of last |
| Using downloaded and verified file: /home/yuqian_fu/.pytorch/SVHN/test_32x32.mat |
| mnist mnist_FA ... usps_FA Avg |
| ShearX 98.879997 98.769997 ... 76.482315 71.892851 |
| ShearY 98.739998 98.629997 ... 76.482315 62.471693 |
| AutoContrast 98.970001 99.000000 ... 76.482315 58.870535 |
| Invert 98.750000 97.680000 ... 76.482315 62.949374 |
| Equalize 98.329994 97.570000 ... 76.482315 69.557647 |
| Solarize 98.070000 97.029999 ... 76.482315 63.104321 |
| SolarizeAdd 98.239998 97.389999 ... 76.482315 69.777911 |
| Posterize 99.010002 98.849998 ... 76.482315 73.372108 |
| Contrast 99.049995 98.970001 ... 76.482315 68.781693 |
| Color 99.089996 99.019997 ... 76.482315 63.887585 |
| Brightness 99.000000 98.979996 ... 76.482315 66.205976 |
| Sharpness 99.019997 98.979996 ... 76.482315 71.537817 |
| NoiseSalt 99.119995 98.970001 ... 76.482315 57.677266 |
| NoiseGaussian 99.110001 99.010002 ... 76.482315 58.686552 |
| w/o do (original x) 99.020000 0.000000 ... 0.000000 72.226558 |
|
|
| [15 rows x 11 columns] |
| mnist svhn mnist_m syndigit usps Avg |
| do 99.04 66.122465 78.591268 75.662096 80.568012 75.23596 |
|
|
