cuda:0
----------------------------------------------------------------
Layer (type) Output Shape Param #
================================================================
LayerNorm-1 [-1, 60, 4000] 8,000
Linear-2 [-1, 60, 12000] 48,000,000
Dropout-3 [-1, 8, 60, 60] 0
Linear-4 [-1, 60, 4000] 16,004,000
Dropout-5 [-1, 60, 4000] 0
Attention-6 [-1, 60, 4000] 0
Identity-7 [-1, 60, 4000] 0
LayerNorm-8 [-1, 60, 4000] 8,000
Linear-9 [-1, 60, 16000] 64,016,000
GELU-10 [-1, 60, 16000] 0
Dropout-11 [-1, 60, 16000] 0
Linear-12 [-1, 60, 4000] 64,004,000
Dropout-13 [-1, 60, 4000] 0
Mlp-14 [-1, 60, 4000] 0
Identity-15 [-1, 60, 4000] 0
Block-16 [-1, 60, 4000] 0
Linear-17 [-1, 512] 2,048,512
ReLU-18 [-1, 512] 0
Linear-19 [-1, 3] 1,539
================================================================
Total params: 194,090,051
Trainable params: 194,090,051
Non-trainable params: 0
----------------------------------------------------------------
Input size (MB): 0.90
Forward/backward pass size (MB): 47.83
Params size (MB): 740.39
Estimated Total Size (MB): 789.13
----------------------------------------------------------------
None
data/BCICIV_calib_ds1e_1000Hz.mat
data/BCICIV_calib_ds1b_1000Hz.mat
data/BCICIV_calib_ds1g_1000Hz.mat
data/BCICIV_calib_ds1a_1000Hz.mat
data/BCICIV_calib_ds1f_1000Hz.mat
data/BCICIV_calib_ds1c_1000Hz.mat
data/BCICIV_calib_ds1d_1000Hz.mat
Epoch 1/50
----------
train Loss: 0.9897 Acc: 0.4710
val Loss: 0.8823 Acc: 0.5381
test Loss: 0.9435 Acc: 0.4700
Epoch 2/50
----------
train Loss: 0.9208 Acc: 0.4870
val Loss: 0.8754 Acc: 0.4873
test Loss: 0.9264 Acc: 0.4650
Epoch 3/50
----------
train Loss: 0.9026 Acc: 0.5130
val Loss: 0.8700 Acc: 0.5584
test Loss: 0.9040 Acc: 0.4850
Epoch 4/50
----------
train Loss: 0.8834 Acc: 0.5180
val Loss: 0.8779 Acc: 0.5330
test Loss: 0.9189 Acc: 0.4950
Epoch 5/50
----------
train Loss: 0.8793 Acc: 0.5160
val Loss: 0.8913 Acc: 0.5025
test Loss: 0.9220 Acc: 0.5100
Epoch 6/50
----------
train Loss: 0.8767 Acc: 0.5150
val Loss: 0.8924 Acc: 0.4721
test Loss: 0.9378 Acc: 0.4800
Epoch 7/50
----------
train Loss: 0.8623 Acc: 0.5440
val Loss: 0.9134 Acc: 0.4873
test Loss: 0.9279 Acc: 0.4950
Epoch 8/50
----------
train Loss: 0.8600 Acc: 0.5560
val Loss: 0.8563 Acc: 0.5330
test Loss: 0.8663 Acc: 0.5100
Epoch 9/50
----------
train Loss: 0.8447 Acc: 0.5550
val Loss: 0.9175 Acc: 0.5127
test Loss: 0.9322 Acc: 0.5050
Epoch 10/50
----------
train Loss: 0.8349 Acc: 0.5590
val Loss: 0.8813 Acc: 0.5025
test Loss: 0.9354 Acc: 0.4900
Epoch 11/50
----------
train Loss: 0.8307 Acc: 0.5530
val Loss: 0.9074 Acc: 0.4822
test Loss: 0.9585 Acc: 0.4350
Epoch 12/50
----------
train Loss: 0.8165 Acc: 0.5650
val Loss: 0.8927 Acc: 0.4873
test Loss: 0.9250 Acc: 0.5150
Epoch 13/50
----------
train Loss: 0.8270 Acc: 0.5560
val Loss: 0.9144 Acc: 0.4822
test Loss: 0.9091 Acc: 0.5000
Epoch 14/50
----------
train Loss: 0.8104 Acc: 0.5890
val Loss: 0.9114 Acc: 0.4873
test Loss: 0.9494 Acc: 0.5050
Epoch 15/50
----------
train Loss: 0.7973 Acc: 0.5780
val Loss: 0.9104 Acc: 0.5178
test Loss: 0.9146 Acc: 0.4750
Epoch 16/50
----------
train Loss: 0.8029 Acc: 0.5750
val Loss: 0.9151 Acc: 0.4924
test Loss: 0.8825 Acc: 0.5400
Epoch 17/50
----------
train Loss: 0.8019 Acc: 0.5960
val Loss: 0.9338 Acc: 0.5127
test Loss: 0.9370 Acc: 0.4550
Epoch 18/50
----------
train Loss: 0.7804 Acc: 0.6010
val Loss: 1.0304 Acc: 0.4873
test Loss: 0.9753 Acc: 0.4850
Epoch 19/50
----------
train Loss: 0.7987 Acc: 0.6190
val Loss: 0.9376 Acc: 0.5127
test Loss: 0.9415 Acc: 0.4750
Epoch 20/50
----------
train Loss: 0.7915 Acc: 0.5950
val Loss: 0.9324 Acc: 0.4772
test Loss: 0.9306 Acc: 0.5050
Epoch 21/50
----------
train Loss: 0.7755 Acc: 0.5870
val Loss: 0.9801 Acc: 0.4467
test Loss: 0.9802 Acc: 0.4650
Epoch 22/50
----------
train Loss: 0.7635 Acc: 0.6240
val Loss: 0.9781 Acc: 0.4873
test Loss: 0.9421 Acc: 0.5050
Epoch 23/50
----------
train Loss: 0.7750 Acc: 0.5970
val Loss: 1.0491 Acc: 0.4569
test Loss: 0.9612 Acc: 0.4850
Epoch 24/50
----------
train Loss: 0.7645 Acc: 0.5910
val Loss: 0.9290 Acc: 0.4924
test Loss: 0.9007 Acc: 0.5350
Epoch 25/50
----------
train Loss: 0.7625 Acc: 0.6190
val Loss: 0.9573 Acc: 0.5127
test Loss: 0.9385 Acc: 0.4950
Epoch 26/50
----------
train Loss: 0.7664 Acc: 0.6320
val Loss: 1.0182 Acc: 0.5127
test Loss: 0.9506 Acc: 0.5150
Epoch 27/50
----------
train Loss: 0.7564 Acc: 0.6180
val Loss: 0.9719 Acc: 0.5127
test Loss: 0.8991 Acc: 0.5400
Epoch 28/50
----------
train Loss: 0.7628 Acc: 0.6390
val Loss: 0.9616 Acc: 0.5127
test Loss: 0.9402 Acc: 0.5150
Epoch 29/50
----------
train Loss: 0.7669 Acc: 0.6040
val Loss: 0.9729 Acc: 0.4975
test Loss: 0.9425 Acc: 0.5050
Epoch 30/50
----------
train Loss: 0.7406 Acc: 0.6200
val Loss: 1.0118 Acc: 0.5482
test Loss: 1.0262 Acc: 0.4700
Epoch 31/50
----------
train Loss: 0.7598 Acc: 0.6250
val Loss: 1.0333 Acc: 0.5025
test Loss: 1.0044 Acc: 0.4500
Epoch 32/50
----------
train Loss: 0.7624 Acc: 0.6190
val Loss: 0.9830 Acc: 0.5330
test Loss: 0.9604 Acc: 0.5050
Epoch 33/50
----------
train Loss: 0.7805 Acc: 0.5790
val Loss: 0.9523 Acc: 0.5025
test Loss: 0.9751 Acc: 0.4750
Epoch 34/50
----------
train Loss: 0.7293 Acc: 0.6170
val Loss: 1.0755 Acc: 0.4924
test Loss: 0.9924 Acc: 0.5100
Epoch 35/50
----------
train Loss: 0.7344 Acc: 0.6130
val Loss: 1.0430 Acc: 0.5025
test Loss: 0.9844 Acc: 0.5200
Epoch 36/50
----------
train Loss: 0.7509 Acc: 0.6200
val Loss: 1.0191 Acc: 0.5127
test Loss: 0.9790 Acc: 0.5600
Epoch 37/50
----------
train Loss: 0.7274 Acc: 0.6510
val Loss: 1.0435 Acc: 0.4619
test Loss: 1.0777 Acc: 0.4400
Epoch 38/50
----------
train Loss: 0.7573 Acc: 0.6340
val Loss: 0.9202 Acc: 0.4822
test Loss: 0.9732 Acc: 0.4950
Epoch 39/50
----------
train Loss: 0.7131 Acc: 0.6470
val Loss: 1.0729 Acc: 0.4873
test Loss: 1.0118 Acc: 0.4650
Epoch 40/50
----------
train Loss: 0.7328 Acc: 0.6540
val Loss: 1.0579 Acc: 0.4873
test Loss: 1.0306 Acc: 0.5100
Epoch 41/50
----------
train Loss: 0.7222 Acc: 0.6410
val Loss: 0.9916 Acc: 0.4975
test Loss: 1.0313 Acc: 0.4800
Epoch 42/50
----------
train Loss: 0.7366 Acc: 0.6420
val Loss: 1.0321 Acc: 0.5025
test Loss: 0.9789 Acc: 0.5050
Epoch 43/50
----------
train Loss: 0.7330 Acc: 0.6360
val Loss: 0.9605 Acc: 0.4721
test Loss: 0.9759 Acc: 0.5250
Epoch 44/50
----------
train Loss: 0.7170 Acc: 0.6320
val Loss: 1.0656 Acc: 0.4619
test Loss: 0.9755 Acc: 0.5100
Epoch 45/50
----------
train Loss: 0.7201 Acc: 0.6510
val Loss: 0.9788 Acc: 0.5025
test Loss: 0.9803 Acc: 0.5150
Epoch 46/50
----------
train Loss: 0.7363 Acc: 0.6610
val Loss: 1.0472 Acc: 0.4721
test Loss: 0.9373 Acc: 0.5400
Epoch 47/50
----------
train Loss: 0.7054 Acc: 0.6650
val Loss: 1.0280 Acc: 0.4873
test Loss: 0.9511 Acc: 0.5200
Epoch 48/50
----------
train Loss: 0.6908 Acc: 0.6690
val Loss: 1.0956 Acc: 0.4924
test Loss: 1.0733 Acc: 0.4500
Epoch 49/50
----------
train Loss: 0.7180 Acc: 0.6470
val Loss: 1.0237 Acc: 0.5076
test Loss: 0.9955 Acc: 0.4700
Epoch 50/50
----------
train Loss: 0.7090 Acc: 0.6440
val Loss: 1.0600 Acc: 0.4975
test Loss: 1.0964 Acc: 0.4500
Training complete in 60m 4s
Best val Acc: 0.558376
Datasets
Models
