Dataset Viewer
Auto-converted to Parquet Duplicate
Filename
string
Depth
int64
Activations-Params
string
Activation Function
string
Total Activations
int64
Total Parameters
int64
Batch Size
int64
Max GPU Memory (MiB)
int64
Avg GPUTL
float64
Avg GRACT
float64
Avg SMACT
float64
Avg SMOCC
float64
Avg FP32A
float64
Conv2d Count
int64
BatchNorm2d Count
int64
Dropout Count
int64
AdaptiveAvgPool2d Count
int64
Linear Count
int64
Status
string
Input Size (MB)
float64
Forward/Backward Pass Size (MB)
float64
Params Size (MB)
float64
Estimated Total Size (MB)
float64
input_channels:3_num_classes:453_depth:22_arch:uniform_base_filters:98_batch:62_input_size:174_act:gelu_dropout:0.3442169060458645_dropout:True_batchnorm:False.out
22
[('conv2d', 98, 174, 174, 62, 2744), ('GELU', 98, 174, 174, 62, 0), ('dropout', 98, 174, 174, 62, 0), ('conv2d', 98, 174, 174, 62, 86534), ('GELU', 98, 174, 174, 62, 0), ('dropout', 98, 174, 174, 62, 0), ('conv2d', 98, 174, 174, 62, 86534), ('GELU', 98, 174, 174, 62, 0), ('dropout', 98, 174, 174, 62, 0), ('conv2d', 98,...
GELU
195,826,172
1,864,805
62
39,517
-1
-1
-1
-1
-1
22
0
22
1
1
SUCCESSFUL
21.7
123,506.48
7.11
123,535.29
input_channels:1_num_classes:167_depth:11_arch:pyramid_base_filters:70_batch:46_input_size:118_act:selu_dropout:0.19151156817540396_dropout:True_batchnorm:True.out
11
[('conv2d', 70, 118, 118, 46, 700), ('batchnorm2d', 70, 118, 118, 46, 140), ('SELU', 70, 118, 118, 46, 0), ('dropout', 70, 118, 118, 46, 0), ('conv2d', 95, 118, 118, 46, 59945), ('batchnorm2d', 95, 118, 118, 46, 190), ('SELU', 95, 118, 118, 46, 0), ('dropout', 95, 118, 118, 46, 0), ('conv2d', 129, 118, 118, 46, 110424)...
SELU
306,385,536
32,924,348
46
36,831
-1
-1
-1
-1
-1
11
11
11
1
1
OOM_CRASH
2.3
134,408.32
125.6
134,536.22
input_channels:3_num_classes:99_depth:14_arch:uniform_base_filters:121_batch:34_input_size:90_act:elu_dropout:0.37457174282124006_dropout:False_batchnorm:False.out
14
[('conv2d', 121, 90, 90, 34, 3388), ('ELU', 121, 90, 90, 34, 0), ('conv2d', 121, 90, 90, 34, 131890), ('ELU', 121, 90, 90, 34, 0), ('conv2d', 121, 90, 90, 34, 131890), ('ELU', 121, 90, 90, 34, 0), ('conv2d', 121, 90, 90, 34, 131890), ('ELU', 121, 90, 90, 34, 0), ('conv2d', 121, 90, 90, 34, 131890), ('ELU', 121, 90, 90,...
ELU
27,443,119
1,730,036
34
6,293
90.222222
0.913875
0.84186
0.44707
0.023073
14
0
0
1
1
SUCCESSFUL
3.06
10,678.04
6.6
10,687.7
input_channels:3_num_classes:95_depth:16_arch:uniform_base_filters:73_batch:10_input_size:160_act:elu_dropout:0.25411316426075414_dropout:True_batchnorm:True.out
16
[('conv2d', 73, 160, 160, 10, 2044), ('batchnorm2d', 73, 160, 160, 10, 146), ('ELU', 73, 160, 160, 10, 0), ('dropout', 73, 160, 160, 10, 0), ('conv2d', 73, 160, 160, 10, 48034), ('batchnorm2d', 73, 160, 160, 10, 146), ('ELU', 73, 160, 160, 10, 0), ('dropout', 73, 160, 160, 10, 0), ('conv2d', 73, 160, 160, 10, 48034), (...
ELU
119,603,463
731,920
10
6,475
93.703125
0.932032
0.884245
0.391189
0.033538
16
16
16
1
1
SUCCESSFUL
2.9
11,406.3
2.79
11,411.99
input_channels:1_num_classes:149_depth:12_arch:gradual_base_filters:64_batch:18_input_size:44_act:relu_dropout:0.4310027159231188_dropout:False_batchnorm:True.out
12
[('conv2d', 64, 44, 44, 18, 640), ('batchnorm2d', 64, 44, 44, 18, 128), ('ReLU', 64, 44, 44, 18, 0), ('conv2d', 73, 44, 44, 18, 42121), ('batchnorm2d', 73, 44, 44, 18, 146), ('ReLU', 73, 44, 44, 18, 0), ('conv2d', 85, 44, 44, 18, 55930), ('batchnorm2d', 85, 44, 44, 18, 170), ('ReLU', 85, 44, 44, 18, 0), ('conv2d', 98, ...
ReLU
11,111,315
2,960,902
18
2,779
88.328125
0.864873
0.758109
0.295291
0.015018
12
12
0
1
1
SUCCESSFUL
0.18
2,034.54
11.29
2,046.01
input_channels:3_num_classes:255_depth:24_arch:pyramid_base_filters:126_batch:2_input_size:50_act:mish_dropout:0.4268541547234518_dropout:True_batchnorm:True.out
24
[('conv2d', 126, 50, 50, 2, 3528), ('batchnorm2d', 126, 50, 50, 2, 252), ('Mish', 126, 50, 50, 2, 0), ('dropout', 126, 50, 50, 2, 0), ('conv2d', 141, 50, 50, 2, 160035), ('batchnorm2d', 141, 50, 50, 2, 282), ('Mish', 141, 50, 50, 2, 0), ('dropout', 141, 50, 50, 2, 0), ('conv2d', 158, 50, 50, 2, 200660), ('batchnorm2d',...
Mish
155,882,333
128,242,341
2
5,075
92.703125
0.915953
0.881
0.280288
0.020702
24
24
24
1
1
SUCCESSFUL
0.06
2,973.22
489.21
3,462.49
input_channels:1_num_classes:981_depth:29_arch:hourglass_base_filters:110_batch:10_input_size:102_act:softplus_dropout:0.170648409186261_dropout:False_batchnorm:False.out
29
[('conv2d', 110, 102, 102, 10, 1100), ('Softplus', 110, 102, 102, 10, 0), ('conv2d', 135, 102, 102, 10, 133785), ('Softplus', 135, 102, 102, 10, 0), ('conv2d', 167, 102, 102, 10, 203072), ('Softplus', 167, 102, 102, 10, 0), ('conv2d', 205, 102, 102, 10, 308320), ('Softplus', 205, 102, 102, 10, 0), ('conv2d', 253, 102, ...
Softplus
360,667,136
148,861,286
10
23,237
-1
-1
-1
-1
-1
29
0
0
1
1
SUCCESSFUL
0.4
41,275
567.86
41,843.26
input_channels:1_num_classes:169_depth:2_arch:uniform_base_filters:32_batch:52_input_size:134_act:mish_dropout:0.4252372788964438_dropout:False_batchnorm:True.out
2
[('conv2d', 32, 134, 134, 52, 320), ('batchnorm2d', 32, 134, 134, 52, 64), ('Mish', 32, 134, 134, 52, 0), ('conv2d', 32, 134, 134, 52, 9248), ('batchnorm2d', 32, 134, 134, 52, 64), ('Mish', 32, 134, 134, 52, 0), ('adaptive_avg_pool2d', 0, 0, 32, 52, 0), ('linear', 0, 0, 169, 52, 5577), ('softmax', 0, 0, 169, 52, 0)]
Mish
3,447,922
15,273
52
2,789
-1
-1
-1
-1
-1
2
2
0
1
1
SUCCESSFUL
3.64
1,823.64
0.06
1,827.34
input_channels:3_num_classes:685_depth:27_arch:residual_base_filters:18_batch:44_input_size:122_act:tanh_dropout:0.27334897812173853_dropout:False_batchnorm:True.out
24
[('conv2d', 18, 122, 122, 44, 72), ('batchnorm2d', 18, 122, 122, 44, 36), ('conv2d', 18, 122, 122, 44, 504), ('batchnorm2d', 18, 122, 122, 44, 36), ('Tanh', 18, 122, 122, 44, 0), ('conv2d', 18, 122, 122, 44, 2934), ('batchnorm2d', 18, 122, 122, 44, 36), ('Tanh', 18, 122, 122, 44, 0), ('conv2d', 21, 61, 61, 44, 399), ('...
Tanh
4,245,207
247,238
44
2,629
-1
-1
-1
-1
-1
24
24
0
1
1
SUCCESSFUL
7.48
1,959.32
0.94
1,967.74
input_channels:1_num_classes:895_depth:17_arch:bottleneck_base_filters:64_batch:56_input_size:16_act:gelu_dropout:0.24397616313556642_dropout:False_batchnorm:True.out
17
[('conv2d', 128, 16, 16, 56, 1280), ('batchnorm2d', 128, 16, 16, 56, 256), ('GELU', 128, 16, 16, 56, 0), ('conv2d', 64, 16, 16, 56, 73792), ('batchnorm2d', 64, 16, 16, 56, 128), ('GELU', 64, 16, 16, 56, 0), ('conv2d', 79, 16, 16, 56, 45583), ('batchnorm2d', 79, 16, 16, 56, 158), ('GELU', 79, 16, 16, 56, 0), ('conv2d', ...
GELU
6,397,039
57,496,575
56
4,659
89.983871
0.857462
0.872545
0.265182
0.018315
17
17
0
1
1
SUCCESSFUL
0
3,643.92
219.33
3,863.25
input_channels:1_num_classes:387_depth:15_arch:residual_base_filters:68_batch:2_input_size:78_act:gelu_dropout:0.15243320688471612_dropout:False_batchnorm:True.out
21
[('conv2d', 68, 78, 78, 2, 136), ('batchnorm2d', 68, 78, 78, 2, 136), ('conv2d', 68, 78, 78, 2, 680), ('batchnorm2d', 68, 78, 78, 2, 136), ('GELU', 68, 78, 78, 2, 0), ('conv2d', 68, 78, 78, 2, 41684), ('batchnorm2d', 68, 78, 78, 2, 136), ('GELU', 68, 78, 78, 2, 0), ('conv2d', 85, 39, 39, 2, 5865), ('batchnorm2d', 85, 3...
GELU
6,994,191
3,201,860
2
1,857
-1
-1
-1
-1
-1
21
21
0
1
1
SUCCESSFUL
0.04
146.74
12.21
158.99
input_channels:1_num_classes:798_depth:13_arch:gradual_base_filters:113_batch:60_input_size:20_act:relu_dropout:0.24301664048682203_dropout:False_batchnorm:False.out
13
[('conv2d', 113, 20, 20, 60, 1130), ('ReLU', 113, 20, 20, 60, 0), ('conv2d', 126, 20, 20, 60, 128268), ('ReLU', 126, 20, 20, 60, 0), ('conv2d', 141, 20, 20, 60, 160035), ('ReLU', 141, 20, 20, 60, 0), ('conv2d', 157, 20, 20, 60, 199390), ('ReLU', 157, 20, 20, 60, 0), ('conv2d', 176, 20, 20, 60, 248864), ('ReLU', 176, 20...
ReLU
2,495,626
7,272,879
60
2,553
78.885246
0.767095
0.694364
0.237618
0.006736
13
0
0
1
1
SUCCESSFUL
0
1,713
27.74
1,740.74
input_channels:1_num_classes:935_depth:13_arch:pyramid_base_filters:50_batch:32_input_size:158_act:prelu_dropout:0.25992523361899594_dropout:True_batchnorm:True.out
13
[('conv2d', 50, 158, 158, 32, 500), ('batchnorm2d', 50, 158, 158, 32, 100), ('PReLU', 50, 158, 158, 32, 0), ('dropout', 50, 158, 158, 32, 0), ('conv2d', 66, 158, 158, 32, 29766), ('batchnorm2d', 66, 158, 158, 32, 132), ('PReLU', 66, 158, 158, 32, 0), ('dropout', 66, 158, 158, 32, 0), ('conv2d', 88, 158, 158, 32, 52360)...
PReLU
603,033,793
38,204,027
32
38,665
32
0.0964
0
0
0.003
13
13
13
1
1
OOM_CRASH
3.2
184,031.04
145.74
184,179.98
input_channels:1_num_classes:939_depth:26_arch:gradual_base_filters:34_batch:58_input_size:32_act:leaky_relu_dropout:0.19926506805091357_dropout:True_batchnorm:False.out
26
[('conv2d', 34, 32, 32, 58, 340), ('LeakyReLU', 34, 32, 32, 58, 0), ('dropout', 34, 32, 32, 58, 0), ('conv2d', 36, 32, 32, 58, 11052), ('LeakyReLU', 36, 32, 32, 58, 0), ('dropout', 36, 32, 32, 58, 0), ('conv2d', 39, 32, 32, 58, 12675), ('LeakyReLU', 39, 32, 32, 58, 0), ('dropout', 39, 32, 32, 58, 0), ('conv2d', 43, 32,...
LeakyReLU
8,566,857
3,529,207
58
4,109
86.253968
0.852492
0.765396
0.315434
0.039926
26
0
26
1
1
SUCCESSFUL
0
5,054.12
13.46
5,067.58
input_channels:1_num_classes:300_depth:27_arch:hourglass_base_filters:122_batch:28_input_size:188_act:leaky_relu_dropout:0.35097002901905217_dropout:True_batchnorm:True.out
27
[('conv2d', 122, 188, 188, 28, 1220), ('batchnorm2d', 122, 188, 188, 28, 244), ('LeakyReLU', 122, 188, 188, 28, 0), ('dropout', 122, 188, 188, 28, 0), ('conv2d', 151, 188, 188, 28, 165949), ('batchnorm2d', 151, 188, 188, 28, 302), ('LeakyReLU', 151, 188, 188, 28, 0), ('dropout', 151, 188, 188, 28, 0), ('conv2d', 188, 1...
LeakyReLU
2,342,035,538
142,278,957
28
37,835
-1
-1
-1
-1
-1
27
27
27
1
1
OOM_CRASH
3.64
625,390.92
542.75
625,937.31
input_channels:3_num_classes:923_depth:5_arch:dense_base_filters:51_batch:52_input_size:80_act:swish_dropout:0.12169761684777597_dropout:True_batchnorm:False.out
4
[('conv2d', 32, 80, 80, 52, 896), ('SiLU', 32, 80, 80, 52, 0), ('dropout', 32, 80, 80, 52, 0), ('conv2d', 32, 80, 80, 52, 10112), ('SiLU', 32, 80, 80, 52, 0), ('dropout', 32, 80, 80, 52, 0), ('conv2d', 32, 80, 80, 52, 19328), ('SiLU', 32, 80, 80, 52, 0), ('dropout', 32, 80, 80, 52, 0), ('conv2d', 32, 80, 80, 52, 28544)...
SiLU
2,459,577
180,716
52
3,053
89.015625
0.685048
0.651792
0.366302
0.099407
4
0
4
1
1
SUCCESSFUL
3.64
1,633.32
0.69
1,637.65
input_channels:1_num_classes:814_depth:6_arch:bottleneck_base_filters:105_batch:56_input_size:184_act:gelu_dropout:0.4043234507861373_dropout:True_batchnorm:True.out
6
[('conv2d', 210, 184, 184, 56, 2100), ('batchnorm2d', 210, 184, 184, 56, 420), ('GELU', 210, 184, 184, 56, 0), ('dropout', 210, 184, 184, 56, 0), ('conv2d', 105, 184, 184, 56, 198555), ('batchnorm2d', 105, 184, 184, 56, 210), ('GELU', 105, 184, 184, 56, 0), ('dropout', 105, 184, 184, 56, 0), ('conv2d', 190, 184, 184, 5...
GELU
352,511,430
10,174,557
56
38,047
21.8
0.0626
0.001
0
0.004333
6
6
6
1
1
OOM_CRASH
7.28
188,261.36
38.81
188,307.45
input_channels:3_num_classes:810_depth:4_arch:uniform_base_filters:59_batch:30_input_size:2_act:prelu_dropout:0.4178472899065261_dropout:False_batchnorm:False.out
0
[]
null
0
0
30
0
48.5
0.298
0.618
0.123
0.017
0
0
0
0
0
SUCCESSFUL
0
0
0
0
input_channels:1_num_classes:76_depth:10_arch:residual_base_filters:93_batch:2_input_size:2_act:mish_dropout:0.3625645145273898_dropout:False_batchnorm:True.out
0
[]
null
0
0
2
0
6
0
0
0
0
0
0
0
0
0
SUCCESSFUL
0
0
0
0
input_channels:1_num_classes:956_depth:22_arch:uniform_base_filters:72_batch:30_input_size:150_act:prelu_dropout:0.16294183649715593_dropout:False_batchnorm:True.out
22
[('conv2d', 72, 150, 150, 30, 720), ('batchnorm2d', 72, 150, 150, 30, 144), ('PReLU', 72, 150, 150, 30, 0), ('conv2d', 72, 150, 150, 30, 46728), ('batchnorm2d', 72, 150, 150, 30, 144), ('PReLU', 72, 150, 150, 30, 0), ('conv2d', 72, 150, 150, 30, 46728), ('batchnorm2d', 72, 150, 150, 30, 144), ('PReLU', 72, 150, 150, 30...
PReLU
106,921,984
1,054,964
30
16,061
-1
-1
-1
-1
-1
22
22
0
1
1
SUCCESSFUL
2.7
32,629.8
4.02
32,636.52
input_channels:3_num_classes:435_depth:16_arch:residual_base_filters:120_batch:16_input_size:200_act:selu_dropout:0.2594242638726377_dropout:False_batchnorm:True.out
24
[('conv2d', 120, 200, 200, 16, 480), ('batchnorm2d', 120, 200, 200, 16, 240), ('conv2d', 120, 200, 200, 16, 3360), ('batchnorm2d', 120, 200, 200, 16, 240), ('SELU', 120, 200, 200, 16, 0), ('conv2d', 120, 200, 200, 16, 129720), ('batchnorm2d', 120, 200, 200, 16, 240), ('SELU', 120, 200, 200, 16, 0), ('conv2d', 143, 100,...
SELU
76,212,365
9,498,236
16
8,843
-1
-1
-1
-1
-1
24
24
0
1
1
SUCCESSFUL
7.36
12,792
36.23
12,835.59
input_channels:1_num_classes:157_depth:4_arch:pyramid_base_filters:61_batch:32_input_size:102_act:tanh_dropout:0.22763112287659637_dropout:False_batchnorm:False.out
4
[('conv2d', 61, 102, 102, 32, 610), ('Tanh', 61, 102, 102, 32, 0), ('conv2d', 146, 102, 102, 32, 80300), ('Tanh', 146, 102, 102, 32, 0), ('conv2d', 353, 102, 102, 32, 464195), ('Tanh', 353, 102, 102, 32, 0), ('conv2d', 850, 102, 102, 32, 2701300), ('Tanh', 850, 102, 102, 32, 0), ('adaptive_avg_pool2d', 0, 0, 850, 32, 0...
Tanh
29,340,444
3,380,012
32
8,899
-1
-1
-1
-1
-1
4
0
0
1
1
SUCCESSFUL
1.28
10,744.64
12.89
10,758.81
input_channels:3_num_classes:730_depth:28_arch:dense_base_filters:25_batch:2_input_size:170_act:leaky_relu_dropout:0.2718365616716558_dropout:False_batchnorm:False.out
29
[('conv2d', 32, 170, 170, 2, 896), ('LeakyReLU', 32, 170, 170, 2, 0), ('conv2d', 32, 170, 170, 2, 10112), ('LeakyReLU', 32, 170, 170, 2, 0), ('conv2d', 32, 170, 170, 2, 19328), ('LeakyReLU', 32, 170, 170, 2, 0), ('conv2d', 32, 170, 170, 2, 28544), ('LeakyReLU', 32, 170, 170, 2, 0), ('conv2d', 65, 170, 170, 2, 8580), ('...
LeakyReLU
22,255,293
3,669,944
2
2,857
-1
-1
-1
-1
-1
29
0
0
1
1
SUCCESSFUL
0.66
920.8
14
935.46
input_channels:1_num_classes:816_depth:21_arch:residual_base_filters:60_batch:22_input_size:158_act:selu_dropout:0.4546621145227434_dropout:True_batchnorm:True.out
30
[('conv2d', 60, 158, 158, 22, 120), ('batchnorm2d', 60, 158, 158, 22, 120), ('conv2d', 60, 158, 158, 22, 600), ('batchnorm2d', 60, 158, 158, 22, 120), ('SELU', 60, 158, 158, 22, 0), ('dropout', 60, 158, 158, 22, 0), ('conv2d', 60, 158, 158, 22, 32460), ('batchnorm2d', 60, 158, 158, 22, 120), ('SELU', 60, 158, 158, 22, ...
SELU
29,689,034
4,561,414
22
4,673
92.203125
0.887828
0.740571
0.354911
0.053309
30
30
20
1
1
SUCCESSFUL
2.2
6,478.12
17.4
6,497.72
input_channels:1_num_classes:431_depth:21_arch:uniform_base_filters:65_batch:18_input_size:76_act:softplus_dropout:0.3133880221557329_dropout:True_batchnorm:True.out
21
[('conv2d', 65, 76, 76, 18, 650), ('batchnorm2d', 65, 76, 76, 18, 130), ('Softplus', 65, 76, 76, 18, 0), ('dropout', 65, 76, 76, 18, 0), ('conv2d', 65, 76, 76, 18, 38090), ('batchnorm2d', 65, 76, 76, 18, 130), ('Softplus', 65, 76, 76, 18, 0), ('dropout', 65, 76, 76, 18, 0), ('conv2d', 65, 76, 76, 18, 38090), ('batchnor...
Softplus
31,537,887
793,626
18
3,825
91.952381
0.909667
0.799
0.32387
0.032
21
21
21
1
1
SUCCESSFUL
0.36
5,413.86
3.03
5,417.25
input_channels:1_num_classes:100_depth:22_arch:residual_base_filters:81_batch:58_input_size:222_act:softplus_dropout:0.3868177844563978_dropout:False_batchnorm:True.out
30
[('conv2d', 81, 222, 222, 58, 162), ('batchnorm2d', 81, 222, 222, 58, 162), ('conv2d', 81, 222, 222, 58, 810), ('batchnorm2d', 81, 222, 222, 58, 162), ('Softplus', 81, 222, 222, 58, 0), ('conv2d', 81, 222, 222, 58, 59130), ('batchnorm2d', 81, 222, 222, 58, 162), ('Softplus', 81, 222, 222, 58, 0), ('conv2d', 93, 111, 11...
Softplus
61,683,159
6,008,182
58
22,297
-1
-1
-1
-1
-1
30
30
0
1
1
SUCCESSFUL
11.02
37,530.64
22.92
37,564.58
input_channels:3_num_classes:85_depth:26_arch:hourglass_base_filters:76_batch:14_input_size:110_act:selu_dropout:0.2079112881861306_dropout:True_batchnorm:True.out
26
[('conv2d', 76, 110, 110, 14, 2128), ('batchnorm2d', 76, 110, 110, 14, 152), ('SELU', 76, 110, 110, 14, 0), ('dropout', 76, 110, 110, 14, 0), ('conv2d', 97, 110, 110, 14, 66445), ('batchnorm2d', 97, 110, 110, 14, 194), ('SELU', 97, 110, 110, 14, 0), ('dropout', 97, 110, 110, 14, 0), ('conv2d', 126, 110, 110, 14, 110124...
SELU
661,241,046
111,230,300
14
34,345
95.53125
0.930609
0.967054
0.227464
0.013055
26
26
26
1
1
SUCCESSFUL
1.96
88,285.26
424.31
88,711.53
input_channels:1_num_classes:387_depth:6_arch:dense_base_filters:100_batch:6_input_size:142_act:tanh_dropout:0.37042170333640667_dropout:True_batchnorm:True.out
4
[('conv2d', 32, 142, 142, 6, 320), ('batchnorm2d', 32, 142, 142, 6, 64), ('Tanh', 32, 142, 142, 6, 0), ('dropout', 32, 142, 142, 6, 0), ('conv2d', 32, 142, 142, 6, 9536), ('batchnorm2d', 32, 142, 142, 6, 64), ('Tanh', 32, 142, 142, 6, 0), ('dropout', 32, 142, 142, 6, 0), ('conv2d', 32, 142, 142, 6, 18752), ('batchnorm2...
Tanh
10,324,871
107,142
6
2,287
83.968254
0.800625
0.547309
0.256855
0.083158
4
4
4
1
1
SUCCESSFUL
0.48
709.86
0.41
710.75
input_channels:1_num_classes:782_depth:7_arch:hourglass_base_filters:55_batch:60_input_size:28_act:swish_dropout:0.4864820138912168_dropout:False_batchnorm:True.out
7
[('conv2d', 55, 28, 28, 60, 550), ('batchnorm2d', 55, 28, 28, 60, 110), ('SiLU', 55, 28, 28, 60, 0), ('conv2d', 183, 28, 28, 60, 90768), ('batchnorm2d', 183, 28, 28, 60, 366), ('SiLU', 183, 28, 28, 60, 0), ('conv2d', 613, 28, 28, 60, 1010224), ('batchnorm2d', 613, 28, 28, 60, 1226), ('SiLU', 613, 28, 28, 60, 0), ('conv...
SiLU
5,187,779
8,697,976
60
4,009
-1
-1
-1
-1
-1
7
7
0
1
1
SUCCESSFUL
0
3,166.2
33.18
3,199.38
input_channels:3_num_classes:633_depth:7_arch:uniform_base_filters:50_batch:34_input_size:182_act:leaky_relu_dropout:0.4001604314152226_dropout:True_batchnorm:True.out
7
[('conv2d', 50, 182, 182, 34, 1400), ('batchnorm2d', 50, 182, 182, 34, 100), ('LeakyReLU', 50, 182, 182, 34, 0), ('dropout', 50, 182, 182, 34, 0), ('conv2d', 50, 182, 182, 34, 22550), ('batchnorm2d', 50, 182, 182, 34, 100), ('LeakyReLU', 50, 182, 182, 34, 0), ('dropout', 50, 182, 182, 34, 0), ('conv2d', 50, 182, 182, 3...
LeakyReLU
46,374,916
169,683
34
8,179
-1
-1
-1
-1
-1
7
7
7
1
1
SUCCESSFUL
12.92
15,036.84
0.65
15,050.41
input_channels:3_num_classes:707_depth:3_arch:uniform_base_filters:43_batch:24_input_size:76_act:selu_dropout:0.1304754404253033_dropout:False_batchnorm:False.out
3
[('conv2d', 43, 76, 76, 24, 1204), ('SELU', 43, 76, 76, 24, 0), ('conv2d', 43, 76, 76, 24, 16684), ('SELU', 43, 76, 76, 24, 0), ('conv2d', 43, 76, 76, 24, 16684), ('SELU', 43, 76, 76, 24, 0), ('adaptive_avg_pool2d', 0, 0, 43, 24, 0), ('linear', 0, 0, 707, 24, 31108), ('softmax', 0, 0, 707, 24, 0)]
SELU
1,491,665
65,680
24
1,969
-1
-1
-1
-1
-1
3
0
0
1
1
SUCCESSFUL
1.68
409.68
0.25
411.61
input_channels:1_num_classes:295_depth:4_arch:gradual_base_filters:64_batch:48_input_size:88_act:softplus_dropout:0.3028756298236742_dropout:False_batchnorm:True.out
4
[('conv2d', 64, 88, 88, 48, 640), ('batchnorm2d', 64, 88, 88, 48, 128), ('Softplus', 64, 88, 88, 48, 0), ('conv2d', 98, 88, 88, 48, 56546), ('batchnorm2d', 98, 88, 88, 48, 196), ('Softplus', 98, 88, 88, 48, 0), ('conv2d', 152, 88, 88, 48, 134216), ('batchnorm2d', 152, 88, 88, 48, 304), ('Softplus', 152, 88, 88, 48, 0),...
Softplus
12,731,960
582,169
48
5,553
86.75
0.844766
0.836145
0.371691
0.046709
4
4
0
1
1
SUCCESSFUL
1.44
6,216.48
2.22
6,220.14
input_channels:1_num_classes:739_depth:7_arch:uniform_base_filters:36_batch:14_input_size:18_act:softplus_dropout:0.3255848303782941_dropout:False_batchnorm:True.out
7
[('conv2d', 36, 18, 18, 14, 360), ('batchnorm2d', 36, 18, 18, 14, 72), ('Softplus', 36, 18, 18, 14, 0), ('conv2d', 36, 18, 18, 14, 11700), ('batchnorm2d', 36, 18, 18, 14, 72), ('Softplus', 36, 18, 18, 14, 0), ('conv2d', 36, 18, 18, 14, 11700), ('batchnorm2d', 36, 18, 18, 14, 72), ('Softplus', 36, 18, 18, 14, 0), ('conv...
Softplus
246,458
98,407
14
1,755
32.129032
0.308333
0.087218
0.012764
0.003509
7
7
0
1
1
SUCCESSFUL
0
35
0.38
35.38
input_channels:1_num_classes:380_depth:3_arch:pyramid_base_filters:48_batch:44_input_size:86_act:gelu_dropout:0.4726838666864841_dropout:False_batchnorm:False.out
3
[('conv2d', 48, 86, 86, 44, 480), ('GELU', 48, 86, 86, 44, 0), ('conv2d', 167, 86, 86, 44, 72311), ('GELU', 167, 86, 86, 44, 0), ('conv2d', 586, 86, 86, 44, 881344), ('GELU', 586, 86, 86, 44, 0), ('adaptive_avg_pool2d', 0, 0, 586, 44, 0), ('linear', 0, 0, 380, 44, 223060), ('softmax', 0, 0, 380, 44, 0)]
GELU
11,849,738
1,177,195
44
5,741
88.246154
0.874469
0.908491
0.387164
0.038655
3
0
0
1
1
SUCCESSFUL
1.32
5,966.4
4.49
5,972.21
input_channels:3_num_classes:104_depth:9_arch:pyramid_base_filters:40_batch:12_input_size:108_act:swish_dropout:0.4487676269184536_dropout:True_batchnorm:False.out
9
[('conv2d', 40, 108, 108, 12, 1120), ('SiLU', 40, 108, 108, 12, 0), ('dropout', 40, 108, 108, 12, 0), ('conv2d', 61, 108, 108, 12, 22021), ('SiLU', 61, 108, 108, 12, 0), ('dropout', 61, 108, 108, 12, 0), ('conv2d', 95, 108, 108, 12, 52250), ('SiLU', 95, 108, 108, 12, 0), ('dropout', 95, 108, 108, 12, 0), ('conv2d', 148...
SiLU
127,967,274
17,547,546
12
8,687
-1
-1
-1
-1
-1
9
0
9
1
1
SUCCESSFUL
1.56
15,621
66.94
15,689.5
input_channels:1_num_classes:354_depth:6_arch:gradual_base_filters:122_batch:6_input_size:60_act:elu_dropout:0.16215820802589556_dropout:False_batchnorm:False.out
6
[('conv2d', 122, 60, 60, 6, 1220), ('ELU', 122, 60, 60, 6, 0), ('conv2d', 154, 60, 60, 6, 169246), ('ELU', 154, 60, 60, 6, 0), ('conv2d', 195, 60, 60, 6, 270465), ('ELU', 195, 60, 60, 6, 0), ('conv2d', 246, 60, 60, 6, 431976), ('ELU', 246, 60, 60, 6, 0), ('conv2d', 312, 60, 60, 6, 691080), ('ELU', 312, 60, 60, 6, 0), (...
ELU
10,253,903
2,813,726
6
2,151
86.84127
0.866762
0.758263
0.261614
0.015103
6
0
0
1
1
SUCCESSFUL
0.06
704.04
10.73
714.83
input_channels:3_num_classes:829_depth:15_arch:uniform_base_filters:122_batch:58_input_size:86_act:gelu_dropout:0.48577245512621836_dropout:True_batchnorm:True.out
15
[('conv2d', 122, 86, 86, 58, 3416), ('batchnorm2d', 122, 86, 86, 58, 244), ('GELU', 122, 86, 86, 58, 0), ('dropout', 122, 86, 86, 58, 0), ('conv2d', 122, 86, 86, 58, 134078), ('batchnorm2d', 122, 86, 86, 58, 244), ('GELU', 122, 86, 86, 58, 0), ('dropout', 122, 86, 86, 58, 0), ('conv2d', 122, 86, 86, 58, 134078), ('batc...
GELU
54,140,500
1,986,135
58
13,057
-1
-1
-1
-1
-1
15
15
15
1
1
SUCCESSFUL
4.64
29,946.56
7.58
29,958.78
input_channels:1_num_classes:479_depth:18_arch:residual_base_filters:89_batch:50_input_size:202_act:gelu_dropout:0.10865719578382925_dropout:True_batchnorm:False.out
27
[('conv2d', 89, 202, 202, 50, 178), ('conv2d', 89, 202, 202, 50, 890), ('GELU', 89, 202, 202, 50, 0), ('dropout', 89, 202, 202, 50, 0), ('conv2d', 89, 202, 202, 50, 71378), ('GELU', 89, 202, 202, 50, 0), ('dropout', 89, 202, 202, 50, 0), ('conv2d', 105, 101, 101, 50, 9450), ('conv2d', 105, 101, 101, 50, 84210), ('GELU'...
GELU
50,869,098
7,363,263
50
14,951
-1
-1
-1
-1
-1
27
0
18
1
1
SUCCESSFUL
8
30,493.5
28.09
30,529.59
input_channels:3_num_classes:959_depth:9_arch:bottleneck_base_filters:77_batch:18_input_size:128_act:relu_dropout:0.4963248953174547_dropout:False_batchnorm:True.out
9
[('conv2d', 154, 128, 128, 18, 4312), ('batchnorm2d', 154, 128, 128, 18, 308), ('ReLU', 154, 128, 128, 18, 0), ('conv2d', 77, 128, 128, 18, 106799), ('batchnorm2d', 77, 128, 128, 18, 154), ('ReLU', 77, 128, 128, 18, 0), ('conv2d', 116, 128, 128, 18, 80504), ('batchnorm2d', 116, 128, 128, 18, 232), ('ReLU', 116, 128, 12...
ReLU
198,528,205
21,046,805
18
17,879
-1
-1
-1
-1
-1
9
9
0
1
1
SUCCESSFUL
3.42
36,351.54
80.29
36,435.25
input_channels:3_num_classes:741_depth:27_arch:hourglass_base_filters:74_batch:50_input_size:166_act:leaky_relu_dropout:0.19747492441512948_dropout:True_batchnorm:False.out
27
[('conv2d', 74, 166, 166, 50, 2072), ('LeakyReLU', 74, 166, 166, 50, 0), ('dropout', 74, 166, 166, 50, 0), ('conv2d', 95, 166, 166, 50, 63365), ('LeakyReLU', 95, 166, 166, 50, 0), ('dropout', 95, 166, 166, 50, 0), ('conv2d', 123, 166, 166, 50, 105288), ('LeakyReLU', 123, 166, 166, 50, 0), ('dropout', 123, 166, 166, 50,...
LeakyReLU
1,169,505,752
115,743,545
50
37,577
21.6
0.124714
0.509
0.153
0.0055
27
0
27
1
1
OOM_CRASH
16
594,841
441.53
595,298.53
input_channels:3_num_classes:975_depth:22_arch:residual_base_filters:93_batch:32_input_size:182_act:softplus_dropout:0.28879752477960674_dropout:False_batchnorm:True.out
30
[('conv2d', 93, 182, 182, 32, 372), ('batchnorm2d', 93, 182, 182, 32, 186), ('conv2d', 93, 182, 182, 32, 2604), ('batchnorm2d', 93, 182, 182, 32, 186), ('Softplus', 93, 182, 182, 32, 0), ('conv2d', 93, 182, 182, 32, 77934), ('batchnorm2d', 93, 182, 182, 32, 186), ('Softplus', 93, 182, 182, 32, 0), ('conv2d', 107, 91, 9...
Softplus
47,445,103
7,604,174
32
10,513
-1
-1
-1
-1
-1
30
30
0
1
1
SUCCESSFUL
12.16
15,926.72
29.01
15,967.89
input_channels:1_num_classes:844_depth:7_arch:uniform_base_filters:62_batch:8_input_size:52_act:elu_dropout:0.3205147600384122_dropout:True_batchnorm:True.out
7
[('conv2d', 62, 52, 52, 8, 620), ('batchnorm2d', 62, 52, 52, 8, 124), ('ELU', 62, 52, 52, 8, 0), ('dropout', 62, 52, 52, 8, 0), ('conv2d', 62, 52, 52, 8, 34658), ('batchnorm2d', 62, 52, 52, 8, 124), ('ELU', 62, 52, 52, 8, 0), ('dropout', 62, 52, 52, 8, 0), ('conv2d', 62, 52, 52, 8, 34658), ('batchnorm2d', 62, 52, 52, 8...
ELU
4,695,894
262,608
8
1,889
49.761905
0.485453
0.272167
0.113983
0.016567
7
7
7
1
1
SUCCESSFUL
0.08
358.24
1
359.32
input_channels:3_num_classes:791_depth:4_arch:dense_base_filters:111_batch:54_input_size:162_act:tanh_dropout:0.3646925180398911_dropout:True_batchnorm:True.out
4
[('conv2d', 32, 162, 162, 54, 896), ('batchnorm2d', 32, 162, 162, 54, 64), ('Tanh', 32, 162, 162, 54, 0), ('dropout', 32, 162, 162, 54, 0), ('conv2d', 32, 162, 162, 54, 10112), ('batchnorm2d', 32, 162, 162, 54, 64), ('Tanh', 32, 162, 162, 54, 0), ('dropout', 32, 162, 162, 54, 0), ('conv2d', 32, 162, 162, 54, 19328), ('...
Tanh
13,438,641
163,548
54
7,915
-1
-1
-1
-1
-1
4
4
4
1
1
SUCCESSFUL
16.2
8,337.06
0.62
8,353.88
input_channels:3_num_classes:858_depth:2_arch:gradual_base_filters:50_batch:46_input_size:46_act:swish_dropout:0.22234240526292215_dropout:False_batchnorm:False.out
2
[('conv2d', 50, 46, 46, 46, 1400), ('SiLU', 50, 46, 46, 46, 0), ('conv2d', 126, 46, 46, 46, 56826), ('SiLU', 126, 46, 46, 46, 0), ('adaptive_avg_pool2d', 0, 0, 126, 46, 0), ('linear', 0, 0, 858, 46, 108966), ('softmax', 0, 0, 858, 46, 0)]
SiLU
746,674
167,192
46
1,983
36.707692
0.359422
0.305364
0.168055
0.021018
2
0
0
1
1
SUCCESSFUL
0.92
392.84
0.64
394.4
input_channels:1_num_classes:219_depth:4_arch:residual_base_filters:74_batch:56_input_size:54_act:leaky_relu_dropout:0.12777064848040343_dropout:False_batchnorm:True.out
6
[('conv2d', 74, 54, 54, 56, 148), ('batchnorm2d', 74, 54, 54, 56, 148), ('conv2d', 74, 54, 54, 56, 740), ('batchnorm2d', 74, 54, 54, 56, 148), ('LeakyReLU', 74, 54, 54, 56, 0), ('conv2d', 74, 54, 54, 56, 49358), ('batchnorm2d', 74, 54, 54, 56, 148), ('LeakyReLU', 74, 54, 54, 56, 0), ('conv2d', 169, 27, 27, 56, 12675), ...
LeakyReLU
2,712,487
471,550
56
2,803
75.242424
0.661738
0.586678
0.270237
0.026017
6
6
0
1
1
SUCCESSFUL
0.56
1,593.2
1.8
1,595.56
input_channels:1_num_classes:296_depth:8_arch:hourglass_base_filters:90_batch:26_input_size:140_act:elu_dropout:0.2904709005233884_dropout:True_batchnorm:False.out
8
[('conv2d', 90, 140, 140, 26, 900), ('ELU', 90, 140, 140, 26, 0), ('dropout', 90, 140, 140, 26, 0), ('conv2d', 196, 140, 140, 26, 158956), ('ELU', 196, 140, 140, 26, 0), ('dropout', 196, 140, 140, 26, 0), ('conv2d', 429, 140, 140, 26, 757185), ('ELU', 429, 140, 140, 26, 0), ('dropout', 429, 140, 140, 26, 0), ('conv2d',...
ELU
194,275,882
16,999,317
26
22,767
92.676923
0.946453
0.970164
0.247527
0.007836
8
0
8
1
1
SUCCESSFUL
1.82
51,383.02
64.85
51,449.69
input_channels:3_num_classes:278_depth:19_arch:hourglass_base_filters:52_batch:44_input_size:18_act:relu_dropout:0.4056898047606978_dropout:True_batchnorm:True.out
19
[('conv2d', 52, 18, 18, 44, 1456), ('batchnorm2d', 52, 18, 18, 44, 104), ('ReLU', 52, 18, 18, 44, 0), ('dropout', 52, 18, 18, 44, 0), ('conv2d', 78, 18, 18, 44, 36582), ('batchnorm2d', 78, 18, 18, 44, 156), ('ReLU', 78, 18, 18, 44, 0), ('dropout', 78, 18, 18, 44, 0), ('conv2d', 117, 18, 18, 44, 82251), ('batchnorm2d', ...
ReLU
10,951,808
61,284,148
44
4,629
-1
-1
-1
-1
-1
19
19
19
1
1
SUCCESSFUL
0
4,595.36
233.78
4,829.14
input_channels:3_num_classes:128_depth:25_arch:gradual_base_filters:51_batch:60_input_size:108_act:leaky_relu_dropout:0.21947271102023425_dropout:False_batchnorm:False.out
25
[('conv2d', 51, 108, 108, 60, 1428), ('LeakyReLU', 51, 108, 108, 60, 0), ('conv2d', 54, 108, 108, 60, 24840), ('LeakyReLU', 54, 108, 108, 60, 0), ('conv2d', 59, 108, 108, 60, 28733), ('LeakyReLU', 59, 108, 108, 60, 0), ('conv2d', 63, 108, 108, 60, 33516), ('LeakyReLU', 63, 108, 108, 60, 0), ('conv2d', 68, 108, 108, 60,...
LeakyReLU
82,535,020
5,336,091
60
32,503
-1
-1
-1
-1
-1
25
0
0
1
1
SUCCESSFUL
7.8
56,672.4
20.36
56,700.56
input_channels:1_num_classes:501_depth:15_arch:pyramid_base_filters:16_batch:52_input_size:16_act:elu_dropout:0.21775297764015017_dropout:False_batchnorm:True.out
15
[('conv2d', 16, 16, 16, 52, 160), ('batchnorm2d', 16, 16, 16, 52, 32), ('ELU', 16, 16, 16, 52, 0), ('conv2d', 22, 16, 16, 52, 3190), ('batchnorm2d', 22, 16, 16, 52, 44), ('ELU', 22, 16, 16, 52, 0), ('conv2d', 30, 16, 16, 52, 5970), ('batchnorm2d', 30, 16, 16, 52, 60), ('ELU', 30, 16, 16, 52, 0), ('conv2d', 42, 16, 16, ...
ELU
4,084,404
30,756,225
52
3,623
-1
-1
-1
-1
-1
15
15
0
1
1
SUCCESSFUL
0
2,160.08
117.33
2,277.41
input_channels:1_num_classes:93_depth:11_arch:hourglass_base_filters:90_batch:44_input_size:68_act:mish_dropout:0.372485838650226_dropout:True_batchnorm:True.out
11
[('conv2d', 90, 68, 68, 44, 900), ('batchnorm2d', 90, 68, 68, 44, 180), ('Mish', 90, 68, 68, 44, 0), ('dropout', 90, 68, 68, 44, 0), ('conv2d', 168, 68, 68, 44, 136248), ('batchnorm2d', 168, 68, 68, 44, 336), ('Mish', 168, 68, 68, 44, 0), ('dropout', 168, 68, 68, 44, 0), ('conv2d', 314, 68, 68, 44, 475082), ('batchnorm...
Mish
94,644,308
32,209,056
44
22,021
-1
-1
-1
-1
-1
11
11
11
1
1
SUCCESSFUL
0.88
39,714.4
122.87
39,838.15
input_channels:3_num_classes:378_depth:16_arch:hourglass_base_filters:16_batch:52_input_size:88_act:tanh_dropout:0.3016367621726166_dropout:False_batchnorm:True.out
16
[('conv2d', 16, 88, 88, 52, 448), ('batchnorm2d', 16, 88, 88, 52, 32), ('Tanh', 16, 88, 88, 52, 0), ('conv2d', 29, 88, 88, 52, 4205), ('batchnorm2d', 29, 88, 88, 52, 58), ('Tanh', 29, 88, 88, 52, 0), ('conv2d', 53, 88, 88, 52, 13886), ('batchnorm2d', 53, 88, 88, 52, 106), ('Tanh', 53, 88, 88, 52, 0), ('conv2d', 98, 88,...
Tanh
113,001,220
28,593,714
52
29,639
-1
-1
-1
-1
-1
16
16
0
1
1
SUCCESSFUL
4.68
59,774.52
109.08
59,888.28
input_channels:3_num_classes:782_depth:8_arch:pyramid_base_filters:47_batch:48_input_size:168_act:elu_dropout:0.12781842986085043_dropout:False_batchnorm:False.out
8
[('conv2d', 47, 168, 168, 48, 1316), ('ELU', 47, 168, 168, 48, 0), ('conv2d', 75, 168, 168, 48, 31800), ('ELU', 75, 168, 168, 48, 0), ('conv2d', 120, 168, 168, 48, 81120), ('ELU', 120, 168, 168, 48, 0), ('conv2d', 193, 168, 168, 48, 208633), ('ELU', 193, 168, 168, 48, 0), ('conv2d', 310, 168, 168, 48, 538780), ('ELU', ...
ELU
187,184,409
15,975,148
48
35,789
31.833333
0.144
0
0
0.228
8
0
0
1
1
OOM_CRASH
15.36
102,822.72
60.94
102,899.02
input_channels:1_num_classes:673_depth:14_arch:residual_base_filters:55_batch:56_input_size:96_act:relu_dropout:0.1329243036031135_dropout:False_batchnorm:True.out
21
[('conv2d', 55, 96, 96, 56, 110), ('batchnorm2d', 55, 96, 96, 56, 110), ('conv2d', 55, 96, 96, 56, 550), ('batchnorm2d', 55, 96, 96, 56, 110), ('ReLU', 55, 96, 96, 56, 0), ('conv2d', 55, 96, 96, 56, 27280), ('batchnorm2d', 55, 96, 96, 56, 110), ('ReLU', 55, 96, 96, 56, 0), ('conv2d', 71, 48, 48, 56, 3976), ('batchnorm2...
ReLU
8,847,813
2,860,582
56
4,429
-1
-1
-1
-1
-1
21
21
0
1
1
SUCCESSFUL
2.24
5,197.36
10.91
5,210.51
input_channels:3_num_classes:325_depth:2_arch:residual_base_filters:102_batch:52_input_size:66_act:relu_dropout:0.39141141159673887_dropout:False_batchnorm:True.out
3
[('conv2d', 102, 66, 66, 52, 408), ('batchnorm2d', 102, 66, 66, 52, 204), ('conv2d', 102, 66, 66, 52, 2856), ('batchnorm2d', 102, 66, 66, 52, 204), ('ReLU', 102, 66, 66, 52, 0), ('conv2d', 102, 66, 66, 52, 93738), ('batchnorm2d', 102, 66, 66, 52, 204), ('ReLU', 102, 66, 66, 52, 0), ('adaptive_avg_pool2d', 0, 0, 102, 52...
ReLU
3,555,248
131,089
52
2,763
64.634921
0.6488
0.649228
0.312579
0.035322
3
3
0
1
1
SUCCESSFUL
2.6
1,939.08
0.5
1,942.18
input_channels:3_num_classes:347_depth:17_arch:hourglass_base_filters:26_batch:24_input_size:138_act:relu_dropout:0.283622497424984_dropout:False_batchnorm:True.out
17
[('conv2d', 26, 138, 138, 24, 728), ('batchnorm2d', 26, 138, 138, 24, 52), ('ReLU', 26, 138, 138, 24, 0), ('conv2d', 44, 138, 138, 24, 10340), ('batchnorm2d', 44, 138, 138, 24, 88), ('ReLU', 44, 138, 138, 24, 0), ('conv2d', 77, 138, 138, 24, 30569), ('batchnorm2d', 77, 138, 138, 24, 154), ('ReLU', 77, 138, 138, 24, 0),...
ReLU
343,706,832
38,659,917
24
36,921
94.242424
0.938246
0.955673
0.221255
0.007055
17
17
0
1
1
SUCCESSFUL
5.28
83,912.64
147.48
84,065.4
input_channels:3_num_classes:342_depth:21_arch:uniform_base_filters:72_batch:54_input_size:44_act:tanh_dropout:0.3590667838150622_dropout:True_batchnorm:True.out
21
[('conv2d', 72, 44, 44, 54, 2016), ('batchnorm2d', 72, 44, 44, 54, 144), ('Tanh', 72, 44, 44, 54, 0), ('dropout', 72, 44, 44, 54, 0), ('conv2d', 72, 44, 44, 54, 46728), ('batchnorm2d', 72, 44, 44, 54, 144), ('Tanh', 72, 44, 44, 54, 0), ('dropout', 72, 44, 44, 54, 0), ('conv2d', 72, 44, 44, 54, 46728), ('batchnorm2d', 7...
Tanh
11,709,684
964,566
54
3,969
-1
-1
-1
-1
-1
21
21
21
1
1
SUCCESSFUL
1.08
6,030.18
3.68
6,034.94
input_channels:1_num_classes:154_depth:2_arch:dense_base_filters:86_batch:60_input_size:184_act:tanh_dropout:0.15712160010655096_dropout:False_batchnorm:True.out
0
[]
null
0
0
60
1,471
6.758065
0.017571
0.001327
0
0
0
0
0
0
0
SUCCESSFUL
0
0
0
0
input_channels:3_num_classes:966_depth:20_arch:bottleneck_base_filters:37_batch:50_input_size:206_act:prelu_dropout:0.30530435953353746_dropout:False_batchnorm:False.out
20
[('conv2d', 74, 206, 206, 50, 2072), ('PReLU', 74, 206, 206, 50, 0), ('conv2d', 37, 206, 206, 50, 24679), ('PReLU', 37, 206, 206, 50, 0), ('conv2d', 45, 206, 206, 50, 15030), ('PReLU', 45, 206, 206, 50, 0), ('conv2d', 56, 206, 206, 50, 22736), ('PReLU', 56, 206, 206, 50, 0), ('conv2d', 69, 206, 206, 50, 34845), ('PReLU...
PReLU
731,090,998
59,664,691
50
38,045
-1
-1
-1
-1
-1
20
0
0
1
1
OOM_CRASH
24.5
418,333
227.6
418,585.1
input_channels:1_num_classes:840_depth:6_arch:pyramid_base_filters:74_batch:22_input_size:36_act:selu_dropout:0.3929660404241916_dropout:False_batchnorm:False.out
6
[('conv2d', 74, 36, 36, 22, 740), ('SELU', 74, 36, 36, 22, 0), ('conv2d', 128, 36, 36, 22, 85376), ('SELU', 128, 36, 36, 22, 0), ('conv2d', 223, 36, 36, 22, 257119), ('SELU', 223, 36, 36, 22, 0), ('conv2d', 389, 36, 36, 22, 781112), ('SELU', 389, 36, 36, 22, 0), ('conv2d', 677, 36, 36, 22, 2370854), ('SELU', 677, 36, 3...
SELU
6,918,313
11,657,359
22
3,041
90.453125
0.890328
0.851509
0.254509
0.012
6
0
0
1
1
SUCCESSFUL
0
1,741.52
44.47
1,785.99
input_channels:1_num_classes:542_depth:11_arch:pyramid_base_filters:40_batch:26_input_size:82_act:swish_dropout:0.24009389316519958_dropout:True_batchnorm:False.out
11
[('conv2d', 40, 82, 82, 26, 400), ('SiLU', 40, 82, 82, 26, 0), ('dropout', 40, 82, 82, 26, 0), ('conv2d', 57, 82, 82, 26, 20577), ('SiLU', 57, 82, 82, 26, 0), ('dropout', 57, 82, 82, 26, 0), ('conv2d', 81, 82, 82, 26, 41634), ('SiLU', 81, 82, 82, 26, 0), ('dropout', 81, 82, 82, 26, 0), ('conv2d', 117, 82, 82, 26, 85410...
SiLU
94,084,723
25,983,265
26
12,841
93.96875
0.947222
0.970673
0.240782
0.011119
11
0
11
1
1
SUCCESSFUL
0.78
24,883.82
99.12
24,983.72
input_channels:1_num_classes:524_depth:25_arch:dense_base_filters:126_batch:60_input_size:122_act:gelu_dropout:0.24097420643746445_dropout:True_batchnorm:True.out
25
[('conv2d', 32, 122, 122, 60, 320), ('batchnorm2d', 32, 122, 122, 60, 64), ('GELU', 32, 122, 122, 60, 0), ('dropout', 32, 122, 122, 60, 0), ('conv2d', 32, 122, 122, 60, 9536), ('batchnorm2d', 32, 122, 122, 60, 64), ('GELU', 32, 122, 122, 60, 0), ('dropout', 32, 122, 122, 60, 0), ('conv2d', 32, 122, 122, 60, 18752), ('b...
GELU
20,005,848
2,556,300
60
16,055
-1
-1
-1
-1
-1
25
25
24
1
1
SUCCESSFUL
3.6
16,468.8
9.75
16,482.15
input_channels:3_num_classes:220_depth:10_arch:gradual_base_filters:86_batch:54_input_size:22_act:elu_dropout:0.37419768011604204_dropout:True_batchnorm:False.out
10
[('conv2d', 86, 22, 22, 54, 2408), ('ELU', 86, 22, 22, 54, 0), ('dropout', 86, 22, 22, 54, 0), ('conv2d', 100, 22, 22, 54, 77500), ('ELU', 100, 22, 22, 54, 0), ('dropout', 100, 22, 22, 54, 0), ('conv2d', 118, 22, 22, 54, 106318), ('ELU', 118, 22, 22, 54, 0), ('dropout', 118, 22, 22, 54, 0), ('conv2d', 138, 22, 22, 54, ...
ELU
2,814,774
3,483,674
54
2,463
-1
-1
-1
-1
-1
10
0
10
1
1
SUCCESSFUL
0.54
1,546.02
13.29
1,559.85
input_channels:3_num_classes:208_depth:26_arch:hourglass_base_filters:81_batch:62_input_size:160_act:swish_dropout:0.2875263141183326_dropout:True_batchnorm:False.out
26
[('conv2d', 81, 160, 160, 62, 2268), ('SiLU', 81, 160, 160, 62, 0), ('dropout', 81, 160, 160, 62, 0), ('conv2d', 103, 160, 160, 62, 75190), ('SiLU', 103, 160, 160, 62, 0), ('dropout', 103, 160, 160, 62, 0), ('conv2d', 133, 160, 160, 62, 123424), ('SiLU', 133, 160, 160, 62, 0), ('dropout', 133, 160, 160, 62, 0), ('conv2...
SiLU
1,070,131,697
114,109,662
62
39,875
-1
-1
-1
-1
-1
26
0
26
1
1
OOM_CRASH
17.98
674,928.28
435.29
675,381.55
input_channels:1_num_classes:190_depth:21_arch:bottleneck_base_filters:110_batch:18_input_size:198_act:leaky_relu_dropout:0.4956058781321332_dropout:False_batchnorm:True.out
21
[('conv2d', 220, 198, 198, 18, 2200), ('batchnorm2d', 220, 198, 198, 18, 440), ('LeakyReLU', 220, 198, 198, 18, 0), ('conv2d', 110, 198, 198, 18, 217910), ('batchnorm2d', 110, 198, 198, 18, 220), ('LeakyReLU', 110, 198, 198, 18, 0), ('conv2d', 127, 198, 198, 18, 125857), ('batchnorm2d', 127, 198, 198, 18, 254), ('Leaky...
LeakyReLU
1,472,504,389
96,120,522
18
38,837
29.25
0.1158
0
0
0
21
21
0
1
1
OOM_CRASH
2.7
269,623.44
366.67
269,992.81
input_channels:3_num_classes:440_depth:2_arch:gradual_base_filters:70_batch:2_input_size:198_act:leaky_relu_dropout:0.2828152190994582_dropout:False_batchnorm:False.out
2
[('conv2d', 70, 198, 198, 2, 1960), ('LeakyReLU', 70, 198, 198, 2, 0), ('conv2d', 162, 198, 198, 2, 102222), ('LeakyReLU', 162, 198, 198, 2, 0), ('adaptive_avg_pool2d', 0, 0, 162, 2, 0), ('linear', 0, 0, 440, 2, 71720), ('softmax', 0, 0, 440, 2, 0)]
LeakyReLU
18,191,698
175,902
2
2,075
80.492308
0.783746
0.683017
0.314153
0.027672
2
0
0
1
1
SUCCESSFUL
0.9
416.36
0.67
417.93
input_channels:1_num_classes:696_depth:21_arch:residual_base_filters:46_batch:60_input_size:176_act:softplus_dropout:0.44759412960394884_dropout:True_batchnorm:False.out
30
[('conv2d', 46, 176, 176, 60, 92), ('conv2d', 46, 176, 176, 60, 460), ('Softplus', 46, 176, 176, 60, 0), ('dropout', 46, 176, 176, 60, 0), ('conv2d', 46, 176, 176, 60, 19090), ('Softplus', 46, 176, 176, 60, 0), ('dropout', 46, 176, 176, 60, 0), ('conv2d', 55, 88, 88, 60, 2585), ('conv2d', 55, 88, 88, 60, 22825), ('Soft...
Softplus
20,085,368
3,187,994
60
7,393
-1
-1
-1
-1
-1
30
0
20
1
1
SUCCESSFUL
7.2
14,448
12.16
14,467.36
input_channels:1_num_classes:117_depth:6_arch:uniform_base_filters:90_batch:42_input_size:76_act:softplus_dropout:0.40557608456188843_dropout:False_batchnorm:False.out
6
[('conv2d', 90, 76, 76, 42, 900), ('Softplus', 90, 76, 76, 42, 0), ('conv2d', 90, 76, 76, 42, 72990), ('Softplus', 90, 76, 76, 42, 0), ('conv2d', 90, 76, 76, 42, 72990), ('Softplus', 90, 76, 76, 42, 0), ('conv2d', 90, 76, 76, 42, 72990), ('Softplus', 90, 76, 76, 42, 0), ('conv2d', 90, 76, 76, 42, 72990), ('Softplus', 9...
Softplus
6,238,404
376,497
42
3,209
-1
-1
-1
-1
-1
6
0
0
1
1
SUCCESSFUL
0.84
2,998.38
1.44
3,000.66
input_channels:3_num_classes:62_depth:27_arch:hourglass_base_filters:110_batch:30_input_size:152_act:gelu_dropout:0.19672866732903366_dropout:True_batchnorm:True.out
27
[('conv2d', 110, 152, 152, 30, 3080), ('batchnorm2d', 110, 152, 152, 30, 220), ('GELU', 110, 152, 152, 30, 0), ('dropout', 110, 152, 152, 30, 0), ('conv2d', 137, 152, 152, 30, 135767), ('batchnorm2d', 137, 152, 152, 30, 274), ('GELU', 137, 152, 152, 30, 0), ('dropout', 137, 152, 152, 30, 0), ('conv2d', 172, 152, 152, 3...
GELU
1,479,950,058
136,011,597
30
37,861
-1
-1
-1
-1
-1
27
27
27
1
1
OOM_CRASH
7.8
423,417
518.84
423,943.64
input_channels:1_num_classes:90_depth:26_arch:dense_base_filters:61_batch:38_input_size:20_act:prelu_dropout:0.1095213924752788_dropout:True_batchnorm:True.out
25
[('conv2d', 32, 20, 20, 38, 320), ('batchnorm2d', 32, 20, 20, 38, 64), ('PReLU', 32, 20, 20, 38, 0), ('dropout', 32, 20, 20, 38, 0), ('conv2d', 32, 20, 20, 38, 9536), ('batchnorm2d', 32, 20, 20, 38, 64), ('PReLU', 32, 20, 20, 38, 0), ('dropout', 32, 20, 20, 38, 0), ('conv2d', 32, 20, 20, 38, 18752), ('batchnorm2d', 32,...
PReLU
538,484
2,250,330
38
2,043
39.369231
0.396154
0.201357
0.067309
0.006286
25
25
24
1
1
SUCCESSFUL
0
280.44
8.58
289.02
input_channels:3_num_classes:919_depth:22_arch:pyramid_base_filters:28_batch:34_input_size:56_act:elu_dropout:0.4834553597708333_dropout:True_batchnorm:True.out
22
[('conv2d', 28, 56, 56, 34, 784), ('batchnorm2d', 28, 56, 56, 34, 56), ('ELU', 28, 56, 56, 34, 0), ('dropout', 28, 56, 56, 34, 0), ('conv2d', 34, 56, 56, 34, 8602), ('batchnorm2d', 34, 56, 56, 34, 68), ('ELU', 34, 56, 56, 34, 0), ('dropout', 34, 56, 56, 34, 0), ('conv2d', 41, 56, 56, 34, 12587), ('batchnorm2d', 41, 56,...
ELU
117,515,714
66,560,318
34
20,121
95.571429
0.925781
0.959948
0.231207
0.014692
22
22
22
1
1
SUCCESSFUL
1.36
38,104.14
253.91
38,359.41
input_channels:1_num_classes:421_depth:22_arch:gradual_base_filters:71_batch:12_input_size:28_act:elu_dropout:0.4972116800368136_dropout:False_batchnorm:False.out
22
[('conv2d', 71, 28, 28, 12, 710), ('ELU', 71, 28, 28, 12, 0), ('conv2d', 76, 28, 28, 12, 48640), ('ELU', 76, 28, 28, 12, 0), ('conv2d', 82, 28, 28, 12, 56170), ('ELU', 82, 28, 28, 12, 0), ('conv2d', 89, 28, 28, 12, 65771), ('ELU', 89, 28, 28, 12, 0), ('conv2d', 96, 28, 28, 12, 76992), ('ELU', 96, 28, 28, 12, 0), ('conv...
ELU
6,114,827
7,171,304
12
2,179
85.793651
0.854952
0.63781
0.195638
0.023667
22
0
0
1
1
SUCCESSFUL
0
839.64
27.36
867
input_channels:3_num_classes:791_depth:10_arch:hourglass_base_filters:66_batch:44_input_size:96_act:elu_dropout:0.26834250170579155_dropout:True_batchnorm:True.out
10
[('conv2d', 66, 96, 96, 44, 1848), ('batchnorm2d', 66, 96, 96, 44, 132), ('ELU', 66, 96, 96, 44, 0), ('dropout', 66, 96, 96, 44, 0), ('conv2d', 131, 96, 96, 44, 77945), ('batchnorm2d', 131, 96, 96, 44, 262), ('ELU', 131, 96, 96, 44, 0), ('dropout', 131, 96, 96, 44, 0), ('conv2d', 260, 96, 96, 44, 306800), ('batchnorm2d...
ELU
147,826,288
22,411,577
44
30,361
95.390625
0.934662
0.962821
0.261875
0.017737
10
10
10
1
1
SUCCESSFUL
4.84
62,030.32
85.49
62,120.65
input_channels:1_num_classes:204_depth:8_arch:gradual_base_filters:61_batch:6_input_size:148_act:softplus_dropout:0.37164570554992904_dropout:False_batchnorm:False.out
8
[('conv2d', 61, 148, 148, 6, 610), ('Softplus', 61, 148, 148, 6, 0), ('conv2d', 75, 148, 148, 6, 41250), ('Softplus', 75, 148, 148, 6, 0), ('conv2d', 94, 148, 148, 6, 63544), ('Softplus', 94, 148, 148, 6, 0), ('conv2d', 117, 148, 148, 6, 99099), ('Softplus', 117, 148, 148, 6, 0), ('conv2d', 146, 148, 148, 6, 153884), (...
Softplus
51,913,171
1,606,158
6
3,911
-1
-1
-1
-1
-1
8
0
0
1
1
SUCCESSFUL
0.48
3,564.6
6.13
3,571.21
input_channels:1_num_classes:313_depth:12_arch:hourglass_base_filters:69_batch:36_input_size:202_act:elu_dropout:0.4929905710845056_dropout:False_batchnorm:True.out
12
[('conv2d', 69, 202, 202, 36, 690), ('batchnorm2d', 69, 202, 202, 36, 138), ('ELU', 69, 202, 202, 36, 0), ('conv2d', 121, 202, 202, 36, 75262), ('batchnorm2d', 121, 202, 202, 36, 242), ('ELU', 121, 202, 202, 36, 0), ('conv2d', 213, 202, 202, 36, 232170), ('batchnorm2d', 213, 202, 202, 36, 426), ('ELU', 213, 202, 202, 3...
ELU
637,032,743
32,562,334
36
37,867
40
0.142333
0.068
0.036
0.0395
12
12
0
1
1
OOM_CRASH
5.76
233,288.28
124.22
233,418.26
input_channels:3_num_classes:476_depth:1_arch:hourglass_base_filters:94_batch:20_input_size:118_act:prelu_dropout:0.4976764538806615_dropout:False_batchnorm:False.out
1
[('conv2d', 94, 118, 118, 20, 2632), ('PReLU', 94, 118, 118, 20, 0), ('adaptive_avg_pool2d', 0, 0, 94, 20, 0), ('linear', 0, 0, 476, 20, 45220), ('softmax', 0, 0, 476, 20, 0)]
PReLU
2,618,758
47,852
20
2,201
36.47619
0.322692
0.295804
0.216571
0.044649
1
0
0
1
1
SUCCESSFUL
3.2
599.4
0.18
602.78
input_channels:1_num_classes:829_depth:27_arch:dense_base_filters:71_batch:38_input_size:110_act:swish_dropout:0.4879000239896045_dropout:False_batchnorm:True.out
25
[('conv2d', 32, 110, 110, 38, 320), ('batchnorm2d', 32, 110, 110, 38, 64), ('SiLU', 32, 110, 110, 38, 0), ('conv2d', 32, 110, 110, 38, 9536), ('batchnorm2d', 32, 110, 110, 38, 64), ('SiLU', 32, 110, 110, 38, 0), ('conv2d', 32, 110, 110, 38, 18752), ('batchnorm2d', 32, 110, 110, 38, 64), ('SiLU', 32, 110, 110, 38, 0), (...
SiLU
12,779,962
2,771,325
38
9,029
92.671875
0.915587
0.813222
0.412667
0.114232
25
25
0
1
1
SUCCESSFUL
1.9
7,469.28
10.57
7,481.75
input_channels:3_num_classes:186_depth:9_arch:uniform_base_filters:35_batch:2_input_size:178_act:relu_dropout:0.229215997093005_dropout:False_batchnorm:False.out
9
[('conv2d', 35, 178, 178, 2, 980), ('ReLU', 35, 178, 178, 2, 0), ('conv2d', 35, 178, 178, 2, 11060), ('ReLU', 35, 178, 178, 2, 0), ('conv2d', 35, 178, 178, 2, 11060), ('ReLU', 35, 178, 178, 2, 0), ('conv2d', 35, 178, 178, 2, 11060), ('ReLU', 35, 178, 178, 2, 0), ('conv2d', 35, 178, 178, 2, 11060), ('ReLU', 35, 178, 178...
ReLU
19,961,327
96,156
2
1,843
-1
-1
-1
-1
-1
9
0
0
1
1
SUCCESSFUL
0.72
456.88
0.37
457.97
input_channels:1_num_classes:464_depth:13_arch:dense_base_filters:122_batch:42_input_size:132_act:selu_dropout:0.1632770865763225_dropout:True_batchnorm:False.out
13
[('conv2d', 32, 132, 132, 42, 320), ('SELU', 32, 132, 132, 42, 0), ('dropout', 32, 132, 132, 42, 0), ('conv2d', 32, 132, 132, 42, 9536), ('SELU', 32, 132, 132, 42, 0), ('dropout', 32, 132, 132, 42, 0), ('conv2d', 32, 132, 132, 42, 18752), ('SELU', 32, 132, 132, 42, 0), ('dropout', 32, 132, 132, 42, 0), ('conv2d', 32, 1...
SELU
12,267,744
619,600
42
6,683
-1
-1
-1
-1
-1
13
0
12
1
1
SUCCESSFUL
2.94
6,973.68
2.36
6,978.98
input_channels:1_num_classes:306_depth:22_arch:gradual_base_filters:79_batch:30_input_size:50_act:mish_dropout:0.25702048201729766_dropout:False_batchnorm:False.out
22
[('conv2d', 79, 50, 50, 30, 790), ('Mish', 79, 50, 50, 30, 0), ('conv2d', 85, 50, 50, 30, 60520), ('Mish', 85, 50, 50, 30, 0), ('conv2d', 91, 50, 50, 30, 69706), ('Mish', 91, 50, 50, 30, 0), ('conv2d', 98, 50, 50, 30, 80360), ('Mish', 98, 50, 50, 30, 0), ('conv2d', 106, 50, 50, 30, 93598), ('Mish', 106, 50, 50, 30, 0),...
Mish
20,995,985
8,183,498
30
5,791
93.109375
0.918127
0.858537
0.322593
0.037473
22
0
0
1
1
SUCCESSFUL
0.3
7,208.4
31.22
7,239.92
input_channels:3_num_classes:864_depth:28_arch:pyramid_base_filters:30_batch:44_input_size:108_act:prelu_dropout:0.27757374025542_dropout:True_batchnorm:True.out
28
[('conv2d', 30, 108, 108, 44, 840), ('batchnorm2d', 30, 108, 108, 44, 60), ('PReLU', 30, 108, 108, 44, 0), ('dropout', 30, 108, 108, 44, 0), ('conv2d', 34, 108, 108, 44, 9214), ('batchnorm2d', 34, 108, 108, 44, 68), ('PReLU', 34, 108, 108, 44, 0), ('dropout', 34, 108, 108, 44, 0), ('conv2d', 40, 108, 108, 44, 12280), (...
PReLU
577,698,081
93,628,218
44
40,145
43.5
0.1112
0.371
0.096
0.006
28
28
28
1
1
OOM_CRASH
5.72
242,411.4
357.16
242,774.28
End of preview. Expand in Data Studio

GPUMemNet and GPUUtilNet Dataset

This dataset accompanies the paper “GPU Memory and Utilization Estimation for Training-Aware Resource Management: Opportunities and Limitations.”

It contains synthetic deep learning training configurations and their measured GPU memory consumption and utilization characteristics.

Dataset configurations

The dataset is divided into separate configurations because MLP, CNN, and Transformer workloads use different feature schemas.

Configuration Rows Description
mlp-memory-step1 3,000 Initial MLP memory and average-utilization measurements
mlp-memory-step2 3,000 MLP measurements with batch-normalization and dropout features
mlp-utilization 3,000 MLP average and maximum utilization measurements
mlp-memory-legacy 1,091 Earlier fully connected network memory dataset
cnn-memory-step1 9,000 CNN measurements including architecture identifiers
cnn-memory-revised 9,000 Revised CNN representation
cnn-utilization 9,000 CNN average and maximum utilization measurements
transformer-memory 5,011 Transformer memory measurements
transformer-utilization 5,011 Transformer average and maximum utilization measurements

Prediction targets

The primary GPU-memory prediction target is:

  • Max GPU Memory (MiB)

The legacy MLP configuration uses:

  • gpumemory_max

The utilization configurations contain average and maximum values for:

  • GPUTL
  • GRACT
  • SMACT
  • SMOCC
  • FP32A
  • DRAMA

Loading

Install the Hugging Face datasets package:

pip install datasets

Load a specific configuration:

from datasets import load_dataset

dataset = load_dataset(
    "ehyo/GPU-Resources-Estimation-for-Deep-Learning-Training-Tasks",
    "cnn-utilization",
)

Each configuration currently provides a train split containing the complete corresponding table.

Dataset characteristics

The workload families contain different features:

  • MLP: depth, activation function, activation counts, parameter counts, batch size, batch-normalization layers, and dropout layers.
  • CNN: depth, activation function, layer-type counts, batch size, parameter counts, activation counts, and analytical memory estimates.
  • Transformer: sequence length, embedding size, number of layers, number of attention heads, parameter counts, activation counts, and layer-type counts.

The configurations should be loaded independently because their schemas are workload-family specific.

Data collection

The measurements were collected from generated deep learning training workloads under controlled execution conditions. The original column names and units are preserved for compatibility with the accompanying code and published experiments.

Further implementation and experimental details are available in the paper and GitHub repository.

Repository

Code, models, and reproducibility artifacts are available at:

https://github.com/itu-rad/GPUMemNet

Citation

@inproceedings{yousefzadehaslmiandoab2026gpumemory,
  author    = {Ehsan Yousefzadeh-Asl-Miandoab and
               Reza Karimzadeh and
               Danyal Yorulmaz and
               Bulat Ibragimov and
               Pınar Tözün},
  title     = {GPU Memory and Utilization Estimation for Training-Aware
               Resource Management: Opportunities and Limitations},
  booktitle = {Proceedings of the Sixth European Workshop on Machine
               Learning and Systems},
  series    = {EuroMLSys '26},
  pages     = {127--138},
  publisher = {Association for Computing Machinery},
  year      = {2026},
  doi       = {10.1145/3805621.3807621}
}

License

This dataset is licensed under the Creative Commons Attribution 4.0 International License.

Downloads last month
73