All Datasets | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 24.10 | 0.37 | 0.32 |
learning_rate | 6.20 | 9.96 | 8.38 |
loss | 0.38 | 0.16 | 0.11 |
batch_size | 3.06 | 39.60 | 36.17 |
Dense | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 18.10 | 0.40 | 0.27 |
learning_rate | 6.53 | 12.81 | 10.60 |
loss | 0.47 | 0.21 | 0.13 |
batch_size | 3.53 | 43.31 | 40.65 |
n_dense_nodes | 2.12 | 10.19 | 0.89 |
dense_activation | 3.33 | 0.26 | 0.97 |
dense_dropout_rate | 3.27 | 0.30 | 0.44 |
n_dense_layers | 8.54 | 8.77 | 27.08 |
Image | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 35.59 | 0.31 | 0.34 |
learning_rate | 6.15 | 8.26 | 7.68 |
loss | 0.19 | 0.07 | 0.05 |
batch_size | 2.30 | 41.44 | 35.51 |
Tabular | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 8.79 | 0.45 | 0.30 |
learning_rate | 6.27 | 12.22 | 9.31 |
loss | 0.64 | 0.28 | 0.20 |
batch_size | 4.08 | 37.13 | 37.05 |
n_dense_nodes | 2.15 | 11.23 | 0.82 |
dense_activation | 3.98 | 0.29 | 1.33 |
dense_dropout_rate | 4.28 | 0.37 | 0.44 |
n_dense_layers | 8.65 | 11.43 | 31.82 |
Tabular Regression | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 0.74 | 0.58 | 0.44 |
learning_rate | 5.91 | 8.23 | 7.78 |
loss | 0.94 | 0.44 | 0.21 |
batch_size | 5.07 | 33.28 | 27.37 |
n_dense_nodes | 2.45 | 12.70 | 1.05 |
dense_activation | 2.87 | 0.37 | 1.67 |
dense_dropout_rate | 5.93 | 0.47 | 0.53 |
n_dense_layers | 1.33 | 14.13 | 40.06 |
Tabular Classification | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 16.85 | 0.31 | 0.15 |
learning_rate | 6.63 | 16.21 | 10.84 |
loss | 0.35 | 0.12 | 0.18 |
batch_size | 3.10 | 40.99 | 46.73 |
n_dense_nodes | 1.85 | 9.75 | 0.59 |
dense_activation | 5.09 | 0.21 | 0.99 |
dense_dropout_rate | 2.64 | 0.27 | 0.36 |
n_dense_layers | 15.96 | 8.73 | 23.58 |
Dense Image | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 32.06 | 0.32 | 0.23 |
learning_rate | 6.92 | 13.69 | 12.55 |
loss | 0.21 | 0.10 | 0.03 |
batch_size | 2.69 | 52.58 | 46.04 |
n_dense_nodes | 2.07 | 8.62 | 0.99 |
dense_activation | 2.36 | 0.21 | 0.44 |
dense_dropout_rate | 1.75 | 0.20 | 0.43 |
n_dense_layers | 8.38 | 4.77 | 19.96 |
CNN Image | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 39.11 | 0.29 | 0.45 |
learning_rate | 5.37 | 2.83 | 2.81 |
loss | 0.17 | 0.03 | 0.07 |
batch_size | 1.91 | 30.31 | 24.99 |
n_kernels | 0.98 | 15.92 | 11.78 |
kernel_size | 0.52 | 0.17 | 0.33 |
pool_size | 1.30 | 0.35 | 0.76 |
conv_activation | 0.53 | 0.20 | 0.47 |
conv_dropout_rate | 1.35 | 0.72 | 0.76 |
n_conv_layers | 3.30 | 13.54 | 18.37 |
n_dense_nodes | 2.01 | 3.98 | 1.15 |
dense_activation | 1.31 | 0.12 | 0.49 |
dense_dropout_rate | 1.41 | 0.27 | 0.54 |
n_dense_layers | 2.32 | 0.84 | 3.75 |
abalone DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 0.28 | 0.60 | 0.51 |
learning_rate | 9.90 | 2.68 | 1.74 |
loss | 1.21 | 0.42 | 0.34 |
n_dense_nodes | 0.71 | 22.42 | 0.86 |
dense_activation | 2.13 | 0.36 | 3.16 |
dense_dropout_rate | 6.61 | 0.37 | 0.57 |
n_dense_layers | 1.69 | 27.35 | 66.91 |
batch_size | 4.23 | 4.56 | 3.11 |
bike_sharing DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 1.65 | 0.36 | 0.72 |
learning_rate | 7.19 | 7.30 | 4.58 |
loss | 0.50 | 0.49 | 0.18 |
n_dense_nodes | 1.12 | 14.58 | 1.21 |
dense_activation | 3.41 | 0.12 | 1.51 |
dense_dropout_rate | 6.37 | 0.30 | 0.42 |
n_dense_layers | 1.74 | 14.78 | 53.18 |
batch_size | 5.45 | 32.70 | 14.40 |
compas DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 15.92 | 0.67 | 0.24 |
learning_rate | 7.70 | 4.85 | 1.31 |
loss | 0.50 | 0.11 | 0.16 |
n_dense_nodes | 2.96 | 23.45 | 0.72 |
dense_activation | 3.41 | 0.34 | 2.26 |
dense_dropout_rate | 3.59 | 0.58 | 0.31 |
n_dense_layers | 16.37 | 22.87 | 69.07 |
batch_size | 1.74 | 4.62 | 5.56 |
covertype DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 12.03 | 0.19 | 0.15 |
learning_rate | 4.51 | 20.96 | 16.07 |
loss | 0.29 | 0.08 | 0.25 |
n_dense_nodes | 0.83 | 3.44 | 0.31 |
dense_activation | 11.24 | 0.10 | 0.61 |
dense_dropout_rate | 1.18 | 0.11 | 0.42 |
n_dense_layers | 11.66 | 2.20 | 1.42 |
batch_size | 5.46 | 60.66 | 68.08 |
delays_zurich DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 0.28 | 0.77 | 0.09 |
learning_rate | 0.65 | 14.72 | 17.02 |
loss | 1.10 | 0.41 | 0.11 |
n_dense_nodes | 5.51 | 1.11 | 1.07 |
dense_activation | 3.06 | 0.64 | 0.34 |
dense_dropout_rate | 4.80 | 0.73 | 0.59 |
n_dense_layers | 0.57 | 0.26 | 0.10 |
batch_size | 5.52 | 62.57 | 64.60 |
higgs DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 22.59 | 0.08 | 0.06 |
learning_rate | 7.69 | 22.82 | 15.14 |
loss | 0.28 | 0.17 | 0.13 |
n_dense_nodes | 1.74 | 2.38 | 0.74 |
dense_activation | 0.63 | 0.19 | 0.08 |
dense_dropout_rate | 3.14 | 0.13 | 0.33 |
n_dense_layers | 19.86 | 1.12 | 0.27 |
batch_size | 2.10 | 57.70 | 66.55 |
cifar10 CNNModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 45.67 | 0.23 | 0.35 |
learning_rate | 4.74 | 2.21 | 1.77 |
loss | 0.20 | 0.06 | 0.14 |
n_kernels | 0.63 | 35.13 | 33.70 |
kernel_size | 0.73 | 0.39 | 0.21 |
pool_size | 1.19 | 0.35 | 0.21 |
conv_activation | 0.22 | 0.20 | 0.10 |
conv_dropout_rate | 1.01 | 0.48 | 0.47 |
n_conv_layers | 3.70 | 25.85 | 28.09 |
n_dense_nodes | 0.72 | 1.65 | 0.40 |
dense_activation | 1.26 | 0.21 | 0.40 |
dense_dropout_rate | 1.04 | 0.26 | 0.72 |
n_dense_layers | 0.69 | 0.13 | 0.67 |
batch_size | 1.56 | 1.35 | 2.95 |
cifar10 DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 38.26 | 0.19 | 0.22 |
learning_rate | 5.90 | 12.42 | 10.19 |
loss | 0.10 | 0.06 | 0.02 |
n_dense_nodes | 2.69 | 10.44 | 1.63 |
dense_activation | 3.52 | 0.34 | 0.55 |
dense_dropout_rate | 0.78 | 0.21 | 0.41 |
n_dense_layers | 2.70 | 4.97 | 24.78 |
batch_size | 1.91 | 51.12 | 40.17 |
mnist CNNModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 33.23 | 0.20 | 0.28 |
learning_rate | 5.29 | 3.54 | 3.62 |
loss | 0.20 | 0.01 | 0.04 |
n_kernels | 0.96 | 1.05 | 0.40 |
kernel_size | 0.12 | 0.02 | 0.16 |
pool_size | 1.62 | 0.51 | 1.30 |
conv_activation | 0.31 | 0.14 | 0.63 |
conv_dropout_rate | 2.66 | 0.60 | 1.06 |
n_conv_layers | 3.55 | 1.71 | 15.56 |
n_dense_nodes | 3.73 | 6.20 | 1.05 |
dense_activation | 2.24 | 0.07 | 0.93 |
dense_dropout_rate | 2.66 | 0.14 | 0.19 |
n_dense_layers | 1.75 | 1.14 | 6.08 |
batch_size | 3.03 | 56.68 | 37.87 |
mnist DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 31.73 | 0.33 | 0.10 |
learning_rate | 6.11 | 10.85 | 12.49 |
loss | 0.31 | 0.19 | 0.04 |
n_dense_nodes | 1.86 | 8.33 | 0.81 |
dense_activation | 2.75 | 0.08 | 0.36 |
dense_dropout_rate | 0.67 | 0.19 | 0.32 |
n_dense_layers | 12.73 | 5.12 | 15.94 |
batch_size | 2.77 | 54.65 | 52.95 |
fashion_mnist CNNModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 34.55 | 0.14 | 0.14 |
learning_rate | 7.99 | 3.23 | 2.20 |
loss | 0.15 | 0.01 | 0.03 |
n_kernels | 1.87 | 0.64 | 0.39 |
kernel_size | 0.57 | 0.03 | 0.35 |
pool_size | 2.01 | 0.30 | 1.22 |
conv_activation | 0.96 | 0.30 | 0.24 |
conv_dropout_rate | 0.89 | 1.45 | 0.65 |
n_conv_layers | 1.13 | 1.36 | 14.14 |
n_dense_nodes | 2.63 | 5.45 | 0.82 |
dense_activation | 0.69 | 0.05 | 0.15 |
dense_dropout_rate | 1.13 | 0.12 | 0.14 |
n_dense_layers | 5.81 | 1.85 | 6.34 |
batch_size | 1.93 | 58.98 | 49.81 |
fashion_mnist DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 28.66 | 0.23 | 0.24 |
learning_rate | 6.78 | 10.04 | 14.26 |
loss | 0.22 | 0.15 | 0.03 |
n_dense_nodes | 1.70 | 8.65 | 0.79 |
dense_activation | 1.05 | 0.29 | 0.36 |
dense_dropout_rate | 1.61 | 0.18 | 0.28 |
n_dense_layers | 12.91 | 3.84 | 19.46 |
batch_size | 2.87 | 57.59 | 46.24 |
cifar100 CNNModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 43.01 | 0.61 | 1.00 |
learning_rate | 3.47 | 2.32 | 3.66 |
loss | 0.13 | 0.05 | 0.07 |
n_kernels | 0.48 | 26.84 | 12.62 |
kernel_size | 0.66 | 0.24 | 0.60 |
pool_size | 0.37 | 0.24 | 0.33 |
conv_activation | 0.61 | 0.16 | 0.93 |
conv_dropout_rate | 0.84 | 0.36 | 0.84 |
n_conv_layers | 4.80 | 25.23 | 15.69 |
n_dense_nodes | 0.97 | 2.60 | 2.32 |
dense_activation | 1.03 | 0.14 | 0.48 |
dense_dropout_rate | 0.82 | 0.55 | 1.11 |
n_dense_layers | 1.03 | 0.22 | 1.92 |
batch_size | 1.10 | 4.24 | 9.34 |
cifar100 DenseModel | |||
---|---|---|---|
Hyperparameter | Performance | Training Time | Inference Time |
optimizer | 29.58 | 0.53 | 0.35 |
learning_rate | 8.88 | 21.45 | 13.25 |
loss | 0.20 | 0.02 | 0.04 |
n_dense_nodes | 2.05 | 7.07 | 0.73 |
dense_activation | 2.12 | 0.14 | 0.49 |
dense_dropout_rate | 3.93 | 0.23 | 0.71 |
n_dense_layers | 5.20 | 5.17 | 19.66 |
batch_size | 3.21 | 46.94 | 44.80 |