Simple repo to train and evaluate models using different activation functions
Train a model with with train.py
. During training you can use tensorboard to observe training loss & validation accuracy. Logs will be stored in the act_fn_experiment folder.
Args | Options | Description |
---|---|---|
dataset | mnist (more to come) | Selects which dataset to train on. |
model | fc (more to come) | Selects which model architecture to use. |
act_fn | sigmoid, tanh, relu, leakyrelu, elu, swish |
Selects activation function used in model |
epochs | [int] | Number of epochs for training. |
lr | [float] | Learning rate. |
batch_size | [int] | Number of samples per batch. |
Train a 2 layer fully connected model with relu activations on mnist
python train.py --dataset mnist --model fc --act_fn relu --epochs 30