selfconfid-classif.yaml

# Data parameters
data:
    dataset: cifar10
    data_dir: data/cifar10-data
    input_size: [32,32]
    input_channels: 3
    num_classes: 10
    valid_size: 0.1

# Training parameters
training:
    output_folder: /home/docker_user/shared/ccorbiere/logs/test/cifar10_selfconfid
    task: classification
    learner: selfconfid
    nb_epochs: 500
    batch_size: 128
    loss:
        name: selfconfid_mse
        weighting: 1
    optimizer:
        name: adam
        lr: 0.0001
        #momentum: 0.9
        #weight_decay: 0.0001
    lr_schedule:
    ft_on_val: False
    metrics: ['accuracy', 'auc', 'ap_success', 'ap_errors']
    pin_memory: False
    num_workers: 3
    augmentations:
        hflip: True
        rotate: 15
        #rcrop: 32
        normalize: [[0.4914, 0.4822, 0.4465], [0.2023, 0.1994, 0.2010]]

# Model parameters
model:
    name: vgg16_selfconfid_classic
    resume: /home/docker_user/shared/ccorbiere/cifar10_pretrained/baseline/model_epoch_197.ckpt
    #uncertainty: /home/docker_user/shared/ccorbiere/logs/test/cifar10_selfconfid/model_epoch_005.ckpt