cifar10_resnet.yaml

# Cifar10 Classification Training:
# Reaches ~94.9 Accuracy after 250 Epochs
# Instructions:
#   0. Make sure that the data is stored in dataset_params.dataset_dir or add "dataset_params.data_dir=<PATH-TO-DATASET>" at the end of the command below (feel free to check ReadMe)
#   1. Move to the project root (where you will find the ReadMe and src folder)
#   2. Run the command:
#       python -m super_gradients.train_from_recipe --config-name=cifar10_resnet +experiment_name=cifar10
#
#   To use equivalent Albumentations transforms pipeline set dataset_params to cifar10_albumentations_dataset_params:
#     python -m super_gradients.train_from_recipe --config-name=cifar10_resnet dataset_params=cifar10_albumentations_dataset_params
defaults:
  - training_hyperparams: cifar10_resnet_train_params
  - dataset_params: cifar10_dataset_params
  - arch_params: resnet18_cifar_arch_params
  - checkpoint_params: default_checkpoint_params
  - _self_
  - variable_setup

train_dataloader: cifar10_train
val_dataloader: cifar10_val
architecture: resnet18_cifar
experiment_name: resnet18_cifar_interpolation_check
multi_gpu: Off
num_gpus: 1