super-gradients/src/super_gradients/recipes/imagenet_regnetY.yaml

#  RegnetY Imagenet classification training:
#  This example trains with batch_size = 256
#  Training time on a single GeForce RTX 2080 Ti, and top1 accuracies:
#  11 days for RegnetY200, 70.88
#  12 days for RegnetY400, 74.74
#  19 days for RegnetY600, 76.18
#  20 days for RegnetY800, 77.07
#  NOTE: Training should probably be lower as resources were shared among the above runs.
#
#  Logs and tensorboards at:
# https://deci-pretrained-models.s3.amazonaws.com/RegnetY800/
# https://deci-pretrained-models.s3.amazonaws.com/RegnetY600/
# https://deci-pretrained-models.s3.amazonaws.com/RegnetY400/
# https://deci-pretrained-models.s3.amazonaws.com/RegnetY200/
#
# 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:
#         regnetY200: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY200
#         regnetY400: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY400
#         regnetY600: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY600
#         regnetY800: python src/super_gradients/examples/train_from_recipe_example/train_from_recipe.py --config-name=imagenet_regnetY architecture=regnetY800

defaults:
  - training_hyperparams: imagenet_regnetY_train_params
  - dataset_params: imagenet_regnetY_dataset_params
  - arch_params: regnetY_arch_params
  - checkpoint_params: default_checkpoint_params
  - _self_

arch_params:   
  num_classes: 1000
  dropout_prob: 0.5
  droppath_prob: 0.0

train_dataloader: imagenet_train
val_dataloader: imagenet_val


load_checkpoint: False
resume: False
training_hyperparams:
  resume: ${resume}


ckpt_root_dir:

multi_gpu: Off


architecture: regnetY800
experiment_name: ${architecture}


# THE FOLLOWING PARAMS ARE DIRECTLY USED BY HYDRA
hydra:
  run:
    # Set the output directory (i.e. where .hydra folder that logs all the input params will be generated)
    dir: ${hydra_output_dir:${ckpt_root_dir}, ${experiment_name}}