# -------------------------------------------------------- # Reversible Column Networks # Copyright (c) 2022 Megvii Inc. # Licensed under TheApache License 2.0 [see LICENSE for details] # Written by Yuxuan Cai # -------------------------------------------------------- import os import yaml from yacs.config import CfgNode as CN _C = CN() # Base config files _C.BASE = [''] # ----------------------------------------------------------------------------- # Data settings # ----------------------------------------------------------------------------- _C.DATA = CN() # Batch size for a single GPU, could be overwritten by command line argument _C.DATA.BATCH_SIZE = 128 # Path to dataset, could be overwritten by command line argument _C.DATA.DATA_PATH = 'path/to/imagenet' # Dataset name _C.DATA.DATASET = 'imagenet' # Input image size _C.DATA.IMG_SIZE = 224 # Interpolation to resize image (random, bilinear, bicubic) _C.DATA.INTERPOLATION = 'bicubic' # Pin CPU memory in DataLoader for more efficient (sometimes) transfer to GPU. _C.DATA.PIN_MEMORY = True # Number of data loading threads _C.DATA.NUM_WORKERS = 8 # Path to evaluation dataset for ImageNet 22k _C.DATA.EVAL_DATA_PATH = 'path/to/eval/data' # ----------------------------------------------------------------------------- # Model settings # ----------------------------------------------------------------------------- _C.MODEL = CN() # Model type _C.MODEL.TYPE = '' # Model name _C.MODEL.NAME = '' # Checkpoint to resume, could be overwritten by command line argument _C.MODEL.RESUME = '' # Checkpoint to finetune, could be overwritten by command line argument _C.MODEL.FINETUNE = '' # Number of classes, overwritten in data preparation _C.MODEL.NUM_CLASSES = 1000 # Label Smoothing _C.MODEL.LABEL_SMOOTHING = 0.0 # ----------------------------------------------------------------------------- # Specific Model settings # ----------------------------------------------------------------------------- _C.REVCOL = CN() _C.REVCOL.INTER_SUPV = True _C.REVCOL.SAVEMM = True _C.REVCOL.FCOE = 4.0 _C.REVCOL.CCOE = 0.8 _C.REVCOL.KERNEL_SIZE = 3 _C.REVCOL.DROP_PATH = 0.1 _C.REVCOL.HEAD_INIT_SCALE = None # ----------------------------------------------------------------------------- # Training settings # ----------------------------------------------------------------------------- _C.TRAIN = CN() _C.TRAIN.START_EPOCH = 0 _C.TRAIN.EPOCHS = 300 _C.TRAIN.WARMUP_EPOCHS = 5 _C.TRAIN.WEIGHT_DECAY = 4e-5 _C.TRAIN.BASE_LR = 0.4 _C.TRAIN.WARMUP_LR = 0.05 _C.TRAIN.MIN_LR = 1e-5 # Clip gradient norm _C.TRAIN.CLIP_GRAD = 10.0 # Auto resume from latest checkpoint _C.TRAIN.AUTO_RESUME = True # Check point _C.TRAIN.USE_CHECKPOINT = False _C.TRAIN.AMP = True # LR scheduler _C.TRAIN.LR_SCHEDULER = CN() # LR scheduler _C.TRAIN.LR_SCHEDULER.NAME = 'cosine' # Epoch interval to decay LR, used in StepLRScheduler _C.TRAIN.LR_SCHEDULER.DECAY_EPOCHS = 30 # LR decay rate, used in StepLRScheduler _C.TRAIN.LR_SCHEDULER.DECAY_RATE = 0.1 # Optimizer _C.TRAIN.OPTIMIZER = CN() _C.TRAIN.OPTIMIZER.NAME = 'sgd' # Optimizer Epsilon fow adamw _C.TRAIN.OPTIMIZER.EPS = 1e-8 # Optimizer Betas fow adamw _C.TRAIN.OPTIMIZER.BETAS = (0.9, 0.999) # SGD momentum _C.TRAIN.OPTIMIZER.MOMENTUM = 0.9 # Layer Decay _C.TRAIN.OPTIMIZER.LAYER_DECAY = 1.0 # ----------------------------------------------------------------------------- # Augmentation settings # ----------------------------------------------------------------------------- _C.AUG = CN() # Color jitter factor _C.AUG.COLOR_JITTER = 0.4 # Use AutoAugment policy. "v0" or "original" _C.AUG.AUTO_AUGMENT = 'rand-m9-mstd0.5-inc1' # Random erase prob _C.AUG.REPROB = 0.25 # Random erase mode _C.AUG.REMODE = 'pixel' # Random erase count _C.AUG.RECOUNT = 1 # Mixup alpha, mixup enabled if > 0 _C.AUG.MIXUP = 0.8 # Cutmix alpha, cutmix enabled if > 0 _C.AUG.CUTMIX = 1.0 # Cutmix min/max ratio, overrides alpha and enables cutmix if set _C.AUG.CUTMIX_MINMAX = None # Probability of performing mixup or cutmix when either/both is enabled _C.AUG.MIXUP_PROB = 1.0 # Probability of switching to cutmix when both mixup and cutmix enabled _C.AUG.MIXUP_SWITCH_PROB = 0.5 # How to apply mixup/cutmix params. Per "batch", "pair", or "elem" _C.AUG.MIXUP_MODE = 'batch' # ----------------------------------------------------------------------------- # Testing settings # ----------------------------------------------------------------------------- _C.TEST = CN() # Whether to use center crop when testing _C.TEST.CROP = True # ----------------------------------------------------------------------------- # Misc # ----------------------------------------------------------------------------- # Path to output folder, overwritten by command line argument _C.OUTPUT = 'outputs/' # Tag of experiment, overwritten by command line argument _C.TAG = 'default' # Frequency to save checkpoint _C.SAVE_FREQ = 1 # Frequency to logging info _C.PRINT_FREQ = 100 # Fixed random seed _C.SEED = 0 # Perform evaluation only, overwritten by command line argument _C.EVAL_MODE = False # Test throughput only, overwritten by command line argument _C.THROUGHPUT_MODE = False # local rank for DistributedDataParallel, given by command line argument _C.LOCAL_RANK = 0 # EMA _C.MODEL_EMA = False _C.MODEL_EMA_DECAY = 0.9999 # Machine _C.MACHINE = CN() _C.MACHINE.MACHINE_WORLD_SIZE = None _C.MACHINE.MACHINE_RANK = None def _update_config_from_file(config, cfg_file): config.defrost() with open(cfg_file, 'r') as f: yaml_cfg = yaml.load(f, Loader=yaml.FullLoader) for cfg in yaml_cfg.setdefault('BASE', ['']): if cfg: _update_config_from_file( config, os.path.join(os.path.dirname(cfg_file), cfg) ) print('=> merge config from {}'.format(cfg_file)) config.merge_from_file(cfg_file) config.freeze() def update_config(config, args): _update_config_from_file(config, args.cfg) config.defrost() if args.opts: config.merge_from_list(args.opts) # merge from specific arguments if args.batch_size: config.DATA.BATCH_SIZE = args.batch_size if args.data_path: config.DATA.DATA_PATH = args.data_path if args.resume: config.MODEL.RESUME = args.resume if args.finetune: config.MODEL.FINETUNE = args.finetune if args.use_checkpoint: config.TRAIN.USE_CHECKPOINT = True if args.output: config.OUTPUT = args.output if args.tag: config.TAG = args.tag if args.eval: config.EVAL_MODE = True if args.model_ema: config.MODEL_EMA = True config.dist_url = args.dist_url # set local rank for distributed training config.LOCAL_RANK = args.local_rank # output folder config.OUTPUT = os.path.join(config.OUTPUT, config.MODEL.NAME, config.TAG) config.freeze() def get_config(args): """Get a yacs CfgNode object with default values.""" # Return a clone so that the defaults will not be altered # This is for the "local variable" use pattern config = _C.clone() update_config(config, args) return config