|
- import argparse
-
-
- def get_argument_parser():
- parser = argparse.ArgumentParser()
- parser.add_argument('--seed', default=1024, type=int,
- help='random seed')
- parser.add_argument('--data_path', default='/tmp/dataset/f30k_precomp',
- help='path to datasets')
- parser.add_argument('--data_name', default='f30k_precomp',
- help='{coco,f30k}_precomp')
- parser.add_argument('--vocab_path', default='/tmp/code/avse_test/vocab/',
- help='Path to saved vocabulary json files.')
- parser.add_argument('--margin', default=0.2, type=float,
- help='Rank loss margin.')
- parser.add_argument('--num_epochs', default=30, type=int,
- help='Number of training epochs.')
- parser.add_argument('--batch_size', default=128, type=int,
- help='Size of a training mini-batch.')
- parser.add_argument('--word_dim', default=300, type=int,
- help='Dimensionality of the word embedding.')
- parser.add_argument('--embed_size', default=1024, type=int,
- help='Dimensionality of the joint embedding.')
- parser.add_argument('--num_layers', default=1, type=int,
- help='Number of GRU layers.')
- parser.add_argument('--grad_clip', default=2., type=float,
- help='Gradient clipping threshold.')
- parser.add_argument('--learning_rate', default=.0005, type=float,
- help='Initial learning rate.')
- parser.add_argument('--lr_update', default=15, type=int,
- help='Number of epochs to update the learning rate.')
- parser.add_argument('--optim', default='adam', type=str,
- help='the optimizer')
- parser.add_argument('--workers', default=4, type=int,
- help='Number of data loader workers.')
- parser.add_argument('--log_step', default=100, type=int,
- help='Number of steps to logger.info and record the log.')
- parser.add_argument('--val_step', default=500, type=int,
- help='Number of steps to run validation.')
- parser.add_argument('--logger_name', default='/tmp/output/runs/baseline_36/log',
- help='Path to save Tensorboard log.')
- parser.add_argument('--model_name', default='/tmp/output/runs/baseline_36/checkpoint',
- help='Path to save the model.')
- parser.add_argument('--resume', default='', type=str, metavar='PATH',
- help='path to latest checkpoint (default: none)')
- parser.add_argument('--max_violation', action='store_true',
- help='Use max instead of sum in the rank loss.')
- parser.add_argument('--img_dim', default=2054, type=int,
- help='Dimensionality of the image embedding.')
- parser.add_argument('--block_dim', default=[512], type=int, nargs="+",
- help='Dimensionality of the image embedding.')
- parser.add_argument('--alpha', default=[1], type=int, nargs="+",
- help='Dimensionality of the image embedding.')
- parser.add_argument('--sim_dim', default=16, type=int,
- help='Dimensionality of the image embedding.')
- parser.add_argument('--no_imgnorm', action='store_true',
- help='Do not normalize the image embeddings.')
- parser.add_argument('--no_txtnorm', action='store_true',
- help='Do not normalize the text embeddings.')
- parser.add_argument('--use_gcn', action='store_true',
- help='Use GCN')
- parser.add_argument('--precomp_enc_type', default="basic",
- help='basic|backbone')
- parser.add_argument('--backbone_path', type=str, default='',
- help='path to the pre-trained backbone net')
- parser.add_argument('--backbone_source', type=str, default='detector',
- help='the source of the backbone model, detector|imagenet')
- parser.add_argument('--vse_mean_warmup_epochs', type=int, default=1,
- help='The number of warmup epochs using mean vse loss')
- parser.add_argument('--reset_start_epoch', action='store_true',
- help='Whether restart the start epoch when load weights')
- parser.add_argument('--backbone_warmup_epochs', type=int, default=5,
- help='The number of epochs for warmup')
- parser.add_argument('--embedding_warmup_epochs', type=int, default=2,
- help='The number of epochs for warming up the embedding layers')
- parser.add_argument('--backbone_lr_factor', default=0.01, type=float,
- help='The lr factor for fine-tuning the backbone, it will be multiplied to the lr of '
- 'the embedding layers')
- parser.add_argument('--input_scale_factor', type=float, default=1,
- help='The factor for scaling the input image')
- parser.add_argument('--drop', type=bool, default=True,
- help='Whether using drop words')
- parser.add_argument('--module_name', default='attn', type=str,
- help='SGR, SAF, GCN, attn')
- parser.add_argument('--feature_path', default='G:/data/flickr30k_vlp/region_feat_gvd_wo_bgd/trainval/',
- type=str, help='path to the pre-computed image features')
- parser.add_argument('--region_bbox_file',
- default='G:/data/flickr30k_vlp/region_feat_gvd_wo_bgd/flickr30k_detection_vg_thresh0.2_feat_gvd_checkpoint_trainvaltest.h5',
- type=str, help='path to the region_bbox_file(.h5)')
-
- return parser
|