MCPcopy
hub / github.com/pytorch/vision / get_args_parser

Function get_args_parser

references/detection/train.py:71–179  ·  view source on GitHub ↗
(add_help=True)

Source from the content-addressed store, hash-verified

69
70
71def get_args_parser(add_help=True):
72 import argparse
73
74 parser = argparse.ArgumentParser(description="PyTorch Detection Training", add_help=add_help)
75
76 parser.add_argument("--data-path", default="/datasets01/COCO/022719/", type=str, help="dataset path")
77 parser.add_argument(
78 "--dataset",
79 default="coco",
80 type=str,
81 help="dataset name. Use coco for object detection and instance segmentation and coco_kp for Keypoint detection",
82 )
83 parser.add_argument("--model", default="maskrcnn_resnet50_fpn", type=str, help="model name")
84 parser.add_argument("--device", default="cuda", type=str, help="device (Use cuda or cpu Default: cuda)")
85 parser.add_argument(
86 "-b", "--batch-size", default=2, type=int, help="images per gpu, the total batch size is $NGPU x batch_size"
87 )
88 parser.add_argument("--epochs", default=26, type=int, metavar="N", help="number of total epochs to run")
89 parser.add_argument(
90 "-j", "--workers", default=4, type=int, metavar="N", help="number of data loading workers (default: 4)"
91 )
92 parser.add_argument("--opt", default="sgd", type=str, help="optimizer")
93 parser.add_argument(
94 "--lr",
95 default=0.02,
96 type=float,
97 help="initial learning rate, 0.02 is the default value for training on 8 gpus and 2 images_per_gpu",
98 )
99 parser.add_argument("--momentum", default=0.9, type=float, metavar="M", help="momentum")
100 parser.add_argument(
101 "--wd",
102 "--weight-decay",
103 default=1e-4,
104 type=float,
105 metavar="W",
106 help="weight decay (default: 1e-4)",
107 dest="weight_decay",
108 )
109 parser.add_argument(
110 "--norm-weight-decay",
111 default=None,
112 type=float,
113 help="weight decay for Normalization layers (default: None, same value as --wd)",
114 )
115 parser.add_argument(
116 "--lr-scheduler", default="multisteplr", type=str, help="name of lr scheduler (default: multisteplr)"
117 )
118 parser.add_argument(
119 "--lr-step-size", default=8, type=int, help="decrease lr every step-size epochs (multisteplr scheduler only)"
120 )
121 parser.add_argument(
122 "--lr-steps",
123 default=[16, 22],
124 nargs="+",
125 type=int,
126 help="decrease lr every step-size epochs (multisteplr scheduler only)",
127 )
128 parser.add_argument(

Callers 1

train.pyFile · 0.70

Calls

no outgoing calls

Tested by

no test coverage detected

Used in the wild real call sites across dependent graphs

searching dependent graphs…