请问怎样运行effedet呢？

请问有没有示例命令呀？我用的--model Effdet --transform frcnn --scheduler 1x --save_freq 1 --tag eff 会直接报错：

Traceback (most recent call last):
  File "/media/mcj/mcj/detection_mcj/train.py", line 56, in <module>
    model = Model(opt, logger)
  File "/media/mcj/mcj/detection_mcj/network/Effdet/Model.py", line 43, in __init__
    self.detector = get_net().to(device=opt.device)
  File "/media/mcj/mcj/detection_mcj/network/Effdet/Model.py", line 36, in get_net
    return DetBenchTrain(net, config)
  File "/media/mcj/mcj/detection_mcj/network/Effdet/effdet/bench.py", line 78, in __init__
    anchors = Anchors(
  File "/media/mcj/mcj/detection_mcj/network/Effdet/effdet/anchors.py", line 259, in __init__
    self.register_buffer('boxes', self._generate_boxes())
  File "/media/mcj/mcj/detection_mcj/network/Effdet/effdet/anchors.py", line 267, in _generate_boxes
    boxes = _generate_anchor_boxes(self.image_size, self.anchor_scale, self.config)
  File "/media/mcj/mcj/detection_mcj/network/Effdet/effdet/anchors.py", line 141, in _generate_anchor_boxes
    if image_size % stride != 0:
TypeError: unsupported operand type(s) for %: 'NoneType' and 'int'

貌似因为opt.scale没有指定，然后我指定比如128，又会报size不匹配的错误。

misads / easy_detection

请问怎样运行effedet呢？ #16