Error in train.py

[ialex2004]

Hello,

I just downloaded your repo, set up conda env with requirements.txt and torch1.0.0. Downloaded datasets by the links in README.md and got the following error when calling train.py:

python run.py --config config/mgif-256.yaml --device_ids 0
/gpfs-volume/miniconda3/envs/fomm/lib/python3.6/importlib/_bootstrap.py:219: RuntimeWarning: numpy.dtype size changed, may indicate binary incompatibility. Expected 96, got 88
  return f(*args, **kwds)
Traceback (most recent call last):
  File "run.py", line 68, in <module>
    dataset = FramesDataset(is_train=(opt.mode == 'train'), **config['dataset_params'])
TypeError: __init__() got an unexpected keyword argument 'image_shape'

[AliaksandrSiarohin]

Hi, change image_shape to the frame_shape in mgif-256.yaml. This is some refactoring error.