running evaluation and layout generation problems:

[bschroedr]

Can you provide some instructions for running evaluation and generating layouts on the vg_msdn dataset (I already trained my own model)? I have been trying to do that using the inference_vg_msdn.py file, however I get the following error:

Traceback (most recent call last):
  File "inference/inference_vg_msdn.py", line 447, in <module>
    infer.check_from_model(100, dataset, model, self_input_dict = self_input_dict)
TypeError: check_from_model() got an unexpected keyword argument 'self_input_dict'

There seems to be a conflict of function parameters for check_from_model() here:

LayoutTransformer/inference/inference_vg_msdn.py

Line 77 in 23c940a

def check_from_model(self, dataset_idx, dataset, model, random=False, layout_save=None):

I am able to run evaluation from train.py but accuracies/stats show 0% and no layout is generated. I used the other script because I could customize my own scene graph.

I have also found vis.py for generating layouts but there are several command line options I am uncertain of how to set. If this is the better place to do layout, can you provide instructions for this (please)?