You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Kind creators. I encountered many configuration issues while running the project.
In scripts/multi30k-train.sh, the parameter arch is set to gcn, but it raised an error: train.py: error: argument --arch/-a: invalid choice: 'gcn' (choose from 'SGHallucUMMT', 'SGHallucUMMT_small', 'SGHallucUMMT_tiny')
--utils, --vis-data, and --vis-data-args might not be correct parameters, because in models/modules: if args.vis_base_encoder == 'RCNN': self.base_model = vision.__dict__[args.vis_encoder_arch]( model_path=args.vis_encoder_model_path, config_path=args.vis_encoder_config_path, codebook=args.vis_encoder_use_codebook, )
This results in: AttributeError: 'Namespace' object has no attribute 'vis_base_encoder'
Moreover, I can't find package vision, leading to NameError: name 'vision' is not defined
Can you provide the correct code?
The text was updated successfully, but these errors were encountered:
Kind creators. I encountered many configuration issues while running the project.
In scripts/multi30k-train.sh, the parameter
arch
is set togcn
, but it raised an error:train.py: error: argument --arch/-a: invalid choice: 'gcn' (choose from 'SGHallucUMMT', 'SGHallucUMMT_small', 'SGHallucUMMT_tiny')
--utils, --vis-data, and --vis-data-args might not be correct parameters, because in models/modules:
if args.vis_base_encoder == 'RCNN': self.base_model = vision.__dict__[args.vis_encoder_arch]( model_path=args.vis_encoder_model_path, config_path=args.vis_encoder_config_path, codebook=args.vis_encoder_use_codebook, )
This results in:
AttributeError: 'Namespace' object has no attribute 'vis_base_encoder'
Moreover, I can't find package
vision
, leading toNameError: name 'vision' is not defined
Can you provide the correct code?
The text was updated successfully, but these errors were encountered: