honestly, I want to reproduce this model. but after I git clone code and install megengine as well as install basecls with source code, I start running commands as follows:
./main_imagenet_test.py -f configs/replknet31_base_224_pt1k.py -w [weights] batch_size 64 data.val_path /path/to/imagenet/val
and then the error is reported as KeyError: "No object named 'DefaultSolver' found in 'solvers' registry!
it seems that something is wrong with the registry, but j just do not know why, as we already define it in the package basecls.
does anyone ever meet the same error as I did? hope for your reply!
honestly, I want to reproduce this model. but after I git clone code and install megengine as well as install basecls with source code, I start running commands as follows:
./main_imagenet_test.py -f configs/replknet31_base_224_pt1k.py -w [weights] batch_size 64 data.val_path /path/to/imagenet/val
and then the error is reported as KeyError: "No object named 'DefaultSolver' found in 'solvers' registry!
it seems that something is wrong with the registry, but j just do not know why, as we already define it in the package basecls.
does anyone ever meet the same error as I did? hope for your reply!