search

RomeroBarata / human_object_interaction

Code for CVPR'21 paper "Learning Asynchronous and Sparse Human-Object Interaction in Videos".
22 stars 3 forks source link

the parameters setting #6

Closed SISTMrL closed 2 years ago

SISTMrL commented 2 years ago

hello, in your released repo, the conf/model/assign.yaml's parameters setting is not same to your release trained model, such as epoch=1, hidden_size=2, they are not same as with your released trained model, should i modify them in assign.yaml and train the model to reproduce your results, looking forward to your reply, thanks!

RomeroBarata commented 2 years ago

Hi SISTMrL,

That's right, you have to modify the values in assign.yaml's file to train a model similar to the pre-trained model provided.

SISTMrL commented 2 years ago

@RomeroBarata thanks for your reply