HelloWorld

Results 8 comments of HelloWorld

1.When I use the provided pre-trained weights, the model behaves normally. 2.We print the ‘pred_wp’ from submission_agent.py, and the results are like: [[-0.6865, -0.1091], [-0.0335, -0.2020], [0.6574, -0.2611], [1.3772, -0.3137]];...

Thank you for your timely reply! 1.After we load the dataset, there are 226115 samples in ‘train_set’ and 4850 samples in ‘val_set’. 2.There are two figures for two training loss:...

Thanks for your timely reply! 1.I used auxiliary losses (‘only_wp=0’) at first training, but set ‘--setting’ = ‘02_05..’ not ‘all’. I think this the reason why the training diverged after...

Sorry for late reply. I refer your suggestion and reset the training configuration. After 41 epochs training, we evaluate the model and find the same problem as previous we meet....

That is very kind of you, thanks a lot. Can I have your email address? or you can sent an email to me to [email protected], so that I can get...

Thanks for your sincere help! The problem has been solved. My own trained model has been worked.

> > I also have the same requirement and hope to obtain the code for the first stage of training. > > I have reimplemented the stage 1 training. If...

> Hi! Thank you for your great work. > > Following your kind explanation, I was going to install the blender package. I've tried several ways but I finally failed...