code/train_preangles.py
@@ -197,8 +197,6 @@ loss_pitch += alpha * loss_reg_pitch loss_roll += alpha * loss_reg_roll loss_yaw *= 1 loss_seq = [loss_yaw, loss_pitch, loss_roll] # loss_seq = [loss_reg_yaw, loss_reg_pitch, loss_reg_roll] grad_seq = [torch.Tensor(1).cuda(gpu) for _ in range(len(loss_seq))]
