Missing weights in two_head.pt

[ghost]

Hi @pb-brainiac @ssegvic @orsic @jsaric, I am trying to use the pretrained two head model. But it seems that the provided checkpoint two_head.pt is missing weights for the outlier detection head. I am getting following message when loading the state dict:

missing keys in state_dict: "{'aux_logits_ood.upsample_1.conv.bias', 'aux_logits_ood.upsample_0.norm.bias', 'aux_logits_ood.upsample_0.norm.weight', 'aux_logits_ood.upsample_0.conv.weight', 'aux_logits_ood.upsample_0.norm.running_mean', 'aux_logits_ood.upsample_2.norm.bias', 'aux_logits_ood.upsample_1.norm.weight', 'aux_logits_ood.upsample_1.conv.weight', 'aux_logits_ood.upsample_2.conv.weight', 'aux_logits_ood.upsample_2.norm.running_mean', 'aux_logits_ood.upsample_2.norm.weight', 'aux_logits_ood.upsample_1.norm.bias', 'aux_logits_ood.upsample_0.norm.num_batches_tracked', 'aux_logits_ood.upsample_1.norm.running_var', 'aux_logits_ood.upsample_2.norm.running_var', 'aux_logits_ood.upsample_2.norm.num_batches_tracked', 'aux_logits_ood.upsample_1.norm.num_batches_tracked', 'aux_logits_ood.upsample_0.conv.bias', 'aux_logits_ood.upsample_2.conv.bias', 'aux_logits_ood.upsample_0.norm.running_var', 'aux_logits_ood.upsample_1.norm.running_mean'}"

Could you please provide model with missing weights?