re-produced training ok, but Error(s) in loading state_dict for Network

[PeterFWS]

Hello,

Thanks for your great work,

I have trained your network on dataset NHR/sport1 myself. However, when I try to load the checkpoint, there is a loading error:

raise RuntimeError('Error(s) in loading state_dict for {}:\n\t{}'.format( RuntimeError: Error(s) in loading state_dict for Network: Missing key(s) in state_dict: "encodings", "occupancy_grids". Unexpected key(s) in state_dict: "latent_vectors.down1.0.0.weight", "latent_vectors.down1.0.0.bias", "latent_vectors.down1.0.1.weight", "latent_vectors.down1.0.1.bias", "latent_vectors.down1.0.2.weight", "latent_vectors.down1.0.2.bias", "latent_vectors.down1.0.3.weight", "latent_vectors.down1.0.3.bias", "latent_vectors.down1.0.4.weight", "latent_vectors.down1.0.4.bias", "latent_vectors.down1.0.5.weight", "latent_vectors.down1.0.5.bias", "latent_vectors.down1.0.6.weight", "latent_vectors.down1.0.6.bias", "latent_vectors.down1.1.0.weight", "latent_vectors.down1.1.0.bias", "latent_vectors.down1.1.1.weight", "latent_vectors.down1.1.1.bias", "latent_vectors.down1.1.2.weight", "latent_vectors.down1.1.2.bias", "latent_vectors.down1.1.3.weight", "latent_vectors.down1.1.3.bias", "latent_vectors.down1.1.4.weight", "latent_vectors.down1.1.4.bias", "latent_vectors.down1.1.5.weight", "latent_vectors.down1.1.5.bias", "latent_vectors.down1.1.6.weight", "latent_vectors.down1.1.6.bias", "latent_vectors.down1.2.0.weight", "latent_vectors.down1.2.0.bias", "latent_vectors.down1.2.1.weight", "latent_vectors.down1.2.1.bias", "latent_vectors.down1.2.2.weight", "latent_vectors.down1.2.2.bias", "latent_vectors.down1.2.3.weight", "latent_vectors.down1.2.3.bias", "latent_vectors.down1.2.4.weight", "latent_vectors.down1.2.4.bias", "latent_vectors.down1.2.5.weight", "latent_vectors.down1.2.5.bias", "latent_vectors.down1.2.6.weight", "latent_vectors.down1.2.6.bias", "latent_vectors.down2.weight", "latent_vectors.down2.bias", "latent_vectors.mu.weight", "latent_vectors.mu.bias", "latent_vectors.logstd.weight", "latent_vectors.logstd.bias".

There was no problem loading the pre-trained checkpoint provided by you. I am very appreciative if you have any feedback about this.

Best regards,

[pengsida]

It seems that you used an incorrect command. Which command did you used?

[PeterFWS]

python run.py --config configs/nhr/sport1.py mode evaluate fast_render True

[PeterFWS]

Oh, sorry,

I forget to post-process the trained model.

Now everything is working.

Thanks.