Neural-Process-Family
Neural-Process-Family copied to clipboard
YannDubs
Code for the Neural Processes website and replication of 4 papers on NPs. Pytorch implementation.
At the tutorial website (https://yanndubs.github.io/Neural-Process-Family), It seems the is_save=False should be save_file=None, of each line of **Make sure you use get_all_gp_datasets(is_save=False) if you use a large number of epochs.**
I could not find in your documentation what _"ELBOFalse_LatLBFalse_SigLBFalse"_ refers to. This is used to name the pretrained LNP models. Could you elaborate this
Do you have code for evaluating the marginal predictive in the case of LNPs for the 1d regression scenario? By this I mean combining the latent samples giving a single...
It seems like the CelebA datasets don't filter out non-matching sizes, which causes problems when trying to run the notebooks provided in `/reproducability` - the default behaviour seems to download...
When trying to run the provided Docker commands, `.` raises errors, and replacing with $PWD is an equivalent fix and should work on most Linux distributions. It also seems like...
How would one want to expand the NPs, lets say the LNP, to the case classification. I cant simply predict the location and scale of the Normal of the output...
I ran into the following dependency issues when trying to install it. I was able to install it by downgrading my python and matplotlib, but maybe you could add them...
`get_activation_name` appears to be expected to return a string using the `mapper`, but I think it was erroneously returning the key (a class, not a string). This leads to a...
