# Regenz, Regenerate!

We have been iterating over the ML models over the past week and have now updated the metadata for the new and improved regenz! The entire dataset has been recreated with more than 5x the training time. We also duplicated the entire dataset and Phunked it to achieve extra accuracy.

We are happy to share these results with the community. Below I put some samples over a London background just because I think its pretty cool. There are way less "weird" results in this dataset. The scoring has been redone.


## The way forward

We have a couple exciting things on our roadmap:

-> We are going to train AI to generate abstract backgrounds for PFPs. Regen NFT holders will be whitelisted to mint their own unique backgrond which we will generate from our biggest training dataset yet

-> There have been many new releases over the past weeks that we want to try our methodology on. We will let the community decide which projects we should mashup next

-> We are now working on our metadata manager dapp which will let holders regenerate their avatars and mash their collection with the AI models.

## FAQ

### What if I liked my old regen better?

For those early regenz we will allow you to revert back to the original since we are saving the current metadata and database.

### Will this keep happening?

This will be the last massive regeneration event. We will work on our DAPP to allow new models to be used by the owner of each regen but we will not change the entire collection again.

## Links

https://y.at/💻🤓💰