neureka VS hypersistence-utils

I am using https://github.com/Gleethos/neureka to do some personal machine learning and also basic data science stuff at my workplace. It is inspired by PyTorch and it's dynamic autograd system (which records the computation graph to then traverse it for backpropagation eagerly). The library is super lightweight, has a nice API and documentation, but it's still very young and not as feature rich.

A lightweight option: Neureka - https://github.com/Gleethos/neureka

https://github.com/Gleethos/neureka A tensor/nd-array library inspired by Numpy and PyTorch. Supports:

If you are into deep learning: I work on an nd-array / tensor library with autograd support. https://github.com/Gleethos/neureka

Unfortunately, you cannot smoothly wrap the sequence-based IDs with value objects. There are solutions, but they are rather cumbersome. I left the proposal to the Hibernate types project of adding support to this feature. You’ve already seen the benefits of user-defined types. So, you can rate up my issue to make it more popular.

The Hibernate Types project.

Spring Data / Hibernate does not support JSON fields out-of-the-box, but there is a library that provides custom types for the most common database systems: hibernate-types. This is a fantastic open source library, please support it!

I will blog about it in the following months. So far, I managed to add a new Hibernate Types module for Hibernate 6 and released the 2.15 version.

For current work, hibernate-types works great to support json/jsonb.

Complex DB structures are not a problem. You can map any database type you want: JSON, HStore, Ranges, Inet, as I did in this hypernate-types project.

I am using the Hibernate Types library for custom Hibernate types.