This is a relatively small release, but it’s an important one! You won’t see any big changes in the functionality (hopefully), but I made the preprocessing function of the images more flexible. This has positive consequences for the model trainer, which should converge quicker, but it also means that I had to modify the existing models.
They should be still compatible with older versions of Dafne, but for optimal incremental learning performance, please update.
Find the latest Dafne release here.
Also, grab the latest model trainer here.
Please bear in mind that this release is still experimental. Contact us to receive your personal API key.