
Orion 1.1, a new version of my little chess engine (50+ Elo at STC, unknown at LTC). Includes a bonus for French speakers
Changes are rather limited to minor improvements in thread/endgame management; the main gains are due to the larger network used.

Cerebrum 2.0, a new version of my NNUE trainer (simpler and more flexible)
Changes made aim to open up the (future) possibility of using engine evaluations directly as inputs for the training phase. Networks now predict scores directly in pawns, avoiding the combination of two predicted values in previous versions (win ratio and material). I hope all of this will be helpful for those who are interested!
Kind regards from France,
David