6 Commits

Author	SHA1	Message	Date
Josh Creek	e56850a236	fix(training): make policy evaluation portable	2026-09-01 18:56:31 +01:00
Josh Creek	4c1ed87344	test(training): verify 2v2 evaluator command	2026-09-01 17:32:10 +01:00
Josh Creek	7b2f9c26f4	feat(training): add opt-in teamplay evaluation	2026-09-01 17:30:40 +01:00
Josh Creek	33952b3cd0	feat(training): pair evaluations across physical sides	2026-08-08 14:55:12 +01:00
Josh Creek	8c15c466ef	fix(*): apply the locomotion mask during in-game/eval inference, not just training ... AIShipController (eval + real gameplay) ran the raw policy output unmasked regardless of allow_vertical/allow_pitch_roll, while ShipAIController (training) correctly discarded those axes for grounded curriculum stages. A grounded-trained model's untrained vertical/pitch-roll output reached the ship as noise during eval, understating it against models that were never handicapped this way.	2026-07-21 22:23:09 +01:00
Josh Creek	85f96eb15e	feat(*): Add self-play RL training pipeline with PPO trainer, in-game GDScript policy inference, and bot opponent support in Match mode	2026-07-18 19:32:51 +01:00