(All data presented are based on publicly available analytics and anonymised expert testimony; no proprietary or confidential information from XNW is disclosed.)
A with experience replay (size = 10⁵) and target network update every 1 000 steps is employed. Training converges after ~2 × 10⁶ steps (≈ 30 min of simulated time). xnxwapcom
No Assets in the basket.