Репост из: Does RL work yet?
Самое время подвести итоги сабмитов на ICLR 2025. Не все успели довести до фулл статей, но и воркшопы тоже хорошо!
Что-то уже есть на архиве (и еще будет обновлятся с новыми результатами), так что делюсь:
1. XLand-100B: A Large-Scale Multi-Task Dataset for In-Context Reinforcement Learning (Main, Poster), by @howuhh and @suessmann and @zzmtsvv
2. Latent Action Learning Requires Supervision in the Presence of Distractors (Workshop, World Models), by @howuhh
3. Object-Centric Latent Action Learning (Workshop, World Models), by @cinemere
4. N-Gram Induction Heads for In-Context RL: Improving Stability and Reducing Data Needs (Workshop, SCOPE), by @suessmann
5. Yes, Q-learning Helps Offline In-Context RL (Workshop, SSI-FM), by @adagrad
Что-то уже есть на архиве (и еще будет обновлятся с новыми результатами), так что делюсь:
1. XLand-100B: A Large-Scale Multi-Task Dataset for In-Context Reinforcement Learning (Main, Poster), by @howuhh and @suessmann and @zzmtsvv
2. Latent Action Learning Requires Supervision in the Presence of Distractors (Workshop, World Models), by @howuhh
3. Object-Centric Latent Action Learning (Workshop, World Models), by @cinemere
4. N-Gram Induction Heads for In-Context RL: Improving Stability and Reducing Data Needs (Workshop, SCOPE), by @suessmann
5. Yes, Q-learning Helps Offline In-Context RL (Workshop, SSI-FM), by @adagrad