@misc{indiciae20b60da751a1,
  title = {KL-Regularised Q-Learning: A Token-level Action-Value perspective on Online RLHF},
  author = {Jason R Brown and Lennie Wells and Edward James Young and Sergio Bacallado},
  year = {2025},
  url = {https://arxiv.org/abs/2508.17000},
  note = {Source identifier: 2508.17000}
}