This is why, whenever I can, I call RLHF/DPO "sequence level calibration" instead of "alignment tuning".
Some precursors to RLHF: https://arxiv.org/abs/2210.00045 https://arxiv.org/abs/2203.16804
Some precursors to RLHF: https://arxiv.org/abs/2210.00045 https://arxiv.org/abs/2203.16804
No comments yet.