5 ms·
Note that this reinforcement finetuning is something different than regular RLHF/DPO post training
by changoplatanero 2y ago
Note that this reinforcement finetuning is something different than regular RLHF/DPO post training
- whimsicalism 2y agoIs it? We have no idea.
- changoplatanero 2y agoYes it is. In RLHF and DPO you are optimizing the model output for human preferences. In the reinforcement fine tuning that was announced today you are optimizing the hidden chain of thought to arrive at a correct answer, as judged by a predefined grader.
- whimsicalism 2y agoI mean i think it could easily be PPO post training. if your point is that the rewards are different, sure