fry69 @fry69.dev · 16/09/2026Homepage for "Beyond Binary Rewards: Training LMs to Reason About Their Uncertainty", which seem to have started all this. -> rl-calibration.github.io 191
fry69 @fry69.dev · 16/09/2026arXiv paper from June 2025 -> arxiv.org/abs/2507.16806arxiv.orgBeyond Binary Rewards: Training LMs to Reason About Their UncertaintyWhen language models (LMs) are trained via reinforcement learning (RL) to generate natural language "reasoning chains", their performance improves on a variety of difficult question answering tasks. T... 030
