Victor Veitch @vveitch.bsky.social · 12/12/2024I'll present this poster tonight at East exhibit hall a-c 2510. 5-7:30 pm. Come chat about alignment! 070
Victor Veitch @vveitch.bsky.social · 10/12/2024I'll be at NeurIPS Thursday-Sunday; send me an email if you'd like to chat :) 020
Victor Veitch @vveitch.bsky.social · 23/11/2024LLM Alignment aims at making model outputs preferred by a ranker while changing as little 'off-target' behavior as possible. Turns out: -best-of-$n$ is the optimal option! -you can contrastively train an LLM to mimic its own best-of-$n$ distribution! BonBon alignment: arxiv.org/abs/2406.00832simons.berkeley.eduOn Spurious Associations and LLM AlignmentLarge language models are `aligned' to bias them towards outputting responses that are good on various measures---e.g., we may want them to be helpful, factual, and polite. Often, alignment procedures... 160
