Alignment is not free: How model upgrades can silence your confidence signals (variance.co) | HN Screenshots

YHacker Newsscreenshots

Screenshot of variance.co: Alignment is not free: How model upgrades can silence your confidence signals

First screen, captured 1 year ago.

Looks similar

Screenshot of seangoedecke.com: Sycophancy is the first LLM "dark pattern"

Sycophancy is the first LLM "dark pattern"(seangoedecke.com)

135 points by jxmorris12 10 months ago | 81 comments

Screenshot of off-policy.com: Alignment is capability

Alignment is capability (off-policy.com)

89 points by drctnlly_crrct 10 months ago | 79 comments

Screenshot of quantamagazine.org: A video tour of the Standard Model (2021)

A video tour of the Standard Model (2021)(quantamagazine.org)

92 points by goodway 1 year ago | 11 comments

Screenshot of ablg.io: No management needed: anti-patterns in early-stage engineering teams

No management needed: anti-patterns in early-stage engineering teams (ablg.io)

275 points by tonioab 9 months ago | 288 comments

Screenshot of minimallysufficient.com: LLM Classification Is Feature Engineering

LLM Classification Is Feature Engineering (minimallysufficient.com)

82 points by minsufficient 1 month ago | 16 comments

Screenshot of willybrauner.com: Signals, the push-pull based algorithm

Signals, the push-pull based algorithm (willybrauner.com)

143 points by mpweiher 6 months ago | 35 comments