[Linkpost] Interpreting Language Model Parameters

·LessWrong··

This is the latest work in our Parameter Decomposition agenda. We introduce a new parameter decomposition method, adVersarial Parameter Decomposition (VPD)[1] and decompose the parameters of a small[2] language model with it. VPD greatly improves on our previous techniques, Stochastic Parameter Decomposition (SPD) and Attribution-based Parameter Decomposition (APD). We think the parameter decomposition approach is now more-or-less ready to be applied at scale to models people care about.Importan...

Read full article →

Related Articles

Probing the loss-band sparsity assumption in Scientist AI
Alejandro Tlaie · LessWrong · 52m ago
SFT Drives Gemini’s Safety Properties
Josh Engels · Alignment Forum · 22d ago
Sequent: scale and automation for higher confidence in alignment
Geoffrey Irving · Alignment Forum · 25d ago
A Mike's-Eye View of ARC's Research
Michael Winer · ARC · 25d ago
My research: a computational cognitive neuroscience perspective on alignment
Seth Herd · Alignment Forum · 1mo ago