x

LESSWRONG

LW

martinveron — LessWrong

martinveron

martinveron

Message

34

2

1

2mo

martinveron

34

2mo

Three Pieces of Evidence that the NLA Reconstructor Does Not Read Semantics or Syntax

TL;DR * The autoreconstructor (AR) in a natural language autoencoder (NLA) is the entire argument for the validity of the method, and at present is poorly understood. * The AR does not condition strongly on English grammar or semantics. Three bits of evidence: 1. Word pairs that are syntactically related...

Another Slice of Swiss Cheese for Untrusted Monitoring

Catching the monitor in a lie by comparing what it perceives against what it reports. TL;DR * I trained a linear probe to directly perceive code back-doors, which are represented linearly in activation space, using honest behaviour that can be elicited even from a scheming model. * I then trained...