INDEX

Explanations

represent

np_max-act · gemini-2.0-flash

This neuron fires on terms denoting deceptive or fraudulent behavior (e.g. “misrepresentations,” “fraud,” “conceal”), highlighting passages about wrongdoing by deception.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

']].

-0.06

gro

-0.06

 contributor

-0.06

 ocur

-0.06

:focus

-0.06

Mozilla

-0.06

steen

-0.06

 Blair

-0.06

/access

-0.06

 THEN

-0.06

POSITIVE LOGITS

RED

0.07

endo

0.06

 Martha

0.06

actic

0.06

 heavyweight

0.06

-face

0.06

країн

0.06

�

0.06

 Bordeaux

0.06

 mandatory

0.06

Activations Density 0.005%

represent

This neuron fires on terms denoting deceptive or fraudulent behavior (e.g. “misrepresentations,” “fraud,” “conceal”), highlighting passages about wrongdoing by deception.

No Comments

No Known Activations