INDEX

Explanations

words related to telling lies or being deceitful

oai_token-act-pair · gpt-3.5-turbo

instances of the word "lying" to identify discussions of dishonesty or deception

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Ultra

-0.77

obs

-0.76

ains

-0.76

aldi

-0.76

ugal

-0.75

iles

-0.73

ISO

-0.72

ilation

-0.71

arthy

-0.69

FN

-0.69

POSITIVE LOGITS

utenant

0.79

 vulner

0.79

 dormant

0.79

sembly

0.74

lie

0.74

 detector

0.73

uten

0.73

 awake

0.71

 lying

0.70

 skelet

0.69

Activations Density 0.012%

words related to telling lies or being deceitful

instances of the word "lying" to identify discussions of dishonesty or deception

No Comments

No Known Activations