INDEX

Explanations

security-related terms and actions involving law enforcement

oai_token-act-pair · gpt-3.5-turbo

references to law enforcement and military presence

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Gutenberg

-0.86

 blasp

-0.75

ãĤ´ãĥ³

-0.74

ãĤ¦ãĤ¹

-0.71

conom

-0.70

olkien

-0.69

etus

-0.69

 UNESCO

-0.69

pring

-0.68

 inheritance

-0.68

POSITIVE LOGITS

 SWAT

1.51

 patrol

1.46

 troopers

1.36

 officers

1.32

 patrols

1.29

 detectives

1.28

 police

1.28

 cops

1.28

 helicopters

1.24

 policemen

1.24

Activations Density 0.422%

security-related terms and actions involving law enforcement

references to law enforcement and military presence

No Comments

No Known Activations