INDEX

Explanations

phrases related to crimes or violent incidents

oai_token-act-pair · gpt-3.5-turbo

references to individuals involved in violent actions or incidents, particularly gunmen and attackers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

umph

-0.86

Balt

-0.74

 detrim

-0.70

payer

-0.69

 Herald

-0.64

Sense

-0.64

olesterol

-0.64

agra

-0.64

utical

-0.63

ensus

-0.63

POSITIVE LOGITS

 penetrated

1.06

 stole

1.06

 detonated

1.03

 stormed

1.01

 gunmen

1.00

 attacked

0.97

 assailants

0.95

 attackers

0.94

 blew

0.93

 intent

0.92

Activations Density 0.121%

phrases related to crimes or violent incidents

references to individuals involved in violent actions or incidents, particularly gunmen and attackers

No Comments

No Known Activations