INDEX

Explanations

mentions of illegal activities or situations

oai_token-act-pair · gpt-3.5-turbo

terms associated with illegal activities and crimes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

rike

-0.82

atche

-0.80

haps

-0.79

igating

-0.76

phasis

-0.75

vet

-0.73

lov

-0.72

attr

-0.71

eting

-0.71

oir

-0.71

POSITIVE LOGITS

 illegal

1.22

 illegally

1.02

 trafficking

0.95

 Illegal

0.95

 unauthorized

0.85

 illeg

0.83

illegal

0.83

 aliens

0.82

 unlawful

0.81

 immigrant

0.80

Activations Density 0.015%

mentions of illegal activities or situations

terms associated with illegal activities and crimes

No Comments

No Known Activations