INDEX

Explanations

terms related to various crimes and misconduct, particularly legal charges and violations of the law

oai_token-act-pair · gpt-3.5-turbo

terms related to legal and criminal activities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

isSpecialOrderable

-0.95

izons

-0.91

soDeliveryDate

-0.86

icon

-0.73

oyer

-0.71

rier

-0.70

Companies

-0.65

riers

-0.65

atorium

-0.64

ellen

-0.64

POSITIVE LOGITS

 offences

1.24

 violations

1.21

 misconduct

1.17

 manslaughter

1.14

 wrongdoing

1.13

 crimes

1.08

 irregularities

1.03

angering

1.03

 poisoning

1.02

 theft

1.01

Activations Density 0.286%

terms related to various crimes and misconduct, particularly legal charges and violations of the law

terms related to legal and criminal activities

No Comments

No Known Activations