INDEX

Explanations

phrases related to criminal activities and legal procedures

oai_token-act-pair · gpt-3.5-turbo

terms related to legal charges and offenses involving harm or assault

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

erella

-0.84

isSpecialOrderable

-0.84

abeth

-0.83

DragonMagazine

-0.78

anmar

-0.76

alth

-0.75

ellen

-0.75

peror

-0.71

ilight

-0.71

aceae

-0.70

POSITIVE LOGITS

 unlawfully

1.18

 unlawful

1.12

 prejud

1.00

 negligent

0.99

 lawfully

0.97

 offender

0.93

 tresp

0.91

 intoxicated

0.91

 improper

0.90

 improperly

0.90

Activations Density 0.906%

phrases related to criminal activities and legal procedures

terms related to legal charges and offenses involving harm or assault

No Comments

No Known Activations