INDEX

Explanations

mentions of criminal activities or investigations

oai_token-act-pair · gpt-3.5-turbo

references to crime and criminal behavior

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

)=(

-0.83

achev

-0.80

ersed

-0.77

arity

-0.77

por

-0.72

zl

-0.71

DCS

-0.70

artney

-0.69

atche

-0.69

chell

-0.68

POSITIVE LOGITS

 spree

1.10

 perpetrated

0.84

fighting

0.84

fighter

0.81

 synd

0.80

 crimes

0.78

 crime

0.77

 unfocusedRange

0.76

 Victim

0.75

 homicide

0.74

Activations Density 0.028%

mentions of criminal activities or investigations

references to crime and criminal behavior

No Comments

No Known Activations