INDEX

Explanations

individuals involved in a conflict or crime

oai_token-act-pair · gpt-3.5-turbo

references to victims and perpetrators in contexts related to crime, law, and violence

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

udeb

-0.58

iban

-0.58

BuyableInstoreAndOnline

-0.58

 supervised

-0.56

 Bald

-0.56

 Hampton

-0.56

 Mush

-0.54

umat

-0.54

cott

-0.53

 democr

-0.53

POSITIVE LOGITS

worthiness

1.12

 accurately

0.97

 differently

0.95

 beforehand

0.90

 correctly

0.85

 accordingly

0.78

 patterns

0.76

vantage

0.76

 whereabouts

0.76

abilities

0.72

Activations Density 0.681%

individuals involved in a conflict or crime

references to victims and perpetrators in contexts related to crime, law, and violence

No Comments

No Known Activations