INDEX

Explanations

mentions of incidents involving being shot

oai_token-act-pair · gpt-3.5-turbo

references to shooting incidents

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

DragonMagazine

-0.99

wcsstore

-0.95

å¥³

-0.85

Else

-0.80

ebook

-0.80

algia

-0.77

é¾įå¥ĳå£«

-0.76

é£

-0.73

DAY

-0.73

hist

-0.72

POSITIVE LOGITS

nikov

0.95

powder

0.93

 rampage

0.84

chens

0.80

 Trayvon

0.79

guns

0.79

gun

0.77

 guns

0.76

 spree

0.76

 fired

0.75

Activations Density 0.035%

mentions of incidents involving being shot

references to shooting incidents

No Comments

No Known Activations