INDEX

Explanations

mentions of injuries or harm to individuals

oai_token-act-pair · gpt-3.5-turbo

references to injuries and casualties

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

SpaceEngineers

-0.69

perm

-0.69

 garn

-0.68

DragonMagazine

-0.66

ramid

-0.62

Recipe

-0.62

unci

-0.62

istries

-0.60

alogy

-0.59

ãĥĪ

-0.59

POSITIVE LOGITS

 bystanders

0.86

jured

0.85

 wounding

0.79

../

0.76

 badly

0.74

lehem

0.73

 sustained

0.71

merga

0.69

 injuring

0.67

mond

0.67

Activations Density 0.045%

mentions of injuries or harm to individuals

references to injuries and casualties

No Comments

No Known Activations