INDEX

Explanations

references to political figures and positions

oai_token-act-pair · gpt-3.5-turbo

references to political leaders, specifically prime ministers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

¥µ

-0.76

wine

-0.74

emis

-0.74

 Paddock

-0.73

ersion

-0.72

oso

-0.72

avery

-0.70

adoes

-0.70

 Twain

-0.69

olescent

-0.69

POSITIVE LOGITS

 minister

1.06

 secretary

1.03

 Minister

0.97

 deputy

0.96

 ministerial

0.94

 treasury

0.92

 appoint

0.92

 treasurer

0.91

 senator

0.91

 prosecutor

0.90

Activations Density 0.065%

references to political figures and positions

references to political leaders, specifically prime ministers

No Comments

No Known Activations