INDEX

Explanations

the word "arian."

oai_token-act-pair · gpt-3.5-turbo

references to different types of -arian ideologies and identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

pressure

-0.69

same

-0.67

Cap

-0.66

 Fargo

-0.64

pler

-0.64

queue

-0.63

bring

-0.62

boards

-0.62

station

-0.62

Track

-0.61

POSITIVE LOGITS

arian

1.61

arians

1.39

ism

0.85

 glances

0.83

alyst

0.80

itarian

0.80

nian

0.79

tarian

0.78

oad

0.78

ity

0.77

Activations Density 0.013%

the word "arian."

references to different types of -arian ideologies and identities

No Comments

No Known Activations