INDEX

Explanations

references to royal figures or titles

oai_token-act-pair · gpt-3.5-turbo

references to queens and related royal terminology

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

iance

-0.77

imaru

-0.75

ints

-0.72

iances

-0.71

ugu

-0.70

 Injury

-0.70

iant

-0.69

angan

-0.68

orsche

-0.68

 Liberties

-0.66

POSITIVE LOGITS

 queens

1.21

 queen

1.20

Queen

0.93

bee

0.87

esses

0.84

 princess

0.82

pin

0.79

board

0.79

maid

0.78

 knight

0.76

Activations Density 0.005%

references to royal figures or titles

references to queens and related royal terminology

No Comments

No Known Activations