INDEX

Explanations

capitalized words "AN" followed by another capitalized word

oai_token-act-pair · gpt-3.5-turbo

the mention of specific acronyms or abbreviations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 shale

-0.76

duct

-0.72

ucket

-0.68

ibles

-0.65

endas

-0.65

emouth

-0.64

 loft

-0.62

rero

-0.62

uration

-0.62

ynasty

-0.61

POSITIVE LOGITS

AN

3.49

AN

1.57

AS

1.43

ANI

1.33

AU

1.33

SAN

1.30

AUT

1.27

MAN

1.25

An

1.24

ART

1.21

Activations Density 0.008%

capitalized words "AN" followed by another capitalized word

the mention of specific acronyms or abbreviations

No Comments

No Known Activations