INDEX

Explanations

directives or suggestions prompting action

oai_token-act-pair · gpt-3.5-turbo

phrases instructing the reader to take specific actions or follow advice

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

deen

-0.65

vell

-0.64

Ö¼

-0.63

ELD

-0.63

 territ

-0.62

 satisf

-0.55

Hum

-0.54

uko

-0.52

onew

-0.52

zb

-0.51

POSITIVE LOGITS

to

0.87

 yourself

0.83

 bookmark

0.79

 checking

0.78

 yourselves

0.78

 Nanto

0.76

 scrolling

0.73

 checkout

0.70

assetsadobe

0.70

 downloading

0.68

Activations Density 0.099%

directives or suggestions prompting action

phrases instructing the reader to take specific actions or follow advice

No Comments

No Known Activations