INDEX

Explanations

references to items being included in a list or collection

oai_token-act-pair · gpt-3.5-turbo

phrases that discuss inclusion

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Lives

-0.62

ater

-0.59

politics

-0.59

stead

-0.58

idium

-0.57

 preaching

-0.55

 powerless

-0.53

 havoc

-0.53

agon

-0.53

Bec

-0.52

POSITIVE LOGITS

 include

3.53

 Include

2.47

 includes

2.30

include

2.19

 included

2.10

 includ

1.91

 encompass

1.85

includes

1.72

 incorporate

1.71

 exclude

1.65

Activations Density 0.021%

references to items being included in a list or collection

phrases that discuss inclusion

No Comments

No Known Activations