INDEX

Explanations

URLs starting with "https"

oai_token-act-pair · gpt-3.5-turbo

hyperlinks or URLs in the text

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

 Sierra

-0.72

 Morales

-0.69

 Morse

-0.67

rane

-0.66

 polyg

-0.64

 abst

-0.64

 Rath

-0.64

 Terra

-0.64

 trave

-0.63

 deceive

-0.62

POSITIVE LOGITS

://

1.56

doi

1.22

https

1.12

:/

1.05

twitter

1.00

soDeliveryDate

0.94

www

0.90

archive

0.89

 https

0.87

natureconservancy

0.83

Activations Density 0.005%

URLs starting with "https"

hyperlinks or URLs in the text

No Comments

No Known Activations