INDEX

Explanations

mentions of individuals by their names or accounts on social media platforms

oai_token-act-pair · gpt-3.5-turbo

references to popular television shows and their characters

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

 Okinawa

-0.99

 Sapp

-0.97

 Buddhism

-0.93

Zen

-0.93

 Buddhist

-0.91

 Dresden

-0.86

 Sega

-0.86

Psy

-0.86

 Helsinki

-0.83

Japan

-0.83

POSITIVE LOGITS

 Oliver

2.26

Ol

1.79

 Arrow

1.71

 Stewart

1.53

 Laurel

1.47

Stew

1.43

 Canary

1.38

 Colbert

1.26

 Olsen

1.17

Stephen

1.17

Activations Density 0.209%

mentions of individuals by their names or accounts on social media platforms

references to popular television shows and their characters

No Comments

No Known Activations