INDEX

Explanations

the word "game" with varying activations

oai_token-act-pair · gpt-3.5-turbo

references to video games

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

eur

-0.70

cliffe

-0.68

acid

-0.66

 Cosponsors

-0.65

 Emerson

-0.65

yll

-0.64

yne

-0.63

Param

-0.63

iewicz

-0.62

 testified

-0.61

POSITIVE LOGITS

 game

3.37

game

2.75

Game

2.56

 GAME

2.36

 games

2.35

 Game

2.31

GAME

2.13

games

1.99

 gameplay

1.92

Games

1.68

Activations Density 0.053%

the word "game" with varying activations

references to video games

No Comments

No Known Activations