INDEX

Explanations

references to specific locations, particularly squares

oai_token-act-pair · gpt-3.5-turbo

references to public squares

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ERAL

-0.83

urally

-0.69

enegger

-0.69

eworld

-0.66

Ì¶

-0.65

netflix

-0.65

é»Ĵ

-0.63

ictional

-0.63

opathy

-0.62

orship

-0.61

POSITIVE LOGITS

 Enix

1.44

 Mile

0.93

pants

0.92

Square

0.89

 Square

0.88

 Feet

0.77

cade

0.74

 Flavoring

0.73

faces

0.72

asaki

0.71

Activations Density 0.013%

references to specific locations, particularly squares

references to public squares

No Comments

No Known Activations