INDEX

Explanations

references to eyeglasses

oai_token-act-pair · gpt-3.5-turbo

references to glasses

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

ILY

-0.96

itative

-0.78

anqu

-0.70

actionDate

-0.68

heny

-0.68

noon

-0.67

ELY

-0.66

ctive

-0.65

TAIN

-0.65

Filename

-0.65

POSITIVE LOGITS

 glasses

1.37

 goggles

1.05

linger

0.96

creen

0.94

 sunglasses

0.89

lace

0.88

lasses

0.88

glass

0.86

 Bottle

0.85

pan

0.79

Activations Density 0.011%

references to eyeglasses

references to glasses

No Comments

No Known Activations