INDEX

Explanations

written descriptions of glasses and their specific features

oai_token-act-pair · gpt-3.5-turbo

references to eyewear, specifically glasses and sunglasses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ILY

-0.96

heny

-0.76

actionDate

-0.74

raltar

-0.74

DAY

-0.71

itative

-0.71

anqu

-0.66

TAIN

-0.65

BTC

-0.64

Mech

-0.63

POSITIVE LOGITS

 glasses

1.37

 goggles

1.02

lasses

0.97

 sunglasses

0.96

creen

0.88

linger

0.87

lace

0.84

 tint

0.80

glass

0.79

 lenses

0.79

Activations Density 0.023%

written descriptions of glasses and their specific features

references to eyewear, specifically glasses and sunglasses

No Comments

No Known Activations