INDEX

Explanations

phrases related to things or people being unpopular

oai_token-act-pair · gpt-3.5-turbo

references to unpopularity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

chn

-0.90

urers

-0.85

ramid

-0.83

uther

-0.75

CONCLUS

-0.74

ult

-0.73

EStreamFrame

-0.71

hens

-0.71

tein

-0.71

anwhile

-0.70

POSITIVE LOGITS

 unpopular

1.23

ity

1.20

 incumbent

0.88

ities

0.87

liest

0.83

 burdens

0.83

 disadvant

0.78

lihood

0.73

 partisan

0.73

 plag

0.72

Activations Density 0.014%

phrases related to things or people being unpopular

references to unpopularity

No Comments

No Known Activations