INDEX

Explanations

references to newsletters

oai_token-act-pair · gpt-3.5-turbo

mentions of newsletters, particularly from the New York Times

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ben

-0.64

OC

-0.62

chance

-0.60

 Hills

-0.59

 mole

-0.57

GB

-0.57

jury

-0.56

eur

-0.56

 reformed

-0.56

jaw

-0.56

POSITIVE LOGITS

azines

0.87

insula

0.85

ļéĨĴ

0.82

 Travels

0.81

ĸļ

0.81

é¾įå¥ĳå£«

0.81

soever

0.80

dayName

0.80

sylv

0.78

idth

0.78

Activations Density 0.013%

references to newsletters

mentions of newsletters, particularly from the New York Times

No Comments

No Known Activations