INDEX

Explanations

words related to nationality and countries, particularly focusing on African and Middle Eastern regions

oai_token-act-pair · gpt-3.5-turbo

references to various nationalities or ethnic identifiers

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

ods

-0.79

oyer

-0.73

utherford

-0.73

ullivan

-0.73

raviolet

-0.72

soDeliveryDate

-0.70

lished

-0.68

irlf

-0.68

squ

-0.68

renheit

-0.66

POSITIVE LOGITS

 Airways

1.15

istan

1.14

 Airlines

1.09

 Nadu

1.06

 nationals

0.99

apolis

0.95

 Affairs

0.85

 Embassy

0.82

oglu

0.81

 Customs

0.81

Activations Density 0.090%

words related to nationality and countries, particularly focusing on African and Middle Eastern regions

references to various nationalities or ethnic identifiers

No Comments

No Known Activations