INDEX

Explanations

mentions of discrimination or inequality against minorities

oai_token-act-pair · gpt-3.5-turbo

references to marginalized and minority groups

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ENA

-0.82

FIN

-0.82

CHA

-0.76

amina

-0.76

rol

-0.74

×ŀ

-0.73

×¨

-0.72

PT

-0.72

rolog

-0.71

×

-0.71

POSITIVE LOGITS

 minorities

1.08

 genders

0.99

rats

0.89

 minority

0.83

 backgrounds

0.80

eatures

0.79

ecided

0.77

 whites

0.77

 unemploy

0.75

 males

0.75

Activations Density 0.005%

mentions of discrimination or inequality against minorities

references to marginalized and minority groups

No Comments

No Known Activations