INDEX

Explanations

mentions of derogatory terms or prejudiced language

oai_token-act-pair · gpt-3.5-turbo

references to racism and bigotry

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

 backup

-0.82

 resil

-0.77

 GOODMAN

-0.74

 rotation

-0.74

 refurb

-0.73

ufact

-0.73

 refin

-0.73

oscopic

-0.70

 remod

-0.69

 renovations

-0.68

POSITIVE LOGITS

 bigot

1.03

 bigotry

0.97

 coward

0.92

Semitic

0.88

 slurs

0.86

 unworthy

0.85

 perpetrated

0.85

 cowardly

0.84

 insin

0.80

 hypocrisy

0.79

Activations Density 0.978%

mentions of derogatory terms or prejudiced language

references to racism and bigotry

No Comments

No Known Activations