INDEX

Explanations

news headlines mentioning newspapers or media outlets

oai_token-act-pair · gpt-3.5-turbo

negative or critical references to media sources or publications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

 hiber

-0.88

achev

-0.83

idious

-0.79

 regress

-0.76

itary

-0.72

milo

-0.71

adequ

-0.71

 adulthood

-0.70

 hitch

-0.70

 analogue

-0.69

POSITIVE LOGITS

University

1.09

Times

1.03

Office

0.91

Press

0.88

Union

0.86

District

0.85

Atlanta

0.85

Metro

0.84

Mail

0.84

Ox

0.83

Activations Density 0.099%

news headlines mentioning newspapers or media outlets

negative or critical references to media sources or publications

No Comments

No Known Activations