INDEX

Explanations

phrases related to violent actions or criminal behavior

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Configuration

neuronpedia/gpt2-small__res_scl-ajt/6-res_scl-ajt

Prompts (Dashboard)

12,288 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

46,080

Data Type

torch.float32

Hook Point

blocks.6.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

apollo-research/Skylion007-openwebtext-tokenizer-gpt2

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

verning

-0.91

igham

-0.86

sonian

-0.83

çīĪ

-0.74

aver

-0.73

lished

-0.73

ortium

-0.72

agall

-0.71

iddy

-0.71

clus

-0.69

POSITIVE LOGITS

 imaginable

0.93

 Syndrome

0.84

 harming

0.73

 syndrome

0.72

 outweigh

0.72

 perpetrated

0.70

 inflicted

0.70

 Boko

0.69

 fentanyl

0.66

ÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤÃĥÃĤ

0.66

Activations Density 15.605%

phrases related to violent actions or criminal behavior

No Comments

No Known Activations