INDEX

Explanations

violent and disturbing actions or events involving people and law enforcement officers

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

•Layer 6 UMAP region: e2e-recon cluster above bottome line touching local cluster - local

No Comments

Negative Logits

 nodd

-0.68

 defe

-0.67

 extinct

-0.66

 quir

-0.65

 everywhere

-0.64

 royalty

-0.63

 renaissance

-0.63

omics

-0.63

 standings

-0.63

 adapt

-0.61

POSITIVE LOGITS

 Afterwards

1.00

 Investigators

1.00

 During

0.91

 According

0.90

 Witnesses

0.89

 Additionally

0.87

 Apparently

0.85

 However

0.84

 Meanwhile

0.83

 Moments

0.83

Activations Density 10.563%

violent and disturbing actions or events involving people and law enforcement officers

No Comments

No Known Activations