INDEX

Explanations

officials reporting on political and military events, specifically related to warnings, confirmations, details, and intelligence assessments

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Interstitial

-0.72

Govern

-0.70

 steril

-0.68

 landfall

-0.67

 Timbers

-0.66

 totality

-0.66

Tour

-0.66

ĸļ

-0.65

 unconstitutional

-0.65

incarn

-0.64

POSITIVE LOGITS

 briefed

0.94

raq

0.84

 knowledgeable

0.73

zinski

0.73

 familiar

0.72

 interviewed

0.71

 told

0.70

 priv

0.69

¿½

0.69

 sergeant

0.69

Activations Density 8.584%

officials reporting on political and military events, specifically related to warnings, confirmations, details, and intelligence assessments

No Comments

No Known Activations