INDEX

Explanations

formal titles and respectful addressing terms like "sir," "captain," and "gentleman."

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

Offline

-0.56

 Clinic

-0.55

NetMessage

-0.55

 largeDownload

-0.52

ARK

-0.51

obyl

-0.50

 Workers

-0.50

 Injury

-0.50

 Victims

-0.50

 Puzzles

-0.49

POSITIVE LOGITS

sir

0.73

 gentleman

0.64

geant

0.61

 colonel

0.59

terday

0.58

xual

0.58

 gentlemen

0.57

mares

0.56

ority

0.55

 repe

0.53

Activations Density 10.014%

formal titles and respectful addressing terms like "sir," "captain," and "gentleman."

No Comments

No Known Activations