INDEX

Explanations

people who use specific words

The neuron is a general “content‐word” detector, lighting up on salient topic nouns or domain‐specific terms (e.g. concrete subjects, proper nouns, specialized vocabulary).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 verur

-1.04

if

-1.04

 beeintr

-1.02

țiile

-0.97

 サム

-0.94

你自己

-0.93

ज्ञ

-0.91

חים

-0.91

ьере

-0.89

 Certifications

-0.88

POSITIVE LOGITS

who

1.92

 their

1.73

or

1.68

 themselves

1.36

themselves

1.28

 quienes

1.19

 их

1.18

are

1.15

who

1.14

 kteří

1.13

Activations Density 0.136%