INDEX

Explanations

completeness or fullness

The neuron strongly fires on words expressing totality or completeness—especially “full,” “complete,” and their cognates.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

1.41

1.20

rit

1.18

Isn

1.16

きた

1.16

daki

1.16

dB

1.15

ro

1.13

are

1.12

으로는

1.12

POSITIVE LOGITS

 blown

1.30

тни

1.29

 complet

1.28

 promul

1.27

blown

1.24

 pleinement

1.23

 كامل

1.23

1.22

 completa

1.21

 জুড়ে

1.21

Activations Density 0.773%