INDEX

Explanations

adverbs ending in `ly`

This neuron detects adverbial suffixes (like English “-ly,” French “-ment,” Greek “–ως,” etc.) that signal adverbs.

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

of

-2.48

咄

-2.16

 beetje

-2.03

 pentru

-2.03

 véritable

-2.03

 hanem

-2.02

忖

-1.95

𝟴

-1.95

ïe

-1.90

 heilig

-1.87

POSITIVE LOGITS

there

2.61

2.52

this

2.33

There

2.28

what

2.22

2.20

2.19

where

2.13

en

2.11

re

2.09

Activations Density 0.012%