INDEX

Explanations

any followed by relationship or activity

The neuron is primarily firing on generalizing “quantifier” words—indefinite determiners and modifiers like “any”, “another”, “additional” (and the occasional related content word)—that mark broad, catch-all references.

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

 tainted

1.40

 thighs

1.25

 hashlib

1.23

 histories

1.18

 bucks

1.18

ם

1.17

 hyperparameters

1.16

 THINGS

1.14

stk

1.10

 somatic

1.09

POSITIVE LOGITS

 qualunque

1.53

place

1.48

 qualquer

1.48

 whatsoever

1.45

任何

1.38

 herhangi

1.37

 autre

1.34

ਰ

1.34

1.30

le

1.29

Activations Density 0.152%