INDEX

Explanations

more than

The neuron is triggered by past‐participle verb forms (e.g. “granted,” “anticipated,” “realized”) and similar adjective‐like constructions.

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

罍

-1.25

 ángel

-1.12

峮

-1.11

だけではなく

-1.06

 also

-1.05

フード

-1.04

だけでなく

-1.04

 assister

-1.03

 kontak

-1.02

乁

-1.02

POSITIVE LOGITS

any

1.82

 than

1.77

 usual

1.42

 before

1.27

 ever

1.25

or

1.21

 possible

1.20

breviations

1.20

ksiä

1.15

 sonst

1.12

Activations Density 0.018%