INDEX

Explanations

of

This neuron detects abstract nominalizations—formal nouns ending in “-ion” or “-ment” that name processes, actions, or concepts.

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

 প্রশংসা

0.66

Sab

0.66

 বুঝিতে

0.64

水を

0.63

せて

0.62

Attached

0.62

 পৌঁছে

0.61

inserted

0.60

opened

0.60

ऊंगा

0.60

POSITIVE LOGITS

of

1.39

 ofthe

1.19

 των

1.14

 شدن

1.11

 τῶν

1.05

 của

1.02

 thereof

1.02

 của

1.02

 hereof

0.96

នូវ

0.91

Activations Density 0.514%