INDEX

Explanations

This neuron detects proper names and titles (capitalized named entities like people, places, or works).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ште

0.26

 정부

0.24

 режима

0.24

ୂ

0.24

阍

0.23

і

0.23

éről

0.23

ﺎ

0.23

 하지만

0.22

тім

0.22

POSITIVE LOGITS

0.36

0.30

 discriminating

0.25

 empat

0.25

 mermaid

0.25

 internship

0.25

 assortment

0.25

 floss

0.25

 ergonom

0.24

 dann

0.24

Activations Density 0.000%