INDEX

Explanations

significant portion

This neuron responds to long, multisyllabic, formal vocabulary—technical or academic–style words typically found in analytical or scholarly prose.

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

ول

1.04

 nhấn

1.04

 வகை

1.03

 Gerade

0.96

 Quels

0.93

ो

0.92

imme

0.91

 lubricants

0.91

ﻥ

0.91

و

0.89

POSITIVE LOGITS

cie

1.27

이면

1.27

ları

1.23

zelf

1.23

ukasz

1.16

ください

1.15

お待ち

1.13

cussion

1.11

KCl

1.11

Activations Density 2.811%