INDEX

Explanations

This neuron fires on relatively rare, content-rich tokens—especially proper names, technical terms, and other low-frequency words.

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

விக்கும்

0.56

 kanssa

0.51

सात

0.49

chsler

0.49

ין

0.49

φό

0.48

nier

0.46

мещение

0.46

ವರೆ

0.46

পাতের

0.46

POSITIVE LOGITS

by

4.38

 oleh

3.84

by

3.53

By

3.30

โดย

3.22

 توسط

3.20

By

3.08

BY

2.67

 द्वारा

2.64

によって

2.61

Activations Density 0.651%