INDEX

Explanations

chess grandmasters and championships

The neuron primarily detects numeric tokens (numbers and digit sequences).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

𝙨

-0.81

 contar

-0.77

抓住

-0.73

atorship

-0.71

 Fara

-0.70

câ

-0.69

福田

-0.69

ólicos

-0.67

 doPost

-0.67

ليب

-0.66

POSITIVE LOGITS

 Alpha

1.09

 chess

1.09

棋

1.02

Alpha

0.91

 checkers

0.86

Chess

0.86

围

0.85

 professional

0.84

chess

0.83

 Chess

0.82

Activations Density 0.047%