INDEX

Explanations

each, different, each one

The neuron primarily detects numeric tokens (especially multi‐digit numbers and decimal values).

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

ор

0.61

ủy

0.60

ु

0.58

曹

0.57

水

0.57

ы

0.56

0.55

ift

0.54

ু

0.54

ip

0.53

POSITIVE LOGITS

 masing

1.05

それぞれの

1.02

 तीनों

1.01

 Each

0.98

 jeweils

0.96

 각각

0.95

それぞれ

0.91

 semuanya

0.90

 drei

0.89

 каждой

0.88

Activations Density 0.084%