INDEX

Explanations

"first"

This neuron detects mentions of inaugural occurrences, especially phrases like “the first time” describing a first-ever event.

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

if

0.66

）

0.62

 就是

0.61

lebih

0.60

smaller

0.60

 }{}_{\

0.60

 আরও

0.58

helpful

0.58

 jest

0.57

is

0.57

POSITIVE LOGITS

 EVER

1.05

本格

0.99

 ever

0.98

真正

0.93

 überhaupt

0.88

ऐसा

0.86

 ooit

0.83

 réellement

0.83

真正的

0.80

 такого

0.79

Activations Density 0.209%