INDEX

Explanations

medical and historical suffixes

The neuron strongly activates on unusually long or rare multi‐character tokens—e.g. proper names, technical terms, long words, and multi‐syllabic tokens.

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

ја

0.83

 تھا

0.77

ونها

0.75

తుంది

0.75

 વચ્ચે

0.73

ون

0.73

 udah

0.72

 accolades

0.72

 واذا

0.72

 نیز

0.71

POSITIVE LOGITS

 của

1.06

of

1.04

 של

0.95

 της

0.91

 của

0.89

ของ

0.88

 Porsche

0.87

 Airbus

0.87

របស់

0.85

 ntawm

0.84

Activations Density 2.358%