INDEX

Explanations

list structure or data classification

The neuron fires strongly on rare or domain‐specific content words—i.e. it flags unusually infrequent terms.

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

 машины

0.89

 lackluster

0.85

берриз

0.81

 объем

0.80

ань

0.79

병

0.79

 anteriores

0.78

uirre

0.77

 мальчика

0.77

 Старки

0.77

POSITIVE LOGITS

માં

0.84

état

0.82

در

0.74

مرة

0.72

אה

0.72

 catégories

0.72

 dans

0.71

 stato

0.71

 adição

0.71

في

0.71

Activations Density 0.000%