INDEX

Explanations

months and years

The neuron activates on calendar date expressions, especially month + year 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

 बातों

0.51

 вверх

0.50

観光

0.48

тическая

0.48

щите

0.48

 Lung

0.47

たくさん

0.47

水中

0.46

 hydroph

0.46

then

0.46

POSITIVE LOGITS

 지난해

0.86

去年

0.83

 पिछले

0.76

 গত

0.74

昨年

0.73

 گزشتہ

0.72

今年

0.72

今年の

0.69

지난

0.68

 கடந்த

0.63

Activations Density 0.011%