INDEX

Explanations

minutes before

The main thing this neuron does is detect clock‐style time expressions (e.g. numeric times with colons and AM/PM).

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

.’’

0.67

,’’

0.67

殞

0.62

日付

0.61

’’

0.60

ecuzione

0.60

 exécut

0.59

 irrepar

0.59

 uitgevoerd

0.59

 executed

0.58

POSITIVE LOGITS

前の

0.69

 перед

0.68

前に

0.66

直し

0.63

 წინ

0.62

前的

0.61

 przed

0.60

Ön

0.60

 मांस

0.60

 Antes

0.59

Activations Density 0.015%