INDEX

Explanations

class

The neuron fires on tokens typical of formal legal or contractual phrasing (e.g. “with respect to,” “limited,” “revocable,” “Overnight”).

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

 clerg

0.91

arabic

0.83

 ########.

0.81

 sharks

0.80

跑到

0.79

कान

0.78

 abroad

0.78

 batsmen

0.77

点了点头

0.76

 bolstered

0.76

POSITIVE LOGITS

 определя

0.73

 телефона

0.70

 извест

0.70

0.68

 изготовления

0.68

 днем

0.67

 может

0.65

 поне

0.65

 це

0.63

вые

0.63

Activations Density 0.001%