INDEX

Explanations

`0` or `last` followed by things like `2` or `straw`

The neuron primarily responds to numeric tokens and measurement references (e.g. years, page numbers, dimensions, statistics).

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

 Banglades

1.13

 Ariana

1.11

 Arach

1.09

 Большая

1.03

 Thay

1.02

 Aleks

1.01

 Cáceres

1.01

 PAOK

1.00

Bxb

0.99

 Aleksandra

0.99

POSITIVE LOGITS

сно

0.75

оло

0.71

 brokerage

0.68

্বার

0.67

barred

0.66

thermal

0.64

 brakes

0.64

ኢ

0.64

ينية

0.63

 engines

0.62

Activations Density 0.000%