INDEX

Explanations

questions and conditionals

The neuron fires on numeric tokens and measurements (e.g. decimal numbers and quantities).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 السياس

-1.77

 hebt

-1.45

 saumon

-1.41

ถ้า

-1.38

หลัง

-1.38

 ราคา

-1.38

kız

-1.34

บริการ

-1.32

 تساعد

-1.30

เปิด

-1.30

POSITIVE LOGITS

1.45

 vectra

1.41

 jetta

1.37

 tfsi

1.35

 简介

1.30

begingroup

1.30

 mondeo

1.29

 also

1.27

 tiguan

1.27

ご理解

1.23

Activations Density 0.073%