INDEX

Explanations

durations and time limits

This neuron activates on numeric tokens, detecting numbers and quantitative measurements in the text.

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

玙

-1.32

 convite

-1.19

 espelho

-1.16

 estiver

-1.16

 sítio

-1.16

秶

-1.16

埽

-1.15

 meilleure

-1.13

瑆

-1.13

 braço

-1.13

POSITIVE LOGITS

or

2.27

 europeos

1.36

になったら

1.23

 Furthermore

1.14

 efectivos

1.12

 alemanes

1.06

 उन्ह

1.05

 Darauf

1.05

 melindungi

1.04

 after

1.03

Activations Density 0.061%