INDEX

Explanations

The neuron is primarily detecting numeric expressions (tokens representing numbers or numeric measurements).

suggestions and opinions

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 during

-1.61

our

-1.48

 where

-1.41

 through

-1.30

 both

-1.30

 when

-1.28

 with

-1.27

 those

-1.23

 these

-1.22

 while

-1.18

POSITIVE LOGITS

 charmant

1.41

ю

1.24

 assez

1.23

с

1.21

 REALLY

1.19

«

1.17

 SaaS

1.16

 SECOND

1.13

жете

1.13

Activations Density 0.010%