INDEX

Explanations

The neuron activates on tokens representing floating‐point numeric values (i.e. numbers with decimal points).

pay-per-view, Equestrian Festival, Hydraulically, Hydroview, Sir Lambton

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

𝒙

-1.40

 still

-1.13

җ

-1.13

蠅

-1.09

 Censo

-1.09

;%%

-1.05

 Diputados

-1.01

还需要

-1.01

";

-1.01

POSITIVE LOGITS

𝑞

1.36

 орден

1.32

roba

1.16

 nessun

1.15

盉

1.13

 fizer

1.11

饪

1.11

Preparazione

1.10

tocks

1.10

but

1.09

Activations Density 0.000%