INDEX

Explanations

for the purposes of

The neuron activates on words ending in “-ing” (i.e., gerunds or present participles).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

issauga

-1.30

描く

-1.28

 reconoce

-1.24

 zwölf

-1.22

小结

-1.21

 spearheaded

-1.20

辈子

-1.18

 wyja

-1.16

凭借

-1.15

 lauded

-1.14

POSITIVE LOGITS

 retten

1.18

 tertentu

1.18

or

1.13

conocimiento

1.11

ری

1.05

 februari

1.05

 kalt

1.05

 alkoh

1.04

ɲ

1.02

 their

1.02

Activations Density 0.106%