INDEX

Explanations

This neuron detects words and phrases that introduce or refer to thematic categories or subjects (e.g. “topics,” “issues,” “related,” “relevant,” “range,” “matters,” “aspects”).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

for

-2.19

on

-1.77

 Since

-1.46

 Because

-1.43

 There

-1.41

 because

-1.34

as

-1.29

 from

-1.27

 noteworthy

-1.26

 herói

-1.26

POSITIVE LOGITS

usually

1.47

 such

1.47

 jabón

1.39

ごとの

1.31

 ciem

1.31

 takie

1.30

Could

1.29

iest

1.28

 этого

1.28

Ð

1.27

Activations Density 0.067%

related topics and matters