INDEX

Explanations

entity name or place

The neuron strongly activates on multi-syllable compound nouns or named‐entity terms (e.g. “Body of Christ,” “pole vault,” “Valley Forge,” “beta-lactamase,” “pinstriping,” etc.).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

you

-1.70

 after

-1.69

 that

-1.66

 they

-1.53

it

-1.47

 selepas

-1.41

any

-1.34

 любом

-1.29

 before

-1.28

 with

-1.23

POSITIVE LOGITS

 metu

1.59

was

1.55

 would

1.40



1.33

想了想

1.32

を使用

1.30



1.30

 fotográfica

1.29

 توسط

1.28

équilibre

1.28

Activations Density 0.072%