INDEX

Explanations

relative clauses describing people or entities

This neuron detects countable noun heads that follow numeric or quantitative expressions (e.g. “people,” “features,” “sleepers,” “factors,” “players”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 peut

0.42

可能是

0.39

0.38

或者是

0.38

খানে

0.38

是一种

0.37

是一个

0.37

 Goods

0.36

 อาจ

0.36

 არის

0.36

POSITIVE LOGITS

ที่มี

0.55

 الذين

0.53

 जिन्होंने

0.51

 που

0.49

 whose

0.48

who

0.47

 that

0.46

 jotka

0.46

 którzy

0.45

 которые

0.45

Activations Density 0.001%