INDEX

Explanations

index

The neuron fires on references to explicit positions or indices—i.e. when text names or numbers a specific position (e.g. “remove from an arbitrary but known position,” “position,” buffer ID, block number, line number).

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

 autocl

0.53

 kvalit

0.52

 kaks

0.52

 aggregating

0.52

 topping

0.51

सरकार

0.51

 пан

0.51

 সরকার

0.50

ringar

0.50

 テーマ

0.50

POSITIVE LOGITS

 index

2.52

index

2.45

 Index

2.38

Index

2.30

索引

2.28

 INDEX

2.25

INDEX

2.23

 indexes

2.19

 indices

2.17

 índice

2.17

Activations Density 0.048%