INDEX

Explanations

code structure

The neuron fires on actual code lines (indented source‐code statements) rather than plain text.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 those

-1.79

 like

-1.52

開設

-1.48

at

-1.45

 these

-1.44

を

-1.37

by

-1.36

？』

-1.35

 after

-1.33

 whose

-1.33

POSITIVE LOGITS

whenever

1.25

 aislados

1.24

 évent

1.24

 précédents

1.23

1.22

 profils

1.20

為に

1.19

verhalten

1.16

Pls

1.15

 gena

1.15

Activations Density 0.008%