INDEX

Explanations

optional fields or lists

The neuron detects membership‐test operations (especially the “in”/“not in” operator and similar membership checks in code).

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

to

-1.55

and

-1.34

nicht

-1.23

への

-1.21

hadores

-1.19

if

-1.17

 рассказать

-1.17

 probleme

-1.13

 antal

-1.13

zett

-1.13

POSITIVE LOGITS

 húmedo

1.61

 ningún

1.58

 soportar

1.49

 cálido

1.48

 когда

1.45

 energías

1.45

 involucra

1.43

 поэтому

1.42

 ninguna

1.41

 והוא

1.41

Activations Density 0.010%