INDEX

Explanations

transforming into

This neuron detects verbs (like “transform,” “turn,” “converted,” etc.) that express changing one thing into another (often in an “X … into Y” construction).

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

 lakini

-1.62

 JAPAN

-1.58

 Houſe

-1.57

还不错

-1.57

teilen

-1.57

TÉ

-1.56

 parfüm

-1.56

ších

-1.55

྅

-1.52

 FRENCH

-1.47

POSITIVE LOGITS

2.23

by

2.05

 Some

1.95

 There

1.95

 What

1.84

 would

1.83

an

1.72

 became

1.68

 didn

1.68

 wouldn

1.59

Activations Density 0.024%