INDEX

Explanations

direction or towards

The neuron fires on words and phrases that indicate movement or spatial direction (e.g., “going in the right direction,” “looking,” “turned in [a place]”).

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

 Crossing

0.62

texttt

0.56

 набор

0.56

žil

0.55

ریع

0.54

 балан

0.52

ADA

0.51

 peça

0.51

 Akademie

0.51

్ఞ

0.50

POSITIVE LOGITS

 direction

2.92

方向

2.89

 방향

2.81

的方向

2.75

Direction

2.72

direction

2.69

 Direction

2.66

 directions

2.58

 DIRECTION

2.48

方向に

2.48

Activations Density 0.439%