INDEX

Explanations

The neuron activates on numeric tokens and quantities—i.e. any explicit numbers, measurements, dates, or counts in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 असते

0.61

 থাকে

0.57

 असतो

0.55

 હોય

0.53

 бывает

0.53

 нравится

0.52

 असतात

0.52

 essere

0.52

 usually

0.51

 ہوتا

0.51

POSITIVE LOGITS

 became

1.55

 went

1.46

 took

1.45

became

1.38

 gave

1.38

 began

1.38

 undertook

1.27

took

1.23

 withdrew

1.20

 drove

1.20

Activations Density 2.848%