INDEX

Explanations

energy levels and eigenvalues

The neuron detects mathematical expressions and LaTeX‐style formula tokens (numbers, variables, operators, and equation delimiters).

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

ensity

-0.72

 coefficient

-0.70

غة

-0.67

 Bode

-0.65

flexDirection

-0.65

Nbr

-0.65

Unlike

-0.64

онов

-0.64

 föres

-0.63

něž

-0.63

POSITIVE LOGITS

 energy

2.28

 energies

1.96

energy

1.94

Energy

1.91

 Energy

1.81

energies

1.48

 ENERGY

1.48

 energie

1.48

 energía

1.47

エネルギー

1.46

Activations Density 0.024%