INDEX

Explanations

matrix dimensions and code

The neuron detects explicit matrix‐dimension specifications (e.g. “4×4”, “3×3”).

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

互联网档案馆

-1.71

in

-1.57

had

-1.52

 Typically

-1.48

 would

-1.48

 Their

-1.45

 exquisite

-1.45

”？

-1.44

anyahu

-1.42

POSITIVE LOGITS

2.00

 they

1.77

of

1.70

 spokoj

1.54

点点

1.53

ſſen

1.52

 två

1.46

bigr

1.45

 alguns

1.42

神

1.41

Activations Density 0.002%