INDEX

Explanations

extension method or class

The neuron is keyed to “extension”-style tokens—things like file‐extension abbreviations (e.g. “js,” “ext,” “php”) and the colon separators surrounding them.

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

܆

-2.39

}$

-2.39

васто

-2.27

횃

-2.11

謌

-2.09

柃

-2.06

菇

-2.00

 ispir

-2.00

ೣ

-1.97

 Infór

-1.96

POSITIVE LOGITS

}$.

2.61

 esimer

2.34

 extend

2.22

 desire

2.05

1.97

ens

1.96

\\

1.95

1.91

型的

1.90

 permissible

1.88

Activations Density 0.025%