INDEX

Explanations

progression and characteristics

This neuron is tuned to spot technical or quantitative content—particularly numbers, measurements, and domain‐specific/scientific terminology.

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

Elaboración

-1.88

 insuffisamment

-1.49

',...

-1.45

 melahirkan

-1.43

Duración

-1.39

ţ

-1.38

Inconvénients

-1.34

 mengakui

-1.32

 menangkap

-1.30

Etimología

-1.30

POSITIVE LOGITS

じゃなくて

1.50

its

1.38

ньому

1.30

不断

1.29

 完了

1.28

):

1.27

of

1.26

inderella

1.25

 последние

1.23

Previews

1.23

Activations Density 0.275%