INDEX

Explanations

The neuron fires on isolated alphanumeric fragments—especially numbers, units, single‐letter codes, and chemical-formula tokens common in technical or patent text.

whiskers

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

 бампер

-2.30

ּוֹ

-1.75

ִּ

-1.65

־ה

-1.53

 Σα

-1.48

ְּ

-1.47

 सत्यापित

-1.41

וּ

-1.38

erobic

-1.37

ַּ

-1.36

POSITIVE LOGITS

ׇ

1.62

rafat

1.52

 ویتامین

1.41

 растение

1.36

"]:

1.35

or

1.34

'];

1.34

 блюдо

1.33

licante

1.30

 primeiras

1.29

Activations Density 0.010%