INDEX

Explanations

proper nouns and technical terms

The neuron strongly activates on multi‐syllable, capitalized words marking named entities—e.g. chemical/drug names, software titles, disease names, fictional character names, and algorithm names.

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

fasterxml

-1.41

 also

-1.33

Ꮸ

-1.32

</u>

-1.30

໊

-1.29

Faults

-1.25

 tiga

-1.22

 juga

-1.20

 arī

-1.17

lepší

-1.17

POSITIVE LOGITS

of

1.23

並不是

1.19

 الآن

1.18

應該是

1.09

的机会

1.05

though

1.05

的过程中

1.05

of

1.04

 возможно

1.03

ionados

1.02

Activations Density 0.365%