INDEX

Explanations

components and physical objects

The neuron activates on rare, domain‐specific technical terms—especially chemical group names (like trimethylsilyl or desilylation) and specialized acronyms or jargon.

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

of

-1.89

 навіть

-1.77

by

-1.75

-1.61

two

-1.55

 three

-1.48

 even

-1.45

-\

-1.45

 five

-1.32

 nawet

-1.32

POSITIVE LOGITS

</code>

1.61

datable

1.46

 dificil

1.44

 XVIII

1.41

')

1.40

”]

1.33

立

1.30

度

1.30

ToMany

1.28

lecht

1.28

Activations Density 0.179%