INDEX

Explanations

processes or events

The neuron fires strongly on words that name actions or procedures—i.e. process-denoting nouns (like “operation,” “diagnosis,” “filling,” “printing,” “rendering,” “use,” etc.).

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

 never

-1.77

not

-1.61

in

-1.32

 hogyan

-1.32

-1.31

 than

-1.30

 получает

-1.29

 then

-1.26

don

-1.25

 front

-1.25

POSITIVE LOGITS

︱

1.70

谞

1.51

ledad

1.50

豋

1.47

 ligar

1.45

 perché

1.43

 your

1.41

från

1.41

⁓

1.41

Activations Density 0.152%