INDEX

Explanations

verbs of understanding or management

The neuron selectively fires on longer, content‐bearing words (not on short function words), effectively picking out semantically rich tokens.

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

現貨

-1.54

 derivados

-1.42

 podía

-1.41

 pureza

-1.38

 intertwined

-1.36

 then

-1.35

 sustancias

-1.34

their

-1.33

 trasero

-1.32

 ampio

-1.32

POSITIVE LOGITS

of

1.35

徨

1.35

৬

1.34

1.31

words

1.29

тику

1.29

 ettei

1.26

สอง

1.24

 письмо

1.24

and

1.22

Activations Density 0.117%