INDEX

Explanations

not the only

This neuron responds to prominent content words—especially nouns and adjectives carrying the main topic or focus of a sentence—while ignoring most function words.

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

 actually

-1.64

 really

-1.59

 only

-1.55

 truly

-1.52

 even

-1.51

 while

-1.47

 until

-1.40

 certamen

-1.38

 evaluation

-1.37

 first

-1.34

POSITIVE LOGITS

 será

1.29

пора

1.22

 Başkanı

1.20

 zult

1.20

 sommige

1.19

meida

1.16

 sguardo

1.16

Rhône

1.15

 ロング

1.15

 blijkt

1.13

Activations Density 0.018%