INDEX

Explanations

dy- prefix words

The neuron fires whenever it sees the two-letter sequence “dy” as a token (e.g. in “dy” itself or as the prefix of words like dystrophic, dystrophin, 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

which

-3.27

ing

-2.95

 incluso

-2.41

pecabe

-2.30

 mismos

-2.27

不论

-2.27

of

-2.25

 zweite

-2.23

ong

-2.23

 comprising

-2.23

POSITIVE LOGITS

2.89

邊

2.63

寬

2.48

2.42

在這

2.41

 sarcastic

2.38

 écoute

2.34

kaç

2.34

 државе

2.33

 Because

2.30

Activations Density 0.019%