INDEX

Explanations

write a

The neuron fires on words denoting the act of writing (e.g., “write,” “wrote,” “writing”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

ᱽ

1.52

да

1.42

 Newly

1.34

ز

1.34

entially

1.32

くる

1.30

av

1.30

ญี่ปุ่น

1.26

近く

1.23

 새로

1.23

POSITIVE LOGITS

 cursive

1.48

লেখ

1.43

 essay

1.43

✍

1.42

 writing

1.30

 write

1.23

 pediatrician

1.23

 journaling

1.21

 journalism

1.20

 calligraphy

1.20

Activations Density 0.090%