INDEX

Explanations

also known as

The neuron fires most strongly on sentence‐initial transition words or adverbial discourse markers (e.g. “Also,” “Interesting,” “Known”).

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

 그래도

1.17

それでも

1.12

ки

1.10

이었

1.08

Herkese

1.03

ᴋ

1.03

reek

1.02

age

1.02

lessly

1.02

しっかりと

1.00

POSITIVE LOGITS

 кстати

1.37

 כן

1.30

 불구하고

1.26

 âmbito

1.22

ments

1.11

 disponible

1.10

 বটে

1.10

 constaté

1.10

 conocido

1.07

 connu

1.07

Activations Density 0.053%