INDEX

Explanations

patterns of gradual progress

The neuron consistently lights up on adverbial (–ly) or manner‐describing words (e.g. “consistently,” “rapidly,” “smoothly,” “gently,” “impartially,” “succinctly,” “informally”).

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

Би

0.42

 στη

0.41

َات

0.40

 Би

0.40

ਸ

0.39

 στην

0.39

 आविष्कार

0.38

 tejto

0.38

ن

0.38

ってます

0.36

POSITIVE LOGITS

uft

0.40

也就是说

0.39

စေ

0.38

มาชิก

0.38

PROTOCOL

0.38

৭০

0.38

 pudi

0.37

 wijze

0.37

INIT

0.36

 учетом

0.35

Activations Density 0.000%