INDEX

Explanations

mathematical or programming contexts

The neuron fires on backslash-escaped sequences (e.g. LaTeX commands, code escapes, or markup tags) in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 Rehabilitation

0.48

Reddit

0.45

 Reddit

0.44

 Rehab

0.44

 reddit

0.42

Bd

0.42

 Paytm

0.42

 Gaur

0.41

 Faculty

0.41

 defunct

0.41

POSITIVE LOGITS

empuan

0.38

idane

0.37

łow

0.37

orem

0.36

longrightarrow

0.36

efficiencies

0.36

orel

0.36

ůj

0.36

ków

0.35

šno

0.35

Activations Density 0.000%