INDEX

Explanations

scenario description generation

The neuron strongly responds to standalone topic words or section‐header–style nouns (often sentence‐initial) that signal the start of a new paragraph or section.

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

 можуть

0.49

 могат

0.49

 possono

0.48

 이러한

0.46

 maaaring

0.45

 môžu

0.44

 mohou

0.44

 peuvent

0.43

 могут

0.43

POSITIVE LOGITS

 appartement

0.59

 tastefully

0.58

 maravilh

0.55

 bungal

0.54

 culmination

0.52

 Altogether

0.52

👍

0.51

 passeio

0.51

 extravaganza

0.51

 😍

0.50

Activations Density 0.142%