INDEX

Explanations

types of questions

This neuron fires on the key topical noun that heads a new section or paragraph (the single-word “headline” that labels what follows, e.g. photography, sample, donation, privacy, prize, meal, etc.).

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

 хотелось

0.55

 सगळे

0.48

 sesuai

0.47

 Them

0.46

สิ่งที่

0.46

 wiem

0.44

 Normalize

0.43

 noastră

0.43

 mamy

0.43

 બધા

0.42

POSITIVE LOGITS

或

0.46

 или

0.43

idene

0.39

 hoặc

0.39

िश्वत

0.39

 ऑफिसर

0.38

or

0.37

leges

0.37

 нови

0.37

 патри

0.37

Activations Density 0.427%