INDEX

Explanations

verse numbers

The neuron strongly activates on tokens that contain digits or numeric sequences—e.g. section or error numbers, years, model names, domain names with numbers, 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

 polypes

0.41

 rejuvenating

0.40

 Obamacare

0.40

 NADPH

0.40

 Taehyung

0.40

 Jungkook

0.39

 polvo

0.39

 paralysie

0.38

 pierres

0.38

 EBITDA

0.38

POSITIVE LOGITS

Data

0.50

Value

0.44

0.42

Path

0.42

Field

0.42

Base

0.41

Info

0.41

ID

0.40

id

0.39

Activations Density 3.316%