INDEX

Explanations

titles of works, terms

The neuron strongly activates on standalone title or heading tokens—i.e. the names of works, products, sections, or list entries (proper‐noun or title words).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

==='

-1.21

 końca

-1.21

راقي

-1.19

peration

-1.16

姈

-1.13

ógico

-1.13

amkeit

-1.13

 nhiệm

-1.11

:";

-1.11

 Botany

-1.09

POSITIVE LOGITS

of

1.41

 about

1.22

 первая

1.19

 Thus

1.11

 گیرد

1.09

 ию

1.05

tudiant

1.05

1.04

 Екатери

1.04

то

1.02

Activations Density 0.015%