INDEX

Explanations

HTML tags and attributes

The neuron is a detector for “technical” or “markup” tokens—things like HTML/LaTeX commands (e.g. `<tt>`, `\begin{align*}`) and citation or bibliography markers (e.g. `[@CIT…]`, `[@bib…]`)—that typically appear in scholarly or formatted documents.

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

...”

-2.31

 mereka

-2.30

蕺

-2.23

他們

-2.22

</b>

-2.11

...?

-2.08

</h4>

-2.03

......”

-1.94

co

-1.88

 anguish

-1.87

POSITIVE LOGITS

).

2.27

2.17

chelsea

1.97

 dirigentes

1.96

 Another

1.91

งิน

1.90

🫅

1.88

饪

1.87

 bekannten

1.86

 paisajes

1.85

Activations Density 0.012%