INDEX

Explanations

XML key-value pairs and code snippets

The neuron fires on markup‐style delimiters—characters like “<”, “</” in XML/HTML tags and “\” in TeX/LaTeX commands.

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

’,

-1.27

 this

-1.24

 that

-1.19

We

-1.05

’-

-1.01

 using

-0.98

 most

-0.98

„

-0.95

 commonly

-0.95

all

-0.93

POSITIVE LOGITS

了

1.30

휀

1.11

 Rezept

1.04

rrggbb

1.00

页面存档备份

0.98

artifactId

0.98

kać

0.97

θει

0.97

 unten

0.96

 relação

0.96

Activations Density 0.029%