INDEX

Explanations

key value pairs

This neuron detects the angle‐bracket characters (“<” and “</”) that mark XML/HTML tags.

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

ꉂ

-0.94

 manna

-0.86

を抑

-0.85

쩍

-0.85

Опыт

-0.82

 jobber

-0.82

 Superhost

-0.81

饗

-0.81

ilingual

-0.80

 truk

-0.80

POSITIVE LOGITS

![

0.98

itus

0.87

 Council

0.84

 latter

0.82

 ThemeData

0.81

oo

0.81

 claros

0.81

Ű

0.80

vyb

0.80

<![

0.80

Activations Density 0.019%