INDEX

Explanations

technical titles

This neuron detects metadata field names (like “title” or “author”) in document front‐matter.

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

вгений

-2.13

饜

-2.11

ギュア

-2.09

很喜欢

-2.06

 giapponese

-2.03

很赞

-2.02



-1.99

 enthou

-1.99



-1.96

 estadounid

-1.95

POSITIVE LOGITS

</i>

2.72

</b>

2.58

2.39

</h2>

2.34

2.27

</h1>

2.16

2.09

 more

2.02

<i>

2.02

1.98

Activations Density 0.006%