INDEX

Explanations

abbreviations and identifiers

The neuron consistently activates on the two-letter uppercase token “TC,” flagging instances of that abbreviation (often in reference citations, section headings, or scientific abbreviations).

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

很

-2.67

 Their

-2.42

</strong>

-2.41

丨

-2.36

 neue

-2.34

原本

-2.31

 注意

-2.23

-2.22

 nimmer

-2.17

镦

-2.14

POSITIVE LOGITS



2.59

 nuestro

2.59

𖤣

2.58

翃

2.47

หน้าน

2.33

 Mühe

2.31

tc

2.22

⤿

2.17

the

2.16

 jedna

2.16

Activations Density 0.008%