INDEX

Explanations

author names

The neuron strongly activates on isolated uppercase letters and periods as they appear in scholarly reference and author–initial patterns (e.g. “R.”, “T.”, “D.”).

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

 silueta

-1.29

 balcón

-1.27

駅から

-1.27

 dormitorio

-1.23

随即

-1.23

 falda

-1.20

 then

-1.20

ERGY

-1.19

 счастли

-1.15

 carcasa

-1.15

POSITIVE LOGITS

Seguir

1.52

Despite

1.49

 незавер

1.38

這

1.35

 чесно

1.31

 роль

1.31

Once

1.30

Significado

1.30

Nasc

1.29

 Wartawan

1.27

Activations Density 0.006%