INDEX

Explanations

network development changes

This neuron selectively activates on capitalized tokens, such as proper names, titles, or uppercase words.

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

 katika

-1.51

 передний

-1.36

 church

-1.35

 lecker

-1.34

 задний

-1.33

 military

-1.29

NhanVien

-1.28

ֱ

-1.28

;\;

-1.27

 окружающей

-1.25

POSITIVE LOGITS

 filós

1.70

рѣ

1.52

ſa

1.48

 掃除

1.38

₀

1.35

 бого

1.34

pushd

1.33

 мона

1.32

֠

1.29

 funcionalidades

1.28

Activations Density 0.262%