INDEX

Explanations

names with suffixes or numbers

The neuron fires on capitalized multi‐word named entities (proper names of people, organizations, events, etc.).

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

the

-1.77

他の

-1.38

 another

-1.36

莚

-1.35

 ſte

-1.35

໊

-1.34

 ovvero

-1.33

Such

-1.31

 ſtre

-1.31

например

-1.30

POSITIVE LOGITS

 only

1.88

 when

1.52

鵬

1.48

all

1.48

 quando

1.41

 cuando

1.35

how

1.33

 عندما

1.31

 nasıl

1.30

 after

1.28

Activations Density 0.355%