INDEX

Explanations

financial fund

The neuron strongly activates on uppercase acronyms or initialisms (multi-letter all-caps tokens).

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

zędu

-1.83

臯

-1.67

Still

-1.63

 trei

-1.59

itettura

-1.56

 there

-1.55

 piese

-1.55

Один

-1.50

又一次

-1.48

ﱠ

-1.46

POSITIVE LOGITS

↵↵

1.74

mohon

1.61

are

1.58

hidup

1.47

 sposob

1.38

 permiten

1.38

for

1.35

Ihm

1.34

 Strukt

1.33

lasyon

1.33

Activations Density 0.006%