INDEX

Explanations

with the help of

This neuron detects acknowledgments of assistance, especially the “with the help of” (or “aid of”) pattern.

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

雳

-1.73

 первая

-1.65

’

-1.63

ってみる

-1.53

there

-1.53

軾

-1.49

縋

-1.49

,:);

-1.49

because

-1.47

↵

-1.44

POSITIVE LOGITS

୩

1.86

亹

1.80

也要

1.78

 decidedly

1.78

ভ

1.77

 изменить

1.77

Of

1.75

烀

1.71

 deliciously

1.70

 بواسطة

1.67

Activations Density 0.013%