INDEX

Explanations

reflecting on

The neuron detects instances of reflective language—especially forms of the verb “reflect” (reflects, reflecting, reflection, reflected) and the accompanying “on.”

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

-2.30

</h1>

-2.23

就跟

-2.19

 idyllic

-2.06

 Another

-2.05

 será

-2.03

are

-2.02

 Here

-1.97

 enfrentan

-1.97

 Either

-1.95

POSITIVE LOGITS

 排骨

2.25

趼

1.97

 插画

1.85

亖

1.84

這

1.75

岘

1.75

ᜀ

1.70

錞

1.70

????????

1.70

なんだ

1.70

Activations Density 0.008%