INDEX

Explanations

posted by

This neuron detects HTML level-2 section headers (the “<h2>…</h2>” tags).

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

-1.85

你的

-1.77

懨

-1.77

 gezet

-1.77

What

-1.77

よくある

-1.77

渽

-1.73

 onsdag

-1.69

To

-1.64

 durchaus

-1.61

POSITIVE LOGITS

 Eventually

1.91

 Their

1.89

 From

1.88

 Importantly

1.77

 They

1.76

攆

1.74

 intér

1.74

Ᏺ

1.72

 начинают

1.72

are

1.72

Activations Density 0.001%