INDEX

Explanations

first, then

This neuron detects common discourse markers and function words used to introduce or sequence statements (e.g. “let me first say that,” “first,” “to,” “say”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 също

1.09

 myös

1.02

 również

0.99

 також

0.97

均为

0.94

 также

0.91

 też

0.91

 مختلفة

0.89

 सारे

0.86

 גם

0.86

POSITIVE LOGITS

Firstly

1.63

 먼저

1.56

 Firstly

1.49

最初に

1.49

First

1.41

我們先

1.41

 First

1.38

 முதலில்

1.38

まず

1.35

 まず

1.33

Activations Density 1.006%