INDEX

Explanations

mixture and blend

The neuron activates most strongly on words that denote a combination or collection of components—e.g. “mixture,” “blend,” or “basket.”

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

in

-1.53

 Instead

-1.52

and

-1.49

 Even

-1.47

For

-1.46

if

-1.44

In

-1.44

',{

-1.43

waran

-1.43

POSITIVE LOGITS

of

2.22

 keinem

1.85

 autrefois

1.69

 ternura

1.68

tanto

1.63

昔の

1.55

 öffnungszeiten

1.55

 manifestó

1.49

 Deiner

1.48

 refirió

1.47

Activations Density 0.083%