INDEX

Explanations

second and third rankings

The neuron activates on ordinal number mentions (e.g. “first,” “second,” “third,” “fourth,” etc.).

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

-1.63

 respectivas

-1.42

to

-1.41

 When

-1.39

for

-1.39

The

-1.38

 While

-1.27

 either

-1.25

리를

-1.24

-1.23

POSITIVE LOGITS

郸

1.53

埛

1.49

ípio

1.49

 dám

1.48

 μετά

1.45

âmica

1.41

 haja

1.40

の使い方

1.39

 Lined

1.38

 могут

1.37

Activations Density 0.043%