INDEX

Explanations

on a

This neuron detects the phrase pattern “on a / on an ” followed by a noun or noun phrase indicating being engaged in some activity or state.

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

lerini

-1.51

出来た

-1.48

of

-1.46

締

-1.42

 tradicionais

-1.37

 "/";

-1.33

 invitamos

-1.33

滯

-1.30

feste

-1.30

 nacionais

-1.30

POSITIVE LOGITS

')

1.82

"]

1.60

 něj

1.55

"];

1.52

 vooral

1.50

']

1.48

 takich

1.47

 explosión

1.45

"),

1.44

reviewed

1.44

Activations Density 0.019%