INDEX

Explanations

role for or place for

This neuron picks up on question‐style wording of “place/role/function/room for …,” i.e. phrases asking whether there is a place (or role/function/room) for something in some context.

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

祧

-1.77

俢

-1.65

鹇

-1.57

odils

-1.52

 citazioni

-1.52

洐

-1.52

橚

-1.50

 fallu

-1.48

 soigneusement

-1.48

 lentement

-1.47

POSITIVE LOGITS

in

1.93

</h4>

1.64

1.45

1.44

1.41

</u>

1.38

</strong>

1.37

1.35

be

1.33

Activations Density 0.038%