INDEX

Explanations

actions and their possessions

The neuron fires on dynamic sports‐reporting verbs and related tokens that describe game events or outcomes (e.g. “stunned,” “snapped,” “pummeled,” “won,” “lost,” “trailed,” “improved,” “broke through”).

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

 your

-1.70

 himself

-1.66

or

-1.55

ждает

-1.47

 vaše

-1.45

Even

-1.45

 there

-1.45

 because

-1.40

bản

-1.38

 beliau

-1.37

POSITIVE LOGITS

 themselves

1.79

 själva

1.51

its

1.48

 ihrer

1.38

 ihrem

1.35

 varandra

1.28

そろ

1.27

 ihren

1.26

を用意

1.23

 każdy

1.19

Activations Density 0.138%