INDEX

Explanations

moment

The neuron is looking for common multi-word function-word patterns that express partitive or quantitative relationships—e.g. phrases like “we have a …,” “moment of …,” or “so many of ….”

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

也

-2.25

.”

-2.16

 geprü

-1.77

 buone

-1.72

就

-1.64

只

-1.58

 zugel

-1.56

 Versuche

-1.56

 basierend

-1.55

 стоко

-1.51

POSITIVE LOGITS

 moments

1.70

诋

1.65

 Moment

1.57

艄

1.52

 MOMENT

1.50

𝄆

1.46

骘

1.46

1.41

၃

1.39

 another

1.38

Activations Density 0.021%