INDEX

Explanations

MUST, important, NEED, order, skip, only

The neuron fires on directive language—words and phrases that issue instructions or requirements (e.g. “must,” “in order to,” “link … with,” “important to,” 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

 maybe

-1.39

 generally

-1.38

 måske

-1.37

ukų

-1.28

 might

-1.28

 easier

-1.23

.}(

-1.22

Peut

-1.18

 perhaps

-1.16

 może

-1.16

POSITIVE LOGITS

1.67

!!!

1.46

絶対に

1.45

!!)

1.41

!)

1.39

!!!)

1.35

!!!!!

1.31

!!

1.30

!!!!

1.30

!!!!!!

1.30

Activations Density 0.036%