INDEX

Explanations

how well things work

This neuron is activated by occurrences of the root “work” (in forms like work, works, worked).

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

 اهم

-1.54

蜮

-1.52

㍉

-1.34

☻

-1.33

 navires

-1.29

 edin

-1.27

Geografía

-1.23

 جمعه

-1.22

ed

-1.20

 chercheurs

-1.20

POSITIVE LOGITS

);

1.63

by

1.55

 Também

1.48

 Talvez

1.44

 baffled

1.42

1.40

 таке

1.40

 Enquanto

1.39

):

1.38

 Isso

1.38

Activations Density 0.026%