INDEX

Explanations

cyclones and cyclins

The neuron fires on any word containing the “cycl-” root (e.g. cyclone, cyclones, cyclonic, cyclin).

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

).

-3.00

which

-2.80

including

-2.64

…….

-2.52

from

-2.48

-2.41

瀦

-2.39

 cuja

-2.38

……..

-2.25

of

-2.23

POSITIVE LOGITS

༛

2.55

𓋼

2.48

 большим

2.23

 amerikan

2.20

 comentar

2.19

 rezer

2.19

 извест

2.16

푢

2.14

 vemos

2.13

2.08

Activations Density 0.003%