INDEX

Explanations

as in examples

The neuron fires on numeric and time‐related tokens—i.e. numbers (like “105.5,” “4”) and temporal markers/phrases (e.g. “upon,” “moment,” “ago”).

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

 What

-1.34

iramente

-1.33

to

-1.30

鏽

-1.27

as

-1.24

out

-1.23

 něj

-1.18

 sowohl

-1.17

さまざま

-1.13

ște

-1.12

POSITIVE LOGITS

 they

1.61

Pratique

1.32

 потом

1.31

៍

1.23

we

1.22

 имеет

1.20

Sonstiges

1.20

dames

1.20

Vidu

1.20

 when

1.20

Activations Density 0.035%