INDEX

Explanations

days of the week, time, duration

The neuron fires on temporal‐reference tokens—words or numerals that specify when something happens (dates, times, “first time,” “last night,” “Sunday,” “earlier,” 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

 into

-1.41

has

-1.37

 pomocou

-1.27

tetur

-1.20

in

-1.16

 through

-1.16

 speziellen

-1.12

 svoju

-1.09

有可能

-1.07

샐

-1.07

POSITIVE LOGITS

gradable

1.35

 there

1.34

}=\

1.23

 these

1.20

AZIONI

1.20

 Italie

1.19

 этом

1.17

 они

1.16

)”

1.16

Mfg

1.16

Activations Density 0.143%