INDEX

Explanations

dates, times, and conjunctions

The neuron flags tokens that are part of date and time stamps (e.g. numeric month/day/year, weekdays, times).

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 with

-2.05

to

-1.60

 which

-1.55

 each

-1.48

and

-1.47

 地図

-1.37

 最終

-1.34

 pastillas

-1.31

 apartament

-1.31

 bawat

-1.30

POSITIVE LOGITS

 ponieważ

1.42

 في

1.41

 kiedy

1.38

 élas

1.34

vì

1.34

дву

1.33

 lebt

1.32

 gdyż

1.32

ཽ

1.27

 dlatego

1.25

Activations Density 0.006%