INDEX

Explanations

Python module setup

The neuron selectively fires on square‐bracket tokens that open array or list literals (e.g. the “\[” starting styleUrls, plugins, dependencies, srcs, rules, 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

 berätt

-1.42

醞

-1.40

違った

-1.31

 trouxe

-1.28

 fazia

-1.26

 hiasan

-1.23

。<

-1.22

árias

-1.20

附

-1.20

 tandis

-1.18

POSITIVE LOGITS

and

2.45

at

1.31

素晴らしい

1.30

 including

1.29

wolle

1.28

1.27

 ۱۴۰۰

1.27

fikasi

1.26

 okol

1.25

 such

1.24

Activations Density 0.010%