INDEX

Explanations

4K videos, new quality

The neuron fires on standalone numeric and related punctuation tokens (e.g. figures, dates, counts, and hyphens in stats).

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

 that

-2.05

but

-1.61

 what

-1.52

 sometimes

-1.43

 there

-1.42

How

-1.40

 here

-1.38

so

-1.32

 problems

-1.31

 think

-1.27

POSITIVE LOGITS

 dahil

1.45

 スリーブ

1.43

凄く

1.40

 nimmer

1.38

 streng

1.32

Presse

1.25

 faciliter

1.25

 APPELL

1.23

 オイル

1.22

 PLAINT

1.22

Activations Density 0.033%