INDEX

Explanations

DeepMind

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

onuclease

1.44

 tunggu

1.29

mesine

1.29

EDITORIAL

1.29

heran

1.27

 Dislocations

1.26

alan

1.25

 reprint

1.23

جهه

1.22

 Thumbnail

1.21

POSITIVE LOGITS

sière

1.40

χρι

1.40

灶

1.36

umers

1.32

 concentric

1.28

アンド

1.27

 streetlight

1.27

μία

1.27

ause

1.26

合わせた

1.25

Activations Density 0.002%

No Known Activations