INDEX

Explanations

tolerances and usual states

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

を

-1.25

<bos>

-1.17

zielle

-1.14

 тільки

-1.12

 только

-1.08

 تنها

-1.05

久保

-1.04

vénients

-1.03

-1.01

それにしても

-0.96

POSITIVE LOGITS

 same

2.27

 desired

1.46

 required

1.27

 appropriate

1.22

 default

1.20

 correct

1.16

same

1.14

 exact

1.12

 recommended

1.08

 usual

1.05

Activations Density 0.124%