INDEX

Explanations

accuracy

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Accurate

-1.63

 accurate

-1.61

accurate

-1.54

Accurate

-1.54

 accurately

-1.38

 accuracy

-1.34

 Accuracy

-1.27

 ACCUR

-1.21

accuracy

-1.17

Accuracy

-1.13

POSITIVE LOGITS

LEncoder

0.46

iastes

0.46

row

0.43

ino

0.40

vay

0.40

:+:

0.40

omod

0.39

σμα

0.39

PCell

0.39

 Munk

0.38

Activations Density 0.070%