INDEX

Explanations

correctness and correction

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

楽しめる

0.64

 محبوب

0.63

 negotiable

0.60

可能性がある

0.60

軼

0.60

ienz

0.59

avip

0.58

rable

0.58

 runnable

0.57

实用

0.57

POSITIVE LOGITS

 correct

5.17

correct

4.88

Correct

4.70

 Correct

4.68

 correctly

4.60

 CORRECT

4.45

正确的

4.39

 correcto

4.31

正确

4.29

 correcta

4.26

Activations Density 1.055%