INDEX

Explanations

references to specific cutoff levels used in diagnostic tests

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 kasarigan

-0.71

Rüyada

-0.63

はじめに

-0.61

 nakalista

-0.59

Tracce

-0.57

теризу

-0.57

 alluminio

-0.56

CppCodeGen

-0.56

ArgsConstructor

-0.56

 piede

-0.55

POSITIVE LOGITS

 cutoff

2.86

cutoff

2.17

utoff

2.13

截止

0.79

بوابة

0.70

 threshold

0.68

threshold

0.59

 Perſ

0.59

orianCalendar

0.57

 qualifying

0.56

Activations Density 0.001%