INDEX

Explanations

the word "mistake"

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

Personendaten

-1.23

 '\\;'

-1.08

WriteTagHelper

-1.01

 CURIAM

-0.93

LookAnd

-0.92

ंदीखरीदारी

-0.91

PHeader

-0.91

abestanden

-0.90

 AssemblyCompany

-0.88

fjspx

-0.88

POSITIVE LOGITS

 Mistakes

1.09

Mistake

1.06

 Mistake

1.00

 mistakes

0.98

 mistake

0.94

Mist

0.68

 MIST

0.66

mist

0.64

like

0.63

б

0.61

Activations Density 0.004%