INDEX

Explanations

conclusions and results

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

ात

2.22

ッパー

2.16

 पात्र

2.04

 kotak

2.03

氓

2.01

nent

2.00

 Zend

1.97

 Señor

1.94

ţia

1.94

arry

1.94

POSITIVE LOGITS

謐

1.96

"-

1.84

">(</

1.74

ੱਸ

1.68

1.66

taskList

1.66

wake

1.60

("'

1.59

 операция

1.58

্তি

1.56

Activations Density 0.688%