INDEX

Explanations

Taking thislow levelsannounced

data bias and instructions

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

多年

0.52

百年

0.50

多年的

0.49

世纪

0.48

ెండు

0.48

 modernize

0.48

 বললো

0.47

四个

0.46

义

0.46

鸡

0.46

POSITIVE LOGITS

 Kurds

0.48

 behaviours

0.44

विरुद्ध

0.44

 Soviet

0.43

$}}

0.42

 Demir

0.42

 louder

0.42

 shadows

0.41

 Seung

0.40

 Pala

0.40

Activations Density 0.001%