INDEX

Explanations

helps and assists

predicts help or improvement

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

 hates

0.46

 sufrió

0.44

 Worse

0.43

失去了

0.43

使用的是

0.43

陥

0.42

 considère

0.42

 thinks

0.42

 complains

0.42

 stubbornly

0.41

POSITIVE LOGITS

 helps

2.39

 giúp

2.19

 помогает

2.16

helps

2.13

 помогают

2.06

 membantu

2.05

 aiuta

2.05

 Helps

2.02

ช่วย

1.96

 يساعد

1.96

Activations Density 0.778%