INDEX

Explanations

explaining how things work

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

殚

1.96

kében

1.95

 kişiler

1.88

 대표

1.88

ądź

1.87

价

1.86

 市場

1.85

及

1.84

銷售

1.84

ümer

1.84

POSITIVE LOGITS

 that

1.65

1.41

 checking

1.38

 when

1.36

 allowing

1.33

.??

1.26

 because

1.26

 visualizing

1.24

 linearization

1.22

 using

1.22

Activations Density 0.298%

No Known Activations