INDEX

Explanations

first / last word in sequence

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

 weiterhin

1.51

continued

1.46

őség

1.44

continue

1.41

continues

1.38

 ebenfalls

1.38

还可以

1.38

也可以

1.35

 вскоре

1.34

preserved

1.34

POSITIVE LOGITS

least

2.31

 first

2.25

 least

2.21

 pertama

2.05

first

2.04

 last

1.97

last

1.95

FIRST

1.86

 eerste

1.83

 lowest

1.83

Activations Density 0.433%