INDEX

Explanations

a/the followed by a concept

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

 நான்கு

1.33

vertex

1.32

encement

1.29

 bahwa

1.27

 ஐந்து

1.25

 хотя

1.24

 kerana

1.23

 пункт

1.23

 சுமார்

1.21

 epiz

1.20

POSITIVE LOGITS

 could

1.30

 might

1.23

 deems

1.23

 tripped

1.21

 humans

1.21

 experts

1.20

could

1.18

 experienced

1.16

 COULD

1.16

 deserves

1.15

Activations Density 0.179%