INDEX

Explanations

window in code

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 pile

1.37

 plays

1.32

 Pile

1.32

 younger

1.30

 play

1.26

 young

1.25

dip

1.22

涌

1.17

 Play

1.15

Thumbnail

1.14

POSITIVE LOGITS

 DIRECTIONS

1.72

ichlor

1.69

 Instructions

1.68

 วิธี

1.56

plementation

1.55

 instructions

1.55

 Eddington

1.54

 sConfig

1.53

érences

1.53

 срока

1.52

Activations Density 0.001%

No Known Activations