INDEX

Explanations

let ... client, ground

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

 vielfält

1.46

 choć

1.46

 muita

1.42

 odborn

1.41

 katk

1.39

 gibt

1.39

 particolarmente

1.35

 doskona

1.35

σετε

1.33

 rekan

1.33

POSITIVE LOGITS

OLS

1.16

 seeming

1.14

spreadsheet

1.14

 apparently

1.13

apparently

1.13

 presumably

1.11

 evaded

1.10

 painful

1.09

 supposedly

1.04

&(

1.03

Activations Density 0.002%