INDEX

Explanations

recommendations and potential outcomes

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 clientèle

-1.18

 sona

-1.17

 många

-1.16

 sige

-1.15

 bijoux

-1.13

 bronz

-1.13

 seksi

-1.13

 exemples

-1.13

 gastron

-1.12

 Wikisource

-1.11

POSITIVE LOGITS

 should

1.98

may

1.44

 design

1.30

and

1.27

 recommend

1.24

 would

1.23

 further

1.20

 recommended

1.19

 will

1.16

 might

1.15

Activations Density 0.156%