INDEX

Explanations

concentrate

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 concentrated

-1.11

 concentrating

-1.05

 poffe

-1.05

 concentrates

-1.05

 command

-1.03

 Shakspeare

-1.03

 myſelf

-1.02

 itſelf

-1.01

 Concentr

-0.98

yntaxException

-0.98

POSITIVE LOGITS

sur

0.59

her

0.56

arik

0.54

ɵ

0.54

an

0.49

zu

0.49

tri

0.49

tra

0.49

Tri

0.48

Activations Density 0.296%