INDEX

Explanations

percentage decrease or loss

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

Always

-1.25

Điện

-1.16

ımda

-1.14

 ͜ʖ

-1.10

ментов

-1.07

 καλύτε

-1.06

いなかった

-1.06

 impon

-1.06

-1.05

always

-1.03

POSITIVE LOGITS

 loss

1.63

 losses

1.49

its

1.38

due

1.27

ternut

1.26

loss

1.24

Loss

1.21

}=-

1.20

 Loss

1.17

)=-

1.17

Activations Density 0.055%