INDEX

Explanations

policy guidelines and making

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

 have

-1.22

 utilize

-1.03

So

-0.97

…

-0.95

or

-0.95

 rise

-0.95

 loss

-0.95

依照

-0.95

end

-0.92

POSITIVE LOGITS

 policy

1.73

 makers

1.38

 POLICY

1.31

policy

1.30

 Policies

1.27

Policy

1.27

 Policy

1.27

"},

1.23

policies

1.21

making

1.20

Activations Density 0.015%