INDEX

Explanations

refusal message completion

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

↵↵

2.96

↵↵↵

2.93

↵↵↵↵

2.65

↵↵↵↵↵

2.53

↵↵↵↵↵↵↵↵↵↵↵

2.38

↵↵↵↵↵↵↵↵

2.36

<sup>

2.32

↵↵↵↵↵↵↵↵↵↵

2.29

↵↵↵↵↵↵↵

2.28

↵↵↵↵↵↵↵↵↵↵↵↵

2.28

POSITIVE LOGITS

.”.

2.85

.".

2.80

。.

2.31

 youll

2.30

."],

2.29

 ….

2.28

 youre

2.26

.").

2.02

\\.

2.02

 ”.

2.00

Activations Density 0.931%