INDEX

Explanations

detecting or missing

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

 observational

1.15

検証

1.12

 निरी

1.08

 Stopwatch

1.06

 empirically

1.05

 empirical

1.05

 intervent

1.05

assertions

1.03

なあ

1.02

実験

1.02

POSITIVE LOGITS

 reve

1.04

rules

1.00

 rule

0.98

reve

0.97

 rules

0.97

 मुना

0.95

规则

0.93

rule

0.93

 Ruling

0.93

 returns

0.92

Activations Density 0.066%