INDEX

Explanations

ability to accomplish

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

 intentar

-1.51

 ability

-1.42

能够

-1.36

attempt

-1.29

 попыта

-1.22

 tente

-1.20

能

-1.20

 попробовать

-1.19

 attempting

-1.17

 LoggerFactory

-1.17

POSITIVE LOGITS

 successfully

1.33

get

1.28

 withstand

1.09

 find

1.09

 convince

1.08

 grasp

1.02

 somewhat

0.98

 einiger

0.98

 hold

0.96

 pretty

0.96

Activations Density 0.040%