INDEX

Explanations

action followed by a consequence

only to a negative outcome

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 包括

0.42

arean

0.40

affen

0.40

ያስ

0.40

lude

0.39

 includes

0.39

 Agents

0.39

 Allows

0.38

📔

0.38

 жөн

0.38

POSITIVE LOGITS

 να

0.77

 kemudian

0.68

 затем

0.65

 akhirnya

0.63

 būti

0.61

然后

0.61

 finally

0.60

 להיות

0.59

 then

0.57

 vervolgens

0.57

Activations Density 0.038%