INDEX

Explanations

perform

New Auto-Interp

Configuration

Prompts (Dashboard)

16,384 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

å¹²

-0.34

åģļ

-0.34

åģļäºĨ

-0.34

 doing

-0.33

isis

-0.32

 Doing

-0.32

åģļçļĦ

-0.31

avn

-0.28

iji

-0.27

æĶ¾å¼ĥäºĨ

-0.27

POSITIVE LOGITS

 tasks

0.49

 task

0.46

 actions

0.43

 action

0.43

ä»»åĬ¡

0.41

 specific

0.39

 duties

0.36

tasks

0.35

 Tasks

0.35

ä»»åĭĻ

0.35

Activations Density 0.894%