INDEX

Explanations

code formatting marks

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

of

-1.63

the

-1.57

to

-1.35

 enters

-1.31

：……

-1.28

 recupera

-1.27

 undertakes

-1.27

 expects

-1.26

 differs

-1.23

шает

-1.23

POSITIVE LOGITS

</strong>

1.98

</i>

1.73

 arXiv

1.66

</em>

1.61

</code>

1.59

</u>

1.58

ateness

1.58

1.55

[]:

1.52

都没

1.52

Activations Density 0.004%