INDEX

Explanations

valid/invalid data formats

The neuron flags domain‐specific technical keywords (e.g. “website,” “url,” “date,” “valid,” “values,” “data,” etc.) in programming questions.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

«

-1.54

&#

-1.49

，「

-1.48

 belki

-1.43

 Inſ

-1.41

 durante

-1.35

-1.32

-1.30

 onze

-1.27

 THOSE

-1.26

POSITIVE LOGITS

 cori

1.51

 Lengkap

1.45

1.41

无效

1.41

Here

1.41

this

1.38

This

1.37

Having

1.34

 gila

1.34

 will

1.33

Activations Density 0.026%