INDEX

Explanations

code and structured text boundaries

The neuron detects long runs of whitespace used for indentation at the beginnings of code or markup lines.

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

 these

-1.53

 their

-1.49

 that

-1.48

 ultimately

-1.42

 then

-1.36

 this

-1.35

 some

-1.30

Concerns

-1.24

 consultez

-1.24

 often

-1.23

POSITIVE LOGITS

1.53

 biß

1.48

 takže

1.45

,'

1.31

};

1.30

*/

1.29

sivamente

1.29

 новых

1.28

ußen

1.28

ylase

1.25

Activations Density 0.064%