INDEX

Explanations

apache, google, eclipse software

The neuron activates on organization/vendor identifiers in Java/Scala package names (e.g. “apache,” “google,” “eclipse,” “spark”) within class‐path or XML snippets.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

 schien

-1.73

'".

-1.60

翆

-1.58

 améli

-1.48

禑

-1.42

''.

-1.41

 tornou

-1.41

 promove

-1.40

'}

-1.39

 pertence

-1.38

POSITIVE LOGITS

 from

2.17

2.03

了一会儿

1.75

槼

1.73

 only

1.64

規則

1.57

憚

1.50

klju

1.49

 while

1.45

 complesso

1.42

Activations Density 0.006%