INDEX

Explanations

software components and identifiers

The neuron fires on code identifiers—especially class/service/type names and dependency-injection tokens (e.g. Options, Logger, Service, Handler, Repository, Context, Registry, etc.).

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

 that

-1.73

or

-1.59

for

-1.55

 Bemerkungen

-1.52

-1.48

 orgullo

-1.38

为什么会

-1.37

 when

-1.34

 которы

-1.34

 decently

-1.32

POSITIVE LOGITS

1.31

bte

1.28

 SEVENT

1.25

odil

1.24

oia

1.24

alakip

1.23

 itali

1.22

 remonter

1.22

 constater

1.22

it

1.22

Activations Density 0.023%