INDEX

Explanations

news wire markers

This neuron detects regular‐expression constructs (e.g. character‐class brackets, grouping parentheses, non-capturing-group markers, anchors).

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

to

-1.84

 from

-1.64

at

-1.63

and

-1.60

as

-1.54

had

-1.49

 went

-1.48

on

-1.40

 gave

-1.35

or

-1.30

POSITIVE LOGITS

 apporter

1.52

 montrer

1.51

totally

1.48

 bén

1.48

 accompagn

1.48

鵙

1.47

沵

1.44

訁

1.40

 faciliter

1.38

 accompagner

1.38

Activations Density 0.048%