INDEX

Explanations

numbers and letters

The neuron strongly fires on standalone alphanumeric ID‐style tokens (e.g. mixed letters and digits like video IDs, access keys, or other unique identifiers).

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

of

-2.28

on

-1.59

and

-1.53

to

-1.47

 juſ

-1.45

 открыть

-1.42

 efectuar

-1.34

뿐

-1.28

ſu

-1.27

boe

-1.27

POSITIVE LOGITS

two

1.70

 more

1.65

 another

1.61

 three

1.49

 very

1.49

 بسیار

1.45

 Lainnya

1.44

 delicioso

1.44

 other

1.34

would

1.34

Activations Density 0.016%