INDEX

Explanations

citations and references

The neuron is triggering on numeric tokens in scholarly references—i.e. volume, issue, and page numbers in citations.

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

ਖ਼

-1.38

 excès

-1.34

 détruire

-1.32

hypothesis

-1.32

midler

-1.29

˗

-1.24

 olhando

-1.23

 desej

-1.23

女の子

-1.23

handleClose

-1.23

POSITIVE LOGITS

it

1.84

 there

1.57

at

1.55

 они

1.54

 they

1.52

to

1.48

 его

1.45

!");

1.45

"));

1.44

by

1.42

Activations Density 0.011%