INDEX

Explanations

History sections

The neuron activates on numeric tokens (e.g. years, counts, or other digit sequences).

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

 zoude

-1.39

鹪

-1.38

僦

-1.37

 accepte

-1.32

 alimente

-1.31

 dezelve

-1.30

︲

-1.30

 plafon

-1.30



-1.30

卺

-1.29

POSITIVE LOGITS

and

1.45

in

1.39

 that

1.38

1.30

 often

1.29

so

1.27

 under

1.27

out

1.24

1.21

 However

1.20

Activations Density 0.000%