INDEX

Explanations

Here's the explanation for the neuron's behavior:numbers following `<sup>`

The neuron strongly activates on numeric tokens—especially years, scores, or other digit sequences—in the text.

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 resolução

0.56

ighed

0.52

ătur

0.50

 ਦੀ

0.50

ᓛ

0.49

 интеллектуа

0.49

 resolución

0.48

īng

0.48

 వంటి

0.48

 플레이

0.48

POSITIVE LOGITS

 other

0.74

 Other

0.59

Other

0.55

 then

0.54

 others

0.54

 OTHER

0.52

 lainnya

0.52

อื่นๆ

0.52

翌

0.52

 outras

0.52

Activations Density 0.061%