INDEX

Explanations

court document dates

The neuron activates on numeric tokens—especially case numbers, dates, and 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

,'

-1.37

,"

-1.32

So

-1.29

 így

-1.26

For

-1.23

end

-1.20

-1.19

pecabe

-1.18

-1.17

POSITIVE LOGITS

<bos>

1.81

پاسخ

1.38

موقع

1.37

 обычно

1.34

したのは

1.30

avajillas

1.27

ılmış

1.27

事態

1.24

ífico

1.23

Вопрос

1.23

Activations Density 0.012%