INDEX

Explanations

addressing the president

This neuron detects numbered tokens (e.g. vote counts, amendment or article numbers, timestamps, and other numeric labels) in parliamentary transcript text.

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

to

-1.49

ׇ

-1.28

\#:

-1.28

⛧

-1.27

 broderie

-1.23

 solides

-1.20

吅

-1.20

источник

-1.18

ἡ

-1.16

 most

-1.16

POSITIVE LOGITS

ύ

1.49

 diverso

1.41

ּוֹ

1.34

viating

1.30

havn

1.28

dding

1.27

hift

1.27

 وزار

1.24

 取り付け

1.24

 правительства

1.18

Activations Density 0.011%