INDEX

Explanations

then followed by a list or transition

The neuron strongly activates on numeric tokens (especially multi‐digit numbers with decimals, like timestamps or years).

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

ح

1.32

alım

1.30

 câte

1.27

ות

1.25

 ciò

1.24

்ட

1.23

ח

1.23

та

1.20

️⃣

1.19

 unscathed

1.19

POSITIVE LOGITS

 പിന്നെ

1.52

 vervolgens

1.51

데

1.41

های

1.36

не

1.32

selves

1.27

ی

1.26

 lintas

1.25

こと

1.18

 anschließend

1.18

Activations Density 0.089%