INDEX

Explanations

expected result or output

The neuron fires on numeric literal tokens (i.e. numbers and numeric values) in the 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

 ahora

-1.47

Colm

-1.31

 lleg

-1.27

 tři

-1.25

就知道

-1.24

 bezpośrednio

-1.24

elser

-1.21

picion

-1.20

Audiobook

-1.19

也算是

-1.18

POSITIVE LOGITS

 that

1.58

 solution

1.47

 still

1.45

 alimenti

1.34

that

1.31

 risultato

1.31

 olmayan

1.21

 solutions

1.20

ings

1.19

<strong>

1.14

Activations Density 0.080%