INDEX

Explanations

a or an followed by a noun

The neuron is selectively detecting the standalone indefinite article “a.”

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

 another

-1.59

另一

-1.55

unno

-1.28

 bancada

-1.24

<bos>

-1.23

ต่างๆ

-1.22

鵯

-1.20

好評

-1.17

!***

-1.16

也是

-1.14

POSITIVE LOGITS

 like

1.45

 such

1.34

 datang

1.23

to

1.23

 ilyen

1.19

chte

1.17

 being

1.15

 soos

1.14

 sådan

1.14

 having

1.13

Activations Density 0.109%