INDEX

Explanations

This neuron activates on code identifiers (e.g. class or module names) that include digits, spotting tokens in import or code lines where an uppercase identifier is immediately followed by numbers.

Http Interceptor

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

-1.29

『

-1.20

-1.16

 also

-1.16

-1.11

all

-1.11

 EconPapers

-1.08

</b>

-1.06

-1.05

POSITIVE LOGITS

 emble

1.27

 пыта

1.24

鹟

1.23

 symptômes

1.21

Lø

1.19

SHIN

1.17

FONTE

1.14

 liefde

1.13

angebote

1.13

DAZ

1.13

Activations Density 0.000%