INDEX

Explanations

.method or .property access

This neuron fires on programming‐language tokens—especially variable names, types, literals, and the “=” of assignment statements—i.e. it detects code/assignment syntax.

New Auto-Interp

Configuration

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Negative Logits

how

-1.97

𝔥

-1.49

 what

-1.34

 zincir

-1.31

 устройств

-1.25

 беже

-1.23

 ἐκ

-1.21

 eğiti

-1.21

RootElement

-1.20

친

-1.20

POSITIVE LOGITS

ּוֹ

1.36

 воин

1.30

 này

1.30

).

1.24

 during

1.22

 卧室

1.21

有一定的

1.21

👊

1.20

ʎ

1.19

";

1.16

Activations Density 0.011%