INDEX

Explanations

code snippets and assignments

The neuron fires strongly on code tokens that signal operations—especially method or function calls and assignments (e.g. “push(…)”, “<-”, “doit(…)”, “render = …”). It is detecting imperative code statements.

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

():

-1.19

as

-1.16

now

-1.12

];

-1.02

"):

-1.02

—”

-1.01

';

-1.01

to

-1.00

 came

-0.99

*}$

-0.98

POSITIVE LOGITS

 also

1.41

 همچنین

1.41

 επίσης

1.28

 także

1.27

 erreich

1.24

 również

1.18

 certainement

1.15

 также

1.10

initiated

1.08

 גם

1.07

Activations Density 0.052%