INDEX

Explanations

java code annotations and declarations

The neuron fires on numeric literal tokens (digit sequences) in code, especially numbers used as annotation parameter values.

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

ərb

-0.93

];

-0.93

 disponibile

-0.84

 informée

-0.84

 Hentet

-0.83

рева

-0.82

();

-0.79

 особенно

-0.78

retanto

-0.76

文化

-0.76

POSITIVE LOGITS

1.37

{"

1.28

{'

1.22

={

1.18

({

1.16

={

1.14

1.07

,{

1.04

+{

1.03

{})

0.88

Activations Density 0.008%