INDEX

Explanations

The neuron strongly activates on programming–style identifiers—especially function or method names (often capitalized and followed by parentheses).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 น้อง

0.62

REET

0.59

ベイ

0.57

 والسلام

0.56

 สวัสดี

0.54

ഞ്ഞി

0.53

peror

0.52

 เลข

0.52

 VARCHAR

0.52

dataGridView

0.51

POSITIVE LOGITS

 function

1.64

function

1.56

 funzione

1.34

函数

1.28

 função

1.26

 функцию

1.25

 función

1.23

 Function

1.20

 functions

1.16

 FUNCTION

1.16

Activations Density 0.008%