INDEX

Explanations

code syntax and indexing

The neuron fires on loop‐oriented control structures (e.g. for, foreach and their related symbols), i.e. it detects iteration constructs.

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

 beide

-0.91

 Нам

-0.90

爆发

-0.85

emény

-0.84

mā

-0.79

льт

-0.79

Txt

-0.78

iso

-0.77

 बनाने

-0.77

remes

-0.77

POSITIVE LOGITS

Individ

0.99

 covenant

0.98

 individually

0.98

individ

0.90

 locul

0.89

 trzeba

0.84

 Covenant

0.83

 viata

0.81

 mismos

0.81

ÍA

0.81

Activations Density 0.004%