INDEX

Explanations

prohibited or permitted

This neuron detects words (and their variants) that signal legal or official permission, prohibition, or requirement (e.g. “allowed,” “illegal,” “license,” “authorization”).

New Auto-Interp

Configuration

Prompts (Dashboard)

392,802 prompts, 256 tokens each

Dataset (Dashboard)

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

Negative Logits

 ઘટ

0.80

 coales

0.76

 خواهد

0.75

 reorganization

0.74

 മികച്ച

0.73

 iteratively

0.71

⁺

0.71

增

0.70

 optimizing

0.70

 ग्रोथ

0.68

POSITIVE LOGITS

 prohibited

1.47

 permitido

1.41

禁止

1.38

 forbidden

1.34

 prohibition

1.30

 prohibitions

1.29

forbidden

1.27

 forbids

1.27

 proib

1.27

 నిషే

1.25

Activations Density 0.001%