INDEX

Explanations

adding or preceding text

The neuron strongly activates on numeric labels followed by a period (e.g. section or reference numbers like “1.,” “23.,” “113.”).

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

0.84

0.83

al

0.78

AL

0.77

0.75

0.74

end

0.72

الم

0.71

ar

0.70

POSITIVE LOGITS

 ды

1.05

 Боль

0.96

 дать

0.94

 увидеть

0.93

 суммы

0.91

UIServer

0.90

 Biết

0.88

 tRNA

0.88

 Пло

0.87

 tweeter

0.85

Activations Density 0.001%