INDEX

Explanations

community, 4D, version control

The neuron never fires—it doesn’t detect any pattern (i.e. it’s effectively inactive).

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

-2.69

ଝ

-2.56

 allow

-2.44

一些

-2.41

end

-2.30

 diligent

-2.19

</b>

-2.17

</h4>

-2.16

-2.11

POSITIVE LOGITS

莇

2.80

蜮

2.70

蹰

2.66

м

2.63

 deliciously

2.58

୬

2.50

輊

2.42

 spearheaded

2.36

 récol

2.36

 обеспечение

2.34

Activations Density 0.005%