INDEX

Explanations

specific nouns or technical terms

The neuron fires on technical, patent‐style nouns naming components or features (e.g. terminals, valves, displays, controllers, pipelines, switches, etc.) that identify parts of an invention.

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

 those

-1.09

 using

-1.05

manı

-0.97

 επίσης

-0.96

 from

-0.93

儀式

-0.91

 частности

-0.91

Sì

-0.90

key

-0.90

していました

-0.89

POSITIVE LOGITS

//{

1.30

;</

1.09

(/^

1.07

thodox

1.05

ícias

1.01

ISTICS

1.00

 $^{\

1.00

Unmanaged

0.99

0.98

similar

0.98

Activations Density 0.079%