INDEX

Explanations

results or components

The neuron fires on plural technical nouns (words ending in “-s,” like applications, precursors, elements, modifications, progeny, derivatives, products, dependencies, outcomes, components, transforms, etc.).

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

of

-1.87

 hunne

-1.06

 восхи

-1.05

 هفته

-0.99

uyt

-0.99

 grijs

-0.99

得多

-0.98

 zilver

-0.98

镲

-0.98

 blauw

-0.97

POSITIVE LOGITS

 sowie

1.28

viève

1.23

When

1.19

 oraz

1.14

 should

1.10

 When

1.09

！！！！

1.08

適量

1.08

litre

1.05

 cuando

1.03

Activations Density 0.070%