INDEX

Explanations

convert unicode

The neuron is keyed to words and phrases that express suggestions, intentions, or calls to action (e.g. modals like “can,” “want to,” “let’s,” and verbs conveying purpose or advice).

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

 alongside

1.00

––

0.93

 gleaned

0.93

 ostensibly

0.93

 garnered

0.88

 insofar

0.88

 reimag

0.87

–

0.87

 brimming

0.87

Alongside

0.83

POSITIVE LOGITS

 advices

1.31

varage

1.26

 loosing

1.21

 benutzen

1.19

 analis

1.15

 meisten

1.14

 bisschen

1.11

 banyak

1.10

 дуже

1.09

 spezielle

1.09

Activations Density 1.293%