INDEX

Explanations

numbered lists or steps

The neuron primarily detects sentence‐ or clause‐boundary punctuation (e.g. commas and periods).

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

 пожалуйста

0.88

 einfach

0.84

 clicking

0.81

 breezy

0.80

 crafty

0.78

 simplemente

0.77

 trusty

0.75

 please

0.75

 cheesy

0.75

 просто

0.74

POSITIVE LOGITS

 Develop

0.93

Develop

0.89

Maintain

0.86

Create

0.85

Calculate

0.81

 Maintain

0.81

 Obtain

0.80

Identify

0.80

 develop

0.79

 Identify

0.78

Activations Density 0.369%