INDEX

Explanations

instead of

This neuron detects reporting or attribution verbs that introduce or frame quoted or referenced statements (e.g., “reminds,” “explains,” “says,” “tells”).

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

Removed

0.60

dalam

0.59

ங்களிலிருந்து

0.57

するように

0.55

中に

0.55

하는데

0.55

在本

0.54

 этот

0.54

ங்களுக்கு

0.54

โยชน์

0.54

POSITIVE LOGITS

of

2.95

 ofthe

2.47

 của

2.30

of

2.17

 của

2.11

ของการ

2.09

ของ

2.03

 thereof

1.91

 של

1.91

Of

1.86

Activations Density 0.857%