INDEX

Explanations

zeros and ones

This neuron responds to the word “Related,” i.e. cues for “related” links or related-content navigation labels.

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

 piek

0.61

 antagonists

0.59

்த்து

0.59

 colleagues

0.56

 aler

0.56

 coworkers

0.55

vação

0.54

ኋላ

0.54

 समुद्र

0.53

 siblings

0.53

POSITIVE LOGITS

Retro

0.58

<h2>

0.54

Clusters

0.54

ällig

0.53

GetUser

0.53

GetType

0.52

Communities

0.52

 সৌজন্য

0.52

চিন্ত

0.51

чным

0.50

Activations Density 0.000%