INDEX

Explanations

unspoken emotions, needs, understanding

The neuron flags unusually rare or low‐frequency tokens in the text, i.e. words that stand out as uncommon vocabulary.

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

不够

0.69

enough

0.66

gt

0.66

iD

0.66

only

0.63

lenme

0.62

க்கும்

0.61

 enough

0.60

hythm

0.60

 ندارد

0.60

POSITIVE LOGITS

 siquiera

0.75

🙅

0.72

 ramb

0.68

 brute

0.64

ीकृत

0.64

 hulk

0.64

પણે

0.64

шими

0.63

 Territory

0.61

 cinqui

0.61

Activations Density 0.351%