INDEX

Explanations

technical documentation review

The neuron fires most strongly on distinctive, domain‐specific keywords—particularly proper names, technical or product jargon, and other specific nouns.

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

ступ

0.79

NAL

0.75

ar

0.74

id

0.73

লের

0.72

earing

0.71

то

0.71

se

0.69

ه

0.69

ки

0.69

POSITIVE LOGITS

 tenemos

0.93

HDD

0.92

 подобрать

0.89

 бонусу

0.89

ROH

0.87

FAU

0.86

 peque

0.85

IRR

0.85

 pequ

0.84

 monomer

0.84

Activations Density 0.001%