INDEX

Explanations

cognitive studies or datasets

This neuron is highly sensitive to the subword sequence “ign” (or its variant “gn”) appearing in words like “ignites,” “igniting,” and “gnarly.”

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

uket

0.66

die

0.59

comm

0.57

 occupants

0.57

 tuyển

0.56

fh

0.55

 indulged

0.55

 เรียก

0.55

da

0.54

naive

0.54

POSITIVE LOGITS

虻

0.66

ificent

0.63

を目

0.58

ratulations

0.57

ificant

0.55

合适

0.54

UARY

0.52

CAUSE

0.52

 BUTTON

0.52

公園

0.52

Activations Density 0.161%