INDEX

Explanations

common, easier, available, harder

This neuron fires on discourse‐level framing or transition words and phrases that kick off or shift topics (e.g. “Your donation,” “it’s not unusual,” “in order to,” “only,” “years”).

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

ñana

0.60

 écrire

0.57

 писать

0.56

 escrever

0.55

 शह

0.53

 स्वतंत्रता

0.52

 provas

0.52

 सापेक्ष

0.52

 különböz

0.52

 жидкости

0.50

POSITIVE LOGITS

for

1.19

dla

1.19

สำหรับ

1.16

 bagi

1.09

for

1.05

 для

1.03

для

1.02

សម្រាប់

1.01

 für

1.00

 עבור

0.96

Activations Density 0.223%