lists of items

The neuron strongly activates on structured numeric labels (e.g. “6.5.7”), section or question numbers, and similar outline‐style digit sequences.

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

 minhas

0.47

 লইয়া

0.46

 semelhantes

0.45

 Elektrokh

0.45

 grosseur

0.45

 supernatants

0.44

фіка

0.43

 perkara

0.43

 acontecimientos

0.43

？”

0.43

POSITIVE LOGITS

0.70

0.61

↵↵

0.60

0.59

0.57

0.55

0.54

0.53

Activations Density 0.304%