6Graph

6Graph converts the seed list into a graph whose vertices describe nybble positions and values. Density-increasing spanning trees identify tightly-coupled substructures, which are then recombined to produce new candidate addresses while an outlier bucket preserves coverage.

• Reference: Placeholder citation (coming soon).

Train

rmap train --seeds seeds/hitlist.txt --output models/6graph.bin six-graph \
  --min-region-size 32 --iterations 3

Generate

rmap generate --model models/6graph.bin --count 250000 \
  --unique --output 6graph.txt

Configuration

• --min-region-size <usize> – minimum seed count that a DHC region must satisfy before mining begins (default 16).
• --distance-threshold <usize> – retained for compatibility; tune when experimenting with alternate distance heuristics (default 12).
• --iterations <usize> – number of outlier refinement passes (default 3).

Model notes

• Training stores both the mined cluster patterns and the original seeds used to build each component so the generator can reintroduce variety when a region runs dry.
• Adjust --iterations to balance runtime against coverage: additional passes revisit outliers but can significantly lengthen training.