v1.1

• Add model M3
• Implement _set_start for all models
• Update benchmark script

v1.0

Fix newline in README