FiLMMeD augments a Transformer with FiLM to create one model for 24 MDVRP variants, adds curriculum learning for multi-depot constraints, shows preference optimization outperforming RL in MTL, and outperforms baselines on experiments including 8 new formulations.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FiLMMeD: Feature-wise Linear Modulation for Cross-Problem Multi-Depot Vehicle Routing
FiLMMeD augments a Transformer with FiLM to create one model for 24 MDVRP variants, adds curriculum learning for multi-depot constraints, shows preference optimization outperforming RL in MTL, and outperforms baselines on experiments including 8 new formulations.