Search papers, labs, and topics across Lattice.
The MARS framework introduces a prompt-only system that leverages specialized agents for different algorithmic techniques in competitive programming, addressing the limitations of traditional multi-agent pipelines. By dynamically selecting a team of topic specialists and iteratively refining solutions through a structured relay process, MARS significantly improves code generation efficiency and effectiveness. The system achieves a pass rate of 0.624 on the CodeContests test split, outperforming direct prompting by 14.4 percentage points while reducing wall-clock costs by a factor of 3.3.
MARS outperforms traditional multi-agent systems in competitive programming by harnessing specialized LLMs, achieving higher pass rates with lower costs.
Large Language Models excel at code generation, yet competitive programming exposes a persistent failure mode: existing multi-agent pipelines distribute work over generic planner, coder, and debugger roles and delegate the choice of algorithmic technique to the backbone alone. We present MARS (Multi-Agent Relay of Specialized LLMs), a prompt-only framework in which each agent is a topic specialist---dynamic programming, graphs, strings, geometry, and so on---grounded by retrieval-augmented generation over an algorithm-theory corpus. Given a problem, retrieval selects a small team of relevant specialists; a starter writes an initial C++17 solution, and each subsequent turn runs the candidate against public examples in a sandbox, lets the active specialist keep, repair, or hand off the draft, and forwards a structured packet to the next specialist. A single infrastructure-fixer pass normalizes boilerplate at the end. On the CodeContests test split with Gemma 4, MARS reaches $0.624 \pm 0.006$ pass rate at $2.3$ recorded pipeline stages per task ($+14.4$ percentage points over direct prompting), closing most of the gap to CodeSIM ($0.731$) at $3.3{\times}$ lower wall-clock cost and substantially smaller variance in per-task token spend. The source code is available on GitHub: https://github.com/fckand/mars.
