Tag
cp-algorithms 上关于 Treap（笛卡尔树）的算法文档，介绍了这种结合二叉搜索树与二叉堆特性的随机化数据结构及其插入、删除、合并等操作的复杂度与实现思路。
该论文提出了一个让模型仅凭题面和源码判断并诊断候选程序正确性、并以生成反例作为可验证证据的任务，构建了 USACO-JUDGE 基准，并训练了 35B MoE 代理验证器 ARBITER，显著提升了代码验证、选择与改进效果。
NVIDIA 展示了如何基于 Nemotron 3 通过 SFT、RL 和 GenCorrect 推理循环进行专业化微调，在 IOI 2026 和 IMO 2026 上同时达到金牌水平，证明了一个可复用的专业化训练配方。
A post sharing the free PDF of the Competitive Programmer's Handbook (2018), a widely used textbook covering algorithmic techniques and problem-solving strategies from cses.fi.
The paper investigates whether Large Language Models fix bugs by making minimal edits or reimplementing solutions from scratch in competitive programming, finding that LLMs often modify more than necessary and solve better when generating new solutions, which has implications for debugging tools.
This paper presents a specialization pipeline for post-training language models to achieve gold-medal performance in coding competitions, demonstrating top scores on IOI benchmarks using techniques like supervised fine-tuning and reinforcement learning.
MARS is a multi-agent LLM framework using specialized agents for different algorithmic topics to solve competitive programming problems, achieving a 0.624 pass rate with lower cost and variance than direct prompting.
This paper introduces CurveShift, an analysis method that separates overall ability gains from difficulty-specific improvements in LLM agents. Using METR time-horizon data and LiveCodeBench, it finds that most apparent shifts toward harder tasks are ceiling effects, though a genuine hard-task effect exists for reasoning models in competitive programming.
A new AI model has achieved superhuman performance in competitive programming, marking a significant milestone in AI capabilities.
Scores update from the ongoing AtCoder heuristics finals, which will continue until July 8, 19:00 JST.
Announces the free availability of the 'Competitive Programmer’s Handbook', covering 30 important DSA topics for tech interview preparation.
CP-Agent presents a calibrated risk-controlled approach for feedback-driven competitive programming using large language models, achieving significant improvements on benchmarks without parameter updates.
GrandCode is a multi-agent reinforcement learning system that achieves grandmaster level in competitive programming, consistently beating all human participants in live Codeforces contests using a novel Agentic GRPO method.
Solvita is an agentic evolution framework that enables continuous learning in code generation through reinforcement learning updates to graph-structured knowledge networks, achieving state-of-the-art performance on competitive programming benchmarks.
This profile interviews Cognition AI co-founder Scott Wu, exploring his background as a competitive programming prodigy and the rapid commercial success of his AI coding agent Devin, which has reached a $445 million revenue run rate and a $25 billion valuation.