Skip to content

NineAbyss/CP-Agent

Repository files navigation

💻🤖️ CP-Agent

CP-Agent: A Calibrated Risk-Controlled Agent for Feedback-Driven Competitive Programming

More Analysis

Additional rebuttal experiments, case studies, and efficiency derivations are available in proof and additional analysis.

🎯Quick Start

Environment Preparation

uv sync
source .venv/bin/activate

Evaluation

Start Phoenix to monitor the traces.

    python -m phoenix.server.main serve

Configure in ./agentflow/configs/config.yaml

  • Test on single problem:

      python ./agentflow/main_agent.py
  • Test on single ICPC-Eval:

      sh ./ICPC-Eval/test_agent.sh
  • Test on single LiveCodeBench Pro:

   sh ./LiveCodeBench-Pro/test_agent.sh

About

Self-Evolving Competitive Programming via Execution-Feedback Loop

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors