{"repo":"ruc-datalab/CoDA-Bench","free":true,"listed":false,"github":"https://github.com/ruc-datalab/CoDA-Bench","clone":"git clone https://github.com/ruc-datalab/CoDA-Bench.git","description":"CoDA-Bench is a benchmark for code agents on data-intensive tasks. 🎈代码智能体能搞定数据密集型任务吗?","language":"Python","stars":43,"topics":["agent","agentic","agentic-ai","ai","benchmark","code-agent","code-generation","coding-agent","data-analysis","data-engineering"],"license":"MIT","category":"data-pipelines","readme_excerpt":"--- Code and data for [ICML 2026] CoDA-Bench: Can Code Agents Handle Data-Intensive Tasks? Authors : Yuxin Zhang, Ju Fan, Meihao Fan, Shaolei Zhang , Xiaoyong Du Renmin University of China 👋 Overview CoDA-Bench (Code and Data-intensive Benchmark) is a benchmark for evaluating AI agents on data-intensive analytical tasks. Given a natural language question and access to a Linux sandbox containing hundreds of data files, an agent must discover relevant data, write code, and produce the correct answer. https://github.com/user-attachments/assets/34e50a62-744b-4079-8988-6a8bbfe166a0 ▶️ Demo video. If it does not play inline, download/watch it here. Unlike existing benchmarks that provide oracle data directly, CoDA-Bench requires agents to: - 🔍 Discover relevant data among hundreds of semantically similar files - 🗂️ Navigate complex file hierarchies in a Linux sandbox - 🔗 Integrate information from multiple heterogeneous data sources - 💻 Generate correct code for data-driven analytical tasks 📰 News [Jun. 8, 2026] : CoDA-Bench v1.0 released with Docker evaluation! [Jun. 1, 2026] : CoDA-Bench paper accepted at ICML 2026! 📊 Dataset Statistics Metric Value -------- ------- Total Tasks 1,009 (full) / 119 (hard subset) Communities 31 (full) / 15 (hard subset) Source Datasets 199 Kaggle datasets Avg Files per Task 980 files Total Size 43 GB (compressed) 🚀 Quick Start Installation Download Dataset This downloads: - Benchmark files ( coda bench.json , coda bench hard.json ) - Communi","default_branch":null,"files":null,"tree":[],"storefront":"/r/ruc-datalab","claimed":false,"request_supported":{"post":"https://gitbuyer.com/r/ruc-datalab/CoDA-Bench/request-supported","requests":0},"note":"indexed from public GitHub; nothing is for sale on this page. Clone it from GitHub. Paid listings live at /search."}