This dataset contains the benchmark tasks for evaluating multi-agent coordination in code collaboration.
Paper: CooperBench: Why Coding Agents Cannot be Your Teammates Yet
Code: github.com/cooperbench/CooperBench
Website: cooperbench.com
dataset/
├── {repo_name}_task/ # Repository-specific tasks
│ └── task{id}/ # Individual task
│ ├── Dockerfile # Container setup for testing
│ ├──… See the full description on the dataset page:
https://huggingface.co/datasets/CooperBench/cooperbench-dataset.
