Paper | GitHub | Leaderboard
Dataset repository for the paper AgentCoMa: A Compositional Benchmark Mixing Commonsense and
Mathematical Reasoning in Real-World Scenarios.
To submit to the Leaderboard, follow the instructions in this README.
AgentCoMa is an Agentic Commonsense and Math benchmark where each compositional task requires both commonsense and mathematical reasoning to be solved. The tasks are set in real-world scenarios: house working, web… See the full description on the dataset page:
https://huggingface.co/datasets/LisaAlaz/AgentCoMa.
