Team Ai
Datasetpublic

ATH-MaaS/HSCodeComp

HSCodeComp: A Realistic and Expert-Level Benchmark for Deep Search Agents in Hierarchical Rule Application Paper | Code | Dataset on Hugging Face ⭐ MarcoPolo Team ⭐ Alibaba Group 🗂️ Data 📌 Overview HSCodeComp is the first realistic, expert-level e-commerce benchmark designed to evaluate deep search agents on their ability to perform Level-3 knowledge—hierarchical rule application—a critical yet overlooked capability in current agent evaluation… See the full description on the dataset page: https://huggingface.co/datasets/ATH-MaaS/HSCodeComp.

sourceHugging Faceapache-2.0updated 3mo agoView on Hugging Face
12likes632downloads
4 commits on main
ce911973mo ago

Update README.md

zifushang
ea1820a1y ago

Improve HSCodeComp dataset card: Add metadata, update paper/code links, and sample usage (#3)

QihuiLT, nielsr
498ae951y ago

Upload folder using huggingface_hub

QihuiLT
04f3cbd1y ago

initial commit

QihuiLT