H&M-eval is an evaluation benchmark for fashion image-text retrieval built from the
public H&M Personalized Fashion Recommendations
catalog. It follows the same BEIR-style structure as
ZooClaw-Fashion so models
can be evaluated on both benchmarks with the same code path. Released as part of the data-agent benchmarks served via
zoodata.ai and used by agents on the
ZooClaw platform.
Released alongside ZooClaw-FashionSigLIP2
to test… See the full description on the dataset page:
https://huggingface.co/datasets/srpone/hm-eval.
