MSMARCO Annotation: Utility Labels for Training Dense Retrievers

Wanglanhuajiaofen annotated the MS MARCO dataset for utility using the Qwen3-32B model. The dataset supports research on multi-positive optimization objectives for dense retrieval. It was last updated on 2026-05-19.

Use Cases

• Training dense retrieval models based on multiple positive passages.
• Benchmarking dense retriever performance based on utility annotations.
• Studying multi-positive optimization objectives for information retrieval.

Strengths

• Annotations are generated using the Qwen3-32B model, a large language model.
• Dataset is associated with a published research paper on multi-positive optimization.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Wanglanhuajiaofen on Hugging Face.

Collection Method

Annotations generated using the Qwen3-32B model on the MS MARCO dataset.

Freshness

Last updated 2026-05-19 08:57:44; freshness should be verified.