T2Retrieval: A Large-Scale Chinese Benchmark for Passage Ranking

T2Retrieval is a dataset from the Massive Text Embedding Benchmark (MTEB) designed for evaluating text embedding models. The dataset is a large-scale Chinese benchmark for passage ranking tasks. It was authored by mteb and the dataset page was last updated on 2025-05-03.

Use Cases

• Benchmarking text embedding models for retrieval tasks based on the MTEB framework.
• Evaluating model performance on Chinese passage ranking based on the dataset's domain coverage.
• Training or fine-tuning retrieval models for specialized domains like medical or financial text.
• Conducting research on cross-domain information retrieval performance in Chinese.

Strengths

• Part of the standardized Massive Text Embedding Benchmark (MTEB) framework.
• Covers multiple specialized domains including Medical, Academic, Financial, Government, and Non-fiction.
• Designed specifically for Chinese language passage ranking, addressing a need for non-English benchmarks.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment for large-scale training.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

mteb

Freshness

Last updated 2025-05-03 20:40:51; freshness should be verified.