BRIDGE-Open: 55 Open-Access Tasks from the BRIDGE Benchmark

The BRIDGE-Open dataset, released by YLab-Open in August 2025, is an open-access subset of the datasets used in the BRIDGE benchmark. It comprises 55 tasks from datasets that can be publicly released, while other tasks involving clinical data remain under regulated access. The original sources and licenses for each included dataset are detailed in the associated research paper.

Use Cases

• Benchmarking model performance across 55 diverse tasks based on the dataset's composition.
• Developing and testing AI models on open-access data subsets based on the described task collection.
• Studying task transfer and generalization using the multi-task structure implied by the benchmark.
• Comparing model capabilities on tasks where clinical data is restricted based on the dataset's stated scope.

Strengths

• Contains 55 distinct tasks for multi-domain evaluation.
• Provides clear provenance by referencing original dataset sources and licenses in the associated paper.
• Last updated on 2025-08-16, indicating recent maintenance.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count, file formats, and dataset size are unknown, which may limit suitability assessment.
• The description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

YLab-Open, as a subset of datasets used in the BRIDGE benchmark.

Collection Method

Curated selection of open-access tasks from a larger benchmark collection.

Time Range

null

Freshness

Last updated 2025-08-16 23:00:44.

Geography

null