SonoInstruct: Large-Scale Ultrasound Instruction Dataset

SonoInstruct is a large-scale instruction dataset designed for ultrasound (sonography) understanding and grounding. The dataset is hosted by author Ssdaizi on Hugging Face, with a preview subset of 700 samples released as of 20260304 and a full training split updated on 20260630.

Use Cases

• Training instruction-following models for ultrasound image analysis based on the described instruction-tuning purpose.
• Developing models for grounding medical concepts in ultrasound images based on the dataset's stated focus on grounding.
• Benchmarking multimodal AI systems on medical imaging tasks based on the dataset's scale and focus on ultrasound understanding.

Strengths

• Preview subset contains 700 samples, providing an initial testbed.
• Full training split was updated on 20260630, indicating active maintenance.

Limitations

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

Provenance

Source

Ssdaizi on Hugging Face

Freshness

Last updated 2026-06-30 04:52:02.