Global Decompression Benchmark
The benchmark compares decompression models against historical dive data with recorded outcomes.
Research use only. Benchmark results are not dive plans or operational decompression guidance.
How to read a result
Each result covers one model version and configuration. It does not extend to the wider model family.
A score means little on its own. Check which dataset produced it, how that dataset defines an outcome such as decompression sickness, and how many such outcomes it records.
A model tuned on the same data it is scored against will tend to score higher than one tested on independent data. These two cases are labelled separately.
Reproducibility
Each published result will list the exact datasets, code version and model versions used to produce it, so the run can be repeated.