Research
Data and code
The evaluation harness, run records and reporting format behind the published results.
Results published here are backed by code and records you can inspect. This page is the index to them.
What is available
| Item | What it is | Status |
|---|---|---|
| Evaluation harness | The runner and task definitions behind the published results | Released |
| Run records | Raw per-trial records, including excluded runs | Released |
| Reporting format | The schema the run records use | Released |
Licence
Open licence, reuse with attribution, unless a specific item says otherwise. If you publish results produced with this code, attaching your run records makes the work more useful to whoever comes next.