The dataset viewer is not available because its heuristics could not detect any supported data files. You can try uploading some data files, or configuring the data files location manually.
SWEFoundry Test
Public smoke data for validating the SWEFoundry task-production contract across multiple software-engineering task families.
Included task families
| Task | Family | Behavioral cases | Starter reward | Oracle reward |
|---|---|---|---|---|
build-feature-flags |
nl2repo-lite |
8 | 0.00 | 1.00 |
repair-jsonl-ledger |
synthetic-repair |
8 | 0.25 | 1.00 |
reproduce-url-router |
repo-reproduction |
10 | 0.10 | 1.00 |
Every task includes a natural-language instruction, Harbor task.toml, Docker
environment, Starter workspace, reference solution, and executable verifier.
The verifier emits both a scalar reward.txt and structured reward.json.
data/tasks.jsonl provides one index row per task for Dataset Viewer. Complete
task artifacts are stored under family-smoke-v1/tasks/.
Validation
The independent audit requires:
- Starter reward below 1.0;
- two fresh Oracle runs with reward exactly 1.0;
- identical structured rewards across repeated Oracle runs;
- no network access in the committed Harbor task policy.
The three tasks pass 26 behavioral checks in total. The source repository's test suite passes 36 tests.
Intended use
Use this dataset to test task ingestion, Harbor execution, verifier behavior, trajectory rollout, and filtering infrastructure. These tasks are deliberately small and their reference solutions are public. They must not be treated as a held-out benchmark or used to make contamination-resistant capability claims.
Source
- Downloads last month
- 62