High-throughput synthetic and pretraining dataset sifter for TypeSafe Jev. Rust streaming core, Parquet and JSONL I/O, typed Choice/Score/Noul judgments, speculative fan-out, 24.0 rows/sec measured.
lejudge-jev-jepa: Natural-language constraints for JEPA
AbdelStark/lejudge-jev-jepa
VerifiedResearchDocumented
lejudge-jev-jepa: Natural-language constraints for JEPA world-model planning, judged by a decision model instead of an LLM. See Jev references and source code.
Repository description: Natural-language constraints for JEPA world-model planning, judged by a decision model instead of an LLM.
Get started
Clone the repository and follow its README for install and usage. JevHunt has not run this project.
git clone https://github.com/AbdelStark/lejudge-jev-jepa.gitProject facts
- Relationship to Jev
- Research
- Evidence
- Documented
- Language
- Python
- License
- MIT
- Origin
- Original repository
- Repository status
- Not archived
- Created
- 2026-09-22
- Last code update
- 2026-09-24
- GitHub stars
- 10
- Evidence checked
- Sep 28, 2026
- Metadata checked
- Sep 28, 2026
- Check status
- Up to date
Stars measure the whole repository, including work unrelated to Jev.
Evidence and scope
Documented records the linked documentation or source. JevHunt has not independently run or benchmarked this project.
`LEJUDGE_MODE=offline` raises on any cache miss, so the commands above make **zero** API calls, and re-running `make paper` on unchanged results leaves every figure byte-identical. To regenerate the studies from scratch (needs `TYPESAFE_API_KEY` and `LEJUDGE_MODE=live`):
Evidence commit: b2e578fb4d0ca0826ca1669bbd4bd73cdbca8ec5
Discovered through: awesome-jev, typesafe-field-guide.
How we review →Suggest a correction · Claim this listing · Submit a project