Zero-shot text classification: TypeSafe Jev vs BERT-family zero-shot (NLI, embeddings) on 6 public tasks + PAWS pairs, with label-equivalence curves and an arXiv contamination control
Stars measure the whole repository, including work unrelated to Jev.
Documented records the linked documentation or source. JevHunt has not independently run or benchmarked this project.
| `jev` | `typesafe/jev-1.13-20260917` via OpenRouter `/api/alpha/decisions`, one `choice` question per text, 20 texts per call. Banking77: pick one of 11 groups, then a label inside it. |
Evidence commit: edbf0713583644bd3f47299fd6b104a8b2073219
Discovered through: github-search.
How we review →