English Hebrew Codeswitching ASR Benchmark
Held-out test sets (v1, v2) the models are scored on; never trained on.
Details
- Type
- Benchmark
- Access
- Gated, automatic: accept the terms on Hugging Face and access is approved instantly.
- Licence
- CC BY 4.0
- Rows
- 324
- Language
- English, Hebrew
- Last updated
- 2026-10-30
- Training use
- Never (evaluation only)
Fields come from the dataset index (built 2026-11-06) and the dataset's Hugging Face card; a field the card does not state is left out. All datasets.