---
tags:
- benchmark
- leaderboard
- embedding
- hebrew
- rabbinic
license: mit
---

# Rabbinic Embedding Benchmark Leaderboard

This dataset stores the leaderboard results for the [Rabbinic Hebrew/Aramaic Embedding Benchmark](https://huggingface.co/spaces/Sefaria/Rabbinic-Embedding-Bench).

## Structure

The `leaderboard.json` file contains an array of evaluation results:

```json
[
  {
    "model_id": "model-org/model-name",
    "model_name": "Model Display Name",
    "mrr": 0.85,
    "recall_at_1": 0.75,
    "recall_at_5": 0.90,
    "recall_at_10": 0.95,
    "bitext_accuracy": 0.92,
    "avg_true_pair_similarity": 0.85,
    "avg_random_pair_similarity": 0.35,
    "num_pairs": 3708,
    "timestamp": "2024-01-15T12:00:00",
    "categories": {"Talmud": 480, "Mishnah": 789, ...}
  }
]
```

## Related

- [Benchmark Dataset](https://huggingface.co/datasets/Sefaria/Rabbinic-Hebrew-English-Pairs)
- [Benchmark Space](https://huggingface.co/spaces/Sefaria/Rabbinic-Embedding-Benchmark)