This dataset stores the leaderboard results for the Rabbinic Hebrew/Aramaic Embedding Benchmark.
The leaderboard.json file contains an array of evaluation results:
[
{
"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… See the full description on the dataset page:
https://huggingface.co/datasets/Sefaria/Rabbinic-Embedding-Leaderboard.