RecipeMatching D2 (Qwen3-Embedding-0.6B)

A checkpoint from the paper Recipe-Matching, Not Equivalence (Ali Habibullah, Mohammad Alshiekh, Yazan Alshoibi, Salman Khan and Naeemullah Khan, 2026), shared so the paper's evaluations of this model can be re-run without retraining. It is not one of the three models the paper releases as its main artefacts. Code, data and results: https://github.com/KAUST-Academy/recipe-matching-not-equivalence.

What it is. Rung D2 of the paper's rewriter-prompt ladder: Qwen3-Embedding-0.6B fine-tuned (all weights) on 6,145 of the 6,553 rows of data/llm_pairs_paraphrase/pairs.jsonl, LLM rewrites and near-miss negatives written under a close paraphrase of MathNet-Retrieve's published rewriter prompt. Seed 42; the full training record, command line included, is run_config.json in this repository, and every training file named here is in the GitHub repository.

Results of this checkpoint (seed 42, R@1 / R@5 / R@10; the paper reports three-seed means (seeds 42–44)):

Evaluation	R@1	R@5	R@10
MathNet-Retrieve easy tier (15,000 queries, 117,088 documents)	65.29	91.37	94.53
MathNet-Retrieve hard tier	7.89	53.43	65.37
Cross-language duplicates (strict)	66.67	81.42	83.46

Usage. Encode queries with the model's query prompt, the setting every number above uses.

from sentence_transformers import SentenceTransformer

model = SentenceTransformer("KAUSTAcademy/RecipeMatching_D2_Qwen3-Embedding-0.6B")
q = model.encode(["Find all real x with x^2 = 2x."], prompt_name="query")
d = model.encode(["Determine every real solution of x^2 - 2x = 0."])
print(model.similarity(q, d))

Citation

@misc{habibullah2026recipematchingequivalence,
      title={Recipe-Matching, Not Equivalence}, 
      author={Ali Habibullah and Mohammad Alshiekh and Yazan Alshoibi and Salman Khan and Naeemullah Khan},
      year={2026},
      eprint={2609.31927},
      archivePrefix={arXiv},
      primaryClass={cs.IR},
      url={https://arxiv.org/abs/2609.31927}, 
}