Summary

Adds Shuu12121/NightOwl-CodeEmbedding to the MTEB model registry.

NightOwl-CodeEmbedding is a Sentence Transformers-compatible ModernBERT model specialized for code retrieval. It uses CLS pooling, cosine similarity, and does not require query or document prefixes.

Evaluation

The model was evaluated on 12 representative code-retrieval tasks using MTEB. The macro-average NDCG@10 was 0.70240.

There is currently no original paper associated with this model. Detailed benchmark results are available in the model card.

For CodeEditSearch-like training data, I used a custom dataset derived from bigcode/commitpackft. Rows overlapping with cassanof/CodeEditSearch were excluded using content-, diff-, commit-, and repo/commit-based hashes. If CodeEditSearch should still be listed in the model metadata as related training data, I would be happy to add it.

Checklist

• I have filled out the ModelMeta object to the extent possible

• I have ensured that my model can be loaded using

  • mteb.get_model(model_name, revision)
  • mteb.get_model_meta(model_name, revision)

• I have tested the implementation works on a representative set of tasks

• The model is public, i.e., is available either as an API or the weights are publicly available to download

• I reproduced results from the original paper (if applicable) on at least one benchmark, and I am including the results in the PR description

  • Not applicable: there is currently no original paper associated with this model.