Open Source

Contributions to major NLP open-source projects.

Sentence-Transformers (2025-2026)

  • Extended the cross-encoder training stack with classic learning-to-rank losses. [PR #6] [PR #7]
  • Implemented EmbedDistillLoss to support direct embedding-level knowledge distillation. [PR #3665]
  • Introduced hardness-weighted contrastive learning to up-weight informative hard negatives. [PR #3667]
  • Implemented CachedSpladeLoss for gradient-cache compatible, memory-efficient SPLADE training. [PR #3670]

MTEB - Massive Text Embeddings Benchmark (2024-2025)

  • Added a Korean retrieval benchmark task to expand multilingual retrieval evaluation. [PR #1388]
  • Improved stability of the OpenAI embeddings by adding sentence trimming. [PR #1526]
  • Fixed NaN embeddings for Jasper models by switching model precision. [PR #2481]

InstructKR (2026)

  • Led the Korean Reranker evaluation and leaderboard project. [GitHub]

FlagEmbedding

  • Fixed a bug related to knowledge distillation when training. [Issue#1170]