NMIXX: Domain-Adapted Neural Embeddings for Cross-Lingual eXploration of Finance
NMIXX is a suite of cross-lingual embedding models fine-tuned with 18.8K high-confidence triplets derived from financial texts and translations. This approach successfully improved performance on Korean financial similarity benchmarks by gaining +0.22 on KorFinSTS, demonstrating that specialized training can overcome the limitations of general models in low-resource languages like Korean.