LFM2.5-Embedding-350M & LFM2.5-ColBERT-350M
LFM2.5-Embedding-350M and LFM2.5-ColBERT-350M are two new models released for fast multilingual retrieval and ranking. LFM2.5-Embedding-350M achieves best-in-class multilingual accuracy as a dense embedder of its size, with inference speed comparable to smaller models. These models can be used as drop-in replacements in existing RAG pipelines for cross-lingual search across 11 languages.
Key takeaways
- Best-in-class multilingual accuracy for a dense embedder of its size.
- Inference speed comparable to much smaller models.
- Can be used as a drop-in replacement in RAG pipelines.