LiquidAI releases two small multilingual retrieval models
LiquidAI released two 350 million parameter retrieval models based on LFM2.5. LFM2.5-Embedding-350M creates one vector for each document, which keeps the small and fast. It is designed for search across 11 languages and is described as highly accurate for a dense of its size.
LFM2.5-ColBERT-350M stores a vector for each token and compares questions with documents using MaxSim, which aims to improve matching accuracy. This can let a system store a document in one language, such as an English product description, and retrieve it from questions in other languages. Both models are described as having similar to much smaller models because of the efficient LFM2 backbone.
GGUF versions are available on and are meant to be used as replacements in existing .
Key points
- LFM2.5-Embedding-350M makes one vector per document for fast multilingual search.
- LFM2.5-ColBERT-350M uses token-level vectors and MaxSim for more detailed matching.
- Both models are presented as supporting retrieval across 11 languages.
- They are as drop-in replacements for existing .
- GGUF model files are available on for local use.