Liquid AILiquid Nanos
LFM2.5-Encoder-230M
This 230M multilingual encoder is a masked language model for task-specific fine-tuning.
What the release provides
Full bidirectional attention supports classification, token labels, retrieval and similarity tasks. The family covers 15 languages with an 8K context. These are encoder foundations rather than general chat checkpoints.LFM2.5-Encoder-230M official model card
How to use it well
Practical guidance. Choose a task head and labeled examples that match the intended output. Keep language coverage and domain shift visible when assessing the fine-tuned model.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | LiquidAI/LFM2.5-Encoder-230M |