Summary

  • Liquid AI has generated LFM2-VL, a new generation of vision-language foundation models designed to work across a variety of hardware, from smartphones to laptops to embedded systems.
  • Linear Input-Varying systems mean the models can generate weights or model settings on the fly for each input and extend to support both text and image inputs.
  • Two variants were released: LFM2-VL-450M, a hyper-efficient model, and LFM2-VL-1.6B, a more capable model.
  • Liquid AI’s flagship innovation, the Liquid Foundation Models (LFMs), are based on principles from dynamical systems, signal processing, and numerical linear algebra, and can handle multiple forms of data.
  • The company’s Liquid Edge AI Platform (LEAP) cross-platform SDK allows developers to run small language models directly on mobile and embedded devices.
  • LFM2-VL models are available on Hugging Face, along with example fine-tuning code in Colab, and are compatible with Hugging Face transformers and TRL.

By Carl Franzen

Original Article