Google's EmbeddingGemma 2 loads only the encoders you need: 270M text, 170M vision, 300M audio
HankYeomans · x · 2026-10-07
EmbeddingGemma 2 has a modular 740M footprint: 270M text, 170M vision, 300M audio. You can disable unused encoders via configkwargs — text-only loads 270M, text+images 440M, text+audio 570M. The design targets phones and laptops, loading only the encoders each app needs.
More from Infra
- Serving & Monitoring Local LLMs Across Three Mixed-GPU Machines Without Duct Tape — ziyaulhuk12 · 2026-10-07
- Disaggregated inference is the future, says e/acc's Beff Jezos after panel with General Compute — beffjezos · 2026-10-07
- What Does 'Owning Your Own AI' Technically Mean? Locally Running Open Weights Debated — Imaginary_Choice_430 · 2026-10-07
- NVIDIA's LoGRA cuts LLM RL training memory by up to 45.7%, enables 27B RL on a single 8-GPU node — nvidia · 2026-10-07
- Ex-NVIDIA engineer tells the story of testing a chip with a broken memory controller — blelbach · 2026-10-07
- Follow the money risk: memory suppliers take deposits, AI chip makers finance their customers — tengyanAI · 2026-10-07