Liquid AI explains designing LFMs for phones: less compute, better architecture and data
JosephJacks_ · x · 2026-10-19
Liquid AI's pre-training team, including Adit Tadimeti and Khoi Nguyen, explains that building models that fit on phones or laptops makes pre-training harder: with fewer parameters, quality must come from better architecture and better data rather than more compute. The team designs LFMs with the target device in mind from day one.
More from Infra
- AMD, NVIDIA, and Google All Ship Major vLLM Inference Optimizations in Three Days — PyTorch · 2026-10-20
- XFreeze Claims SpaceX Orbital Data Centers Are the Endgame for Scaling Intelligence — XFreeze · 2026-10-20
- Intel's Crescent Island GPU D270: A 160GB PCIe 5.0 Inference Card — lemire · 2026-10-20
- Mirai's uzu hits 123 tok/s decode on Apple M5 Max 128GB, fastest for Qwen-3.8 27B — teortaxesTex · 2026-10-20
- turbopuffer cuts disk read I/O by up to 80% with smarter vector rescoring — DanielLockyer · 2026-10-20
- vLLM Speculators v0.9.0: 2x faster speculative decoding training, 50% less storage — vllm_project · 2026-10-20