AIHugging Faceabout 2 hours ago
Liquid AI Releases DSpark Drafter for Vision Models
4 min read
Liquid AI released an experimental DSpark draft model for its LFM2.5-VL-3B vision-language model, adding speculative decoding that speeds inference up to 3.13x on device and 2.66x on an H100 without changing output quality. The drafter adds 280M parameters, 8.9% over the 3B target, and ships with day-one support for llama.cpp, MLX-VLM and SGLang.
Level
Hype check
Tap to vote and see what everyone thinks.
