Topic
Speculative Decoding
2 stories tagged Speculative Decoding, newest first.
Sep 24, 2026, 2:08 PMArtificial Intelligence
Liquid AI LFM2.5-VL-DSpark Speeds Vision AI Decoding Up to 3.13x
Liquid AI's 280M-parameter LFM2.5-VL-DSpark accelerates LFM2.5-VL-3B decoding up to 3.13x on devices and 2.66x on H100 with exact output across three runtimes.
DateSectionHeadlineSource
Aug 20
Artificial Intelligence
Liquid AI DSpark Speeds LFM2.5 Inference Up to 3.18x
Hugging Face - Blog
▲