Liquid AI Releases LFM2.5-DSpark Draft Models That Deliver Up to 3.18x Faster Decoding Without Changing Model Outputs
Liquid AI has launched DSpark draft mannequin checkpoints for 3 fashions in its LFM2.5 household: LFM2.5-1.2B-Instruct, LFM2.5-2.6B, and LFM2.5-8B-A1B. Each drafter provides a speculative decoding path to an current goal mannequin. A roughly 300M-parameter draft proposes a block of 9 candidate tokens, and the goal mannequin verifies the entire block in a single ahead move….
