LFM2.5-2.6B β€” Core AI closures for iPhone and iPad

Ahead-of-time compiled closures of mlboydaisuke/LFM2.5-2.6B-CoreAI (revision 0e43b1a408515f3937a20a5bfa0e25d8f24f9704, int8 block-32 export of LiquidAI/LFM2.5-2.6B) for Apple Core AI on iOS 27.2+. The source .aimodel cannot be compiled on an 8 GB device, so each folder holds the closure for one iOS GPU architecture, compiled on a Mac with coreai-build (--platform iOS --min-deployment-version 27.2 --preferred-compute gpu --expect-frequent-reshapes):

h13g M1, h14g M2, h15g M3, h16g M4, h16p A17 Pro, h17g M5, h17p A18 / A18 Pro, h18p A19 / A19 Pro, h19p A20 Pro.

Each folder carries a descriptor.json with the size and SHA-256 of every file. Tokenizer files are the source export's, unchanged. Only the h14g (M2) closure has been run on a physical device; the others are compiled from the same verified source without device inference evidence.

License: LFM Open License v1.0 (file LICENSE in every folder), which permits redistribution of derivatives. Not an official Liquid AI release.

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for wolfram-pro/lfm2.5-2.6b-coreai-ios

Finetuned
(75)
this model