5 days ago
-
Card: document the experimental ane/ build (FluidAudio #49)
Adds the Experimental: Neural Engine LM section and a Changelog row for ane/ (HF commit e5259314). -
Add experimental all-ANE LM build in ane/ (FluidAudio #49)
LFM3-Audio-2B backbone (2 x 12 layers) + head, LUT8 group 8, decode q=1, span 1024; 100% Neural Engine. Not loaded by FluidAudio. GPU pack context: https://github.com/FluidInference/FluidAudio/pull/1076 , cold-load hardening https://github.com/FluidInference/FluidAudio/pull/1078
6 days ago
-
Card: Core AI repo pointer, TTS RTFx range (FluidAudio #49)
https://github.com/FluidInference/FluidAudio/pull/1076 -
Card: link FluidAudio PR #1076 (FluidAudio #49)
https://github.com/FluidInference/FluidAudio/pull/1076 -
LFM3-Audio-2B Core ML: initial release for FluidAudio LFM3AudioManager (FluidAudio #49)
ASR, LiveASR and TTS for FluidAudio (FluidInference/FluidAudio branch feat/lfm3-audio, PR to follow). Request: https://github.com/FluidInference/FluidAudio/issues/49#issuecomment-6501658263. Source port: model-lab models/lfm3. -
initial commit
