Nemotron 3 Diarization, Core ML low preset (mirror)
A pinned mirror of the files Moxie downloads for live speaker-change detection on Apple silicon.
- Model: NVIDIA Nemotron 3 Diarization (8-speaker streaming Sortformer).
- Core ML conversion: FluidInference/nemotron-3-diarization-coreml at revision
25a90f97f254428d4b30374b76af9c74fdee8327, unchanged. - Contents:
monolithic/v2/Nemotron3Diarizer_low.mlmodelc(1.04 s latency preset) and the root assets FluidAudio loads (learnable_sil_emb.bin,pre_encode_proj_t.bin,config.json). Same layout as upstream, so FluidAudio'sNemotron3Modelscan load it from a local directory. - Licence: OpenMDW-1.1, as upstream. The upstream model card is kept as
UPSTREAM_README.md.
- Downloads last month
- 18
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for volantis-labs/nemotron-3-diarization-coreml
Base model
nvidia/Nemotron-3-Diarization