Modelpublic

dchevell/diar_streaming_sortformer_4spk-v2.1-GGUF

sourceHugging Faceotherupdated 1mo agoView on Hugging Face

0likes85downloads

Model Card

Streaming Sortformer 4-speaker v2.1 — NeMo-Speech.cpp GGUF

This repository contains an F32 GGUF conversion of `nvidia/diar_streaming_sortformer_4spk-v2.1` for NVIDIA's `NeMo-Speech.cpp` runtime.

It is intended for NeMo-Speech.cpp v0.1.0. The GGUF schema is runtime-specific and is not interchangeable with conversions made for OpenResearchTools Engine.

Artifact provenance

Item	Value
Source model	`nvidia/diar_streaming_sortformer_4spk-v2.1`
Source revision	`fafaab5faa1617a0ca52d38dd3dc4bd636800d3d`
Source `.nemo` SHA-256	`8abd32832159c6ac1148c926b7276f35ba34582c444e559dce1f1253fea42ef8`
Converter	`NVIDIA/NeMo-Speech.cpp` `v0.1.0` (`4f9676226f667d14608487df744f375db87127f8`)
Output type	F32
Output size	491,094,752 bytes
Output SHA-256	`901aee2822050fe639bb805b8f72067dfd5cd89b952c88ec4dce80ab740ec807`

The conversion used NVIDIA's supported entry point:

python convert_model.py diar_streaming_sortformer_4spk-v2.1.nemo \
  --outfile diar_streaming_sortformer_4spk-v2.1.f32.gguf \
  --outtype f32

The resulting artifact is used by a production NeMo-Speech.cpp v0.1.0 diarization service on an NVIDIA RTX PRO 6000 Blackwell GPU.

License

The source model and this converted artifact are governed by the NVIDIA Open Model License Agreement. The repository includes a copy of that agreement and the attribution notice required for redistribution.