Instructions to use littlebearlabs/witness-resnet293-lm-mlx with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- MLX
How to use littlebearlabs/witness-resnet293-lm-mlx with MLX:
# Download the model from the Hub pip install huggingface_hub[hf_xet] huggingface-cli download --local-dir witness-resnet293-lm-mlx littlebearlabs/witness-resnet293-lm-mlx
- Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- LM Studio
metadata
license: cc-by-4.0
library_name: mlx
tags:
- mlx
- speaker-diarization
- witness
witness-resnet293-lm-mlx
WeSpeaker VoxCeleb ResNet293-LM speaker embedder (256-d), converted to MLX safetensors for witness's on-device speaker diarization.
Upstream attribution
- Model: WeSpeaker VoxCeleb ResNet293-LM (deep Bottleneck ResNet → TSTP pooling → linear head → L2-norm).
- Authors / source: WeSpeaker —
Wespeaker/wespeaker-voxceleb-resnet293-LM. - License: CC-BY-4.0 (inherited from the WeSpeaker VoxCeleb checkpoint).
What's in this repo
model.safetensors— converted from the upstream WeSpeaker checkpoint by.research/diarization/gen_resnet293_embed_fixture.py. 256-d, L2-normalized by the witness crate after the linear head.
Converted to MLX for witness, an
open-source Rust toolkit for on-device system capture on macOS. Generated by
.research/diarization/publish_weights.sh.