Downloads · 30 days
0
lelegu/omni-router-speechcrawl-streaming-asr-0.6b-v1
omni-router-speechcrawl-streaming-asr-0.6b-v1 is a automatic speech recognition model from lelegu. Use it when you need speech turned into text. The card lists the license as apple-amlr.
Omni-router Transformer is a new Mixture-of-Experts (MoE) architecture that explicitly couples routing across layers using a shared router to learn strong and specialized experts. Omni-router's routing decisions appea…
Downloads · 30 days
0
Access
Public
Updated Oct 15, 2025
Repo size
14.7 GB
Likes
1
Public
Click a slice to open those files.
.pth7.4 GB · 100%
From the Hugging Face model README
Omni-router Transformer is a new Mixture-of-Experts (MoE) architecture that explicitly couples routing across layers using a shared router to learn strong and specialized experts. Omni-router's routing decisions appear to form consistent temporal segments and strutured usage across model depth, suggesting meaningful coordination between layers. Please refer to the paper for details.
This ASR model is a 4-expert MoE model (total 613M with 200M activate parameters). The model is streaming which transcribes speech conditioned only on past and current speech.
This model is a speech recognition model.
Please refer to the github page for detailed usage.
The training data is a large-scale conversational audio dataset collected from publicly accessible sources, named SpeechCrawl. Please refer to the paper for details.
If you find this work useful, please cite our paper:
@article{gu2025omnirouter,
title={Omni-router: Sharing Routing Decisions in Sparse Mixture-of-Experts for Speech Recognition},
author={Gu, Zijin and Likhomanenko, Tatiana and Jaitly, Navdeep},
journal={arXiv preprint arXiv:2507.05724},
year={2025}
}
Contact [email protected] for any issues.