This checkpoint was trained by MAGO (https://huggingface.co/mago-ai).

Weights
  Derived from PersonaPlex (nvidia/personaplex-7b-v1),
  Copyright NVIDIA Corporation & Affiliates.
  https://huggingface.co/nvidia/personaplex-7b-v1
  License: NVIDIA Open Model License.
  https://www.nvidia.com/en-us/agreements/enterprise-software/nvidia-open-model-license/

  PersonaPlex is a finetune of Moshi by Kyutai,
  Copyright Kyutai. License: CC-BY-4.0.
  https://github.com/kyutai-labs/moshi
  https://arxiv.org/abs/2410.00037

Training speech
  Dataset v4 audio was synthesized with Boson AI Higgs Audio v3 TTS 4B
  (also called Higgs TTS 3), using the Transformers port
  multimodalart/higgs-audio-v3-tts-4b-transformers.
  Official weights: https://huggingface.co/bosonai/higgs-tts-3-4b

  Derived from Higgs TTS 3, licensed from Boson AI USA, Inc.
  Copyright Boson AI USA, Inc. All Rights Reserved.
  License text: LICENSE.higgs in this repository.

Dialogue text for that dataset was drafted with Qwen2.5-3B-Instruct
(Qwen/Qwen2.5-3B-Instruct, Apache-2.0).

If you use this checkpoint, please also cite PersonaPlex:

@misc{roy2026personaplexvoicerolecontrol,
  title={PersonaPlex: Voice and Role Control for Full Duplex Conversational Speech Models}, 
  author={Rajarshi Roy and Jonathan Raiman and Sang-gil Lee and Teodor-Dumitru Ene and Robert Kirby and Sungwon Kim and Jaehyeon Kim and Bryan Catanzaro},
  year={2026},
  eprint={2602.06053},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url={https://arxiv.org/abs/2602.06053}, 
}