Saltar al contenido
ModelPublicadoExterno
Browse files

SDiaReward 3B

SDiaReward-3B is a lightweight reward model for spoken dialogue, built on Qwen2.5-Omni-3B with a pooling layer and a linear scoring head (QwenOmniThinkerReward). Given a multi-turn conversation containing interleaved...

Fuentehuggingface.co/MYJOKERML/SDiaReward-3B/tree/0e36b88616a70352c0d913d6b6928f61c78b782e
TipoOtherModelo baseQwen/Qwen2.5-Omni-3BVersiónRevision 0e36b88Bcalidad

Fuente del modelo

Tipo
Other
Tamaño del archivo
4,11–4,65 GB
Modelo base
Qwen/Qwen2.5-Omni-3B
Versión
vRevision 0e36b88
Actualizado
31 oct 2026
Descargas
32
Fuente
Hugging Face

Descripción de la fuente

SDiaReward-3B

SDiaReward-3B is a lightweight reward model for spoken dialogue, built on Qwen2.5-Omni-3B with a pooling layer and a linear scoring head (QwenOmniThinkerReward). Given a multi-turn conversation containing interleaved speech and text, it produces a scalar reward reflecting two qualities:

  • Modality-awareness — prosody, emotion, and acoustic naturalness (real human speech vs. synthetic TTS).

  • Colloquialness — spontaneous spoken style vs. scripted written style.

  • 📄 Paper: Modeling and Benchmarking Spoken Dialogue Rewards with Modality and Colloquialness (ACL 2026 Main Conference) — arXiv:2603.14889

  • 💻 Code: https://github.com/MM-Speech/SDiaReward

  • 📚 Training data (gated): SDiaReward dataset.

  • 🧩 Larger variant: SDiaReward-7B.

Evaluation

On the ESDR-Bench validation set:

ModelAccuracyEval lossMargin
SDiaReward-7B0.9710.3581.167
SDiaReward-3B0.9170.4190.869

Usage

This checkpoint uses a custom reward architecture (QwenOmniThinkerReward: a Qwen2.5-Omni backbone with a pooling layer and a scalar reward head). The custom modeling code is not bundled in this repository, so the weights cannot be loaded with a plain AutoModel/pipeline call. To load the model and score conversations, follow the loading and inference instructions in the official code repository:

👉 https://github.com/MM-Speech/SDiaReward (checkpoint id: MYJOKERML/SDiaReward-3B)

Training

Trained with TRL's reward trainer on the SDiaReward preference dataset (11,630 episode-level preference pairs, ~200 hours of paired speech). Backbone: Qwen2.5-Omni-3B; pooling variant: mean_center; 2 epochs.

License & intended use

Released under Apache-2.0 for research use. The reward signal is intended for evaluating and improving spoken-dialogue systems; it is not a safety classifier.

Citation

@article{lu2026modeling,
  title={Modeling and benchmarking spoken dialogue rewards with modality and colloquialness},
  author={Lu, Jingyu and Wang, Yuhan and Zhuo, Fan and Cheng, Xize and Pan, Changhao and Pu, Xueyi and Chen, Yifu and Wen, Chenyuhao and Liang, Tianle and Zhao, Zhou},
  journal={arXiv preprint arXiv:2603.14889},
  year={2026}
}

Fuentes

1 fuente
Hugging FacePrincipalVersión 0e36b88616a70352c0d913d6b6928f61c78b782e
huggingface.co/MYJOKERML/SDiaReward-3B/tree/0e36b88616a70352c0d913d6b6928f61c78b782e

Verificado 31 oct

Download files

Model · 2 weight files

All files listed in the provider index are present.

Repository files

File inventory only; completeness is unverified.

Artefactos del modelo

2 artefactos

model-00001-of-00002.safetensors

safetensors · 4,65 GB · SHA-256 e61f7f60ab64…da57 · Hugging Face

Download shard

model-00002-of-00002.safetensors

safetensors · 4,11 GB · SHA-256 117d5403cbd9…c30c · Hugging Face

Download shard

Extractos de fuentes

3 extractos
Readme15 agohuggingface.co/MYJOKERML/SDiaReward-3B

--- license: apache-2.0 language: - en base_model: - Qwen/Qwen2.5-Omni-3B pipeline_tag: audio-classification library_name: transformers tags: - reward-model - spoken-dialogue - speech - preference - qwen2_5_omni - modality-awareness - colloquialness --- # SDiaReward-3B **SDiaReward-3B** is a lightweight reward model for **spoken dialogue**, built on **Qwen2.5-Omni-3B** with a pooling layer and a linear scoring head (`QwenOmniThinkerReward`). Given a multi-turn conversation containing interleaved speech and text, it produces a scalar reward reflecting two qualities: - **Modality-awareness** — prosody, emotion, and acoustic naturalness (real human speech vs. synthetic TTS). - **Colloquialness** — spontaneous *spoken* style vs. scripted *written* style. - 📄 Paper: *Modeling and Benchmarking Spoken Dialogue Rewards with Modality and Colloquialness* (ACL 2026 Main Conference) — [arXiv:2603.14889](https://arxiv.org/abs/2603.14889) - 💻 Code: https://github.com/MM-Speech/SDiaReward - 📚 Training data (gated): [`SDiaReward` dataset](https://huggingface.co/datasets/MYJOKERML/SDiaReward). - 🧩 Larger variant: [`SDiaReward-7B`](https://huggingface.co/MYJOKERML/SDiaReward-7B). ## Evaluation On the ESDR-Bench validation set: | Model | Accuracy | Eval loss | Margin | |---|---|---|---| | SDiaReward-7B | 0.971 | 0.358 | 1.167 | | **SDiaReward-3B** | **0.917** | 0.419 | 0.869 | ## Usage This checkpoint uses a **custom reward architecture** (`QwenOmniThinkerReward`: a Qwen2.5-Omni backbone with a pooling layer and a scalar reward head). The custom modeling code is **not** bundled in this repository, so the weights cannot be loaded with a plain `AutoModel`/`pipeline` call. To load the model and score conversations, follow the loading and inference instructions in the official code repository: 👉 **https://github.com/MM-Speech/SDiaReward** (checkpoint id: `MYJOKERML/SDiaReward-3B`) ## Training Trained with TRL's reward trainer on the [SDiaReward preference dataset](https://huggingface.co/datasets/MYJOKERML/SDiaReward) (11,630 episode-level preference pairs, ~200 hours of paired speech). Backbone: Qwen2.5-Omni-3B; pooling variant: `mean_center`; 2 epochs. ## License & intended use Released under Apache-2.0 for research use. The reward signal is intended for evaluating and improving spoken-dialogue systems; it is not a safety classifier. ## Citation ```bibtex...

Note31 octhuggingface.co/MYJOKERML/SDiaReward-3B

Source context: 32 downloads · 0 likes · Pipeline audio-classification · Library transformers · Repo MYJOKERML/SDiaReward-3B

Note15 agohuggingface.co/MYJOKERML/SDiaReward-3B

Source context: 8 downloads · 0 likes · Pipeline audio-classification · Library transformers · Repo MYJOKERML/SDiaReward-3B