jordanpainter/diallm-gemma-grpo-all

DiaLLM — Gemma 3-4B-it — Broad — GRPO

Built with Gemma.

Part of DiaLLM: An Investigation into the Robustness-Generation Gap in English Dialect Adaptation (EMNLP 2026 Main).

• Base model: Gemma 3-4B-it
• Target variety: broad — no variety targeting
• Adaptation thread: implicit (broad)
• Alignment method: GRPO

Fine-tuned from jordanpainter/diallm-gemma-sft-all via GRPO with the pooled, variety-agnostic preference set.

Code, checkpoints, preference datasets, linguistic-analysis toolkit: https://github.com/surrey-nlp/diallm

Paper: https://arxiv.org/abs/2607.07669

This model is a fine-tuned version of jordanpainter/diallm-gemma-sft-all, trained using TRL.

Training procedure

This model was trained with GRPO.

Framework versions

• TRL: 0.28.0
• Transformers: 4.57.6
• Pytorch: 2.5.1+cu121
• Datasets: 4.5.0
• Tokenizers: 0.22.2

Citation

@article{painter2026diallm,
  title     = {DiaLLM: An Investigation into the Robustness-Generation Gap in English Dialect Adaptation},
  author    = {Painter, Jordan and Srirag, Dipankar and Kappiyath, Adarsh and Kanojia, Diptesh and Joshi, Aditya and Yin, Lu},
  year      = {2026},
  eprint    = {2607.07669},
  archivePrefix = {arXiv}
}

@article{shao2024deepseekmath,
    title  = {{DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Models}},
    author = {Zhihong Shao and Peiyi Wang and Qihao Zhu and Runxin Xu and Junxiao Song and Mingchuan Zhang and Y. K. Li and Y. Wu and Daya Guo},
    year   = 2024,
    eprint = {arXiv:2402.03300}
}

@software{vonwerra2020trl,
  title   = {{TRL: Transformers Reinforcement Learning}},
  author  = {von Werra, Leandro and Belkada, Younes and Tunstall, Lewis and Beeching, Edward and Thrush, Tristan and Lambert, Nathan and Huang, Shengyi and Rasul, Kashif and Gallouédec, Quentin},
  license = {Apache-2.0},
  url     = {https://github.com/huggingface/trl},
  year    = {2020}
}