diallm-llama-sft-aus

DiaLLM — Llama 3.1-8B — en-AU — SFT

Built with Llama.

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

• Base model: Llama 3.1-8B
• Target variety: en-AU (Australian English)
• Adaptation thread: explicit (variety-targeted)
• Pipeline stage: SFT (from the CPT checkpoint)

Fine-tuned from jordanpainter/diallm-llama-cpt on the dialect-specific preference dataset for en-AU (Australian English), so the chosen completions already contain Multi-VALUE dialectal features.

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

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

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}
}