Jiajun Fan

ICLR 2026

Incentivizing Consistent, Effective and Scalable Reasoning Capability in Audio LLMs via Reasoning Process Rewards

In brief

What the paper does.

We propose CESAR, an online RL framework (GRPO) with multi-faceted reasoning process rewards incentivizing consistency, structured analytical patterns, and calibrated depth. Resolves test-time inverse scaling in Audio LLMs; achieves SOTA on MMAU Test-mini, substantially outperforming Gemini 2.5 Pro and GPT-4o Audio.Also featured on Amazon Science.

Published in
International Conference on Learning Representations 2026 (ICLR 2026)
Paper
openreview.net/forum?id=DUr48hxO2h
Project page
jiajunfan.com/projects/cesar/

Cite

Cite the paper.

@inproceedings{fan2026incentivizing,
  title     = {Incentivizing Consistent, Effective and Scalable Reasoning
  Capability in Audio {LLM}s via Reasoning Process Rewards},
  author    = {Jiajun Fan and Roger Ren and Jingyuan Li and Rahul Pandey and
  Prashanth Gurunath Shivakumar and Ivan Bulyko and Ankur Gandhe
  and Ge Liu and Yile Gu},
  booktitle = {The Fourteenth International Conference on Learning Representations},
  year      = {2026},
  url       = {https://openreview.net/forum?id=DUr48hxO2h}
}