“A CaLLM Judge is a Better Judge: Mitigating Bias in LLM-as-a-judge” is on arXiv
Our work on mitigating bias in LLM-as-a-judge systems is now available on arXiv.
•
1 min read
Our work on mitigating bias in LLM-as-a-judge systems is now available on arXiv.
Our work on learning when to selectively disclose information to decision-makers is now available on arXiv.
Our work on stable local calibration via discrete representations was accepted at NeurIPS 2026.
Presented our work on interpretable, rule-based label exclusion with conformal prediction at the HLDM workshop at ECML PKDD 2026.
At AISTATS 2026 I presented and discussed three of our calibration papers during the poster sessions.
A preliminary version of our work on stable local calibration via discrete representations was accepted at the AISTATS 2026 workshop on trustworthy calibration.
Our work on global and local calibration in graph neural networks was accepted at the AISTATS 2026 workshop on trustworthy calibration.
Our work on consistent multiclass local calibration with LocalNets was accepted at AISTATS 2026.
Our work on consistent multiclass local calibration with LocalNets is now available on arXiv.
Our work on cross-lingual disparities in multilingual LLM-based healthcare Q&A is now available on arXiv.