@article{inferencetimeconsensusformitigatinghidde, title = {Inference-Time Consensus for Mitigating Hidden Behaviors from LLM Fine-Tuning}, author = {Adhyyan Narang and Artin Tajdini and Claire Zhang and Jamie Morgenstern}, year = {2026}, eprint = {2607.23394}, archivePrefix = {arXiv}, url = {https://arxiv.org/abs/2607.23394}, }