Estimating the correctness of large language model (LLM) predictions without access to ground truth is critical for reliable deployment in high-stakes settin…
机构:华为
来源:arXiv 2610.34537 | AI4Papers 论文推荐平台