别只调温度了！双层优化为LLM校准带来更精准的建模方案。
arXiv:2608.07419v1 Announce Type: new Abstract: Preference alignment often makes large language models (LLMs) overconfident and poorly calibrated. Tra…
破解AI推理中的病态思维链，这篇论文为提升大模型可靠性提供诊断新方法。
arXiv:2602.13904v2 Announce Type: replace Abstract: Chain-of-thought (CoT) reasoning is fundamental to modern LLM architectures and represents a criti…
揭秘LLM作为评判者时的位置偏差，点式与对式评估究竟何者更准确？
arXiv:2602.02219v2 Announce Type: replace Abstract: Large language models are widely employed as evaluators, a paradigm commonly referred to as LLM-as…
针对大语言模型在知识图谱推理中易产生幻觉的痛点，提出了一套新的检测框架与评估方法。
arXiv:2606.19351v1 Announce Type: cross Abstract: Knowledge graph (KG) reasoning infers new knowledge from existing facts and is widely applied in que…
大模型能否精准识别失语症话语中的信息单元？这项研究评估指令调优LLM在临床语篇评估中的可靠性。
arXiv:2606.15696v1 Announce Type: new Abstract: Correct Information Units (CIUs) are central to discourse assessment in aphasia because they quantify …
自动化诊断大模型失败原因的新框架，用原理性方法定位LLM错误根源。
arXiv:2602.12966v2 Announce Type: replace Abstract: Understanding how and why large language models (LLMs) fail is becoming a central challenge as mod…
跨释义不变性学习新方法精准检测大模型幻觉，已被ICASSP 2026接收。
arXiv:2606.08157v1 Announce Type: new Abstract: Large language models (LLMs) frequently generate hallucinations, which are unsupported by a source doc…
用掷骰子测试LLM的“概率直觉”，揭示大模型在简单随机任务中的可靠性短板。
arXiv:2606.07515v1 Announce Type: cross Abstract: We investigate the probabilistic reasoning capabilities of large language models through a controlle…
语音大模型推理中实体绑定出错了怎么办？这篇论文系统诊断问题并提出思维链干预方案。
arXiv:2606.04474v1 Announce Type: new Abstract: Speech Large Language Models (SLLMs) underperform their text counterparts on complex reasoning. We rev…
探讨AI自主agent能否突破人类最后验证防线，引发对智能体安全边界的深度思考。
arXiv:2606.02449v1 Announce Type: cross Abstract: Multimodal agents are increasingly expected to operate interfaces on behalf of users, raising a cent…
用3.8B小模型实现7倍参数大模型的数据解析可靠性，开源免费可即用！
https://github.com/pcoz/llm-feedback-control Comments URL: https://news.ycombinator.com/item?id=48359939 Points: 2 # Comments: 1
Claude直接执行任务完美，但生成的代码却屡屡翻车，揭示AI模型在“做”与“教”之间的行为鸿沟。
I asked Claude to do some heavy work, and it was done perfectly. When I asked Claude to write a Python script to do the same task with proper prompts …