多模态大模型不止看懂RGB，高光谱成像理解迎来全新基准与免训练推理框架。
arXiv:2604.08884v2 Announce Type: replace-cross Abstract: Multimodal Large Language Models (MLLMs) have achieved strong performance on RGB image under…
让大模型按需调整推理深度，AdaR框架为复杂问题动态分配计算资源，兼顾效率与准确。
arXiv:2510.04617v3 Announce Type: replace Abstract: Mathematical reasoning is a primary indicator of large language models (LLMs) intelligence. Howeve…
大模型理解因果难题？这个智能体框架专攻因果推理，让LLM真正掌握因果逻辑
arXiv:2408.06849v3 Announce Type: replace Abstract: The large language model (LLM) has achieved significant success across various domains. However, t…
OpenForgeRL框架让AI智能体在任何环境中都能进行原生强化学习训练，打破现有推理工具的局限。
arXiv:2607.21557v1 Announce Type: new Abstract: Modern AI agents rely on elaborate inference harnesses such as Claude Code, Codex, and OpenClaw to dri…
提出双假设推理框架增强LLM安全护栏，创新方法兼顾性能与防护。
arXiv:2607.17575v1 Announce Type: new Abstract: We propose ARBITER, a novel LLM guardrail framework that introduces two key ideas: (i) dual-hypothesis…
开源框架为AI Agent注入系统思考能力，用Dart实现可组合的推理模块，适合构建复杂决策系统。
Article URL: https://github.com/asofyan/dart-framework Comments URL: https://news.ycombinator.com/item?id=48973766 Points: 1 # Comments: 0
单卡10GB显存就能跑满血DeepSeek-R1-671B，支持全系N卡/A卡甚至国产卡，MOE混合推理更是绝活。
Article URL: https://github.com/ztxz16/fastllm Comments URL: https://news.ycombinator.com/item?id=48728290 Points: 3 # Comments: 0
给AI装“逻辑拆弹器”：用形式化验证揪出论证中的谬误，可解释性拉满。
arXiv:2606.21867v1 Announce Type: new Abstract: Current evaluations of Large Language Models (LLMs) on logical fallacy detection focus on predicted la…
LLM化身验证器，用证据链提升多跳推理的准确性与可信度
arXiv:2604.01993v2 Announce Type: replace-cross Abstract: Multi-hop QA benchmarks often reward Large Language Models (LLMs) for spurious correctness, …
双轨知识图谱验证框架提升多跳问答推理可靠性，已被ICML 2026收录。
arXiv:2510.16302v2 Announce Type: replace Abstract: Multi-hop reasoning for question answering (QA) plays a critical role in retrieval-augmented gener…
通过视觉推理提升过程奖励建模精度，为复杂任务训练提供新思路。
arXiv:2508.03556v3 Announce Type: replace Abstract: Process Reward Model (PRM) is widely used in the post-training of Large Language Model (LLM) becau…