1
IBM 量子处理器仅用 19 秒完成百万次采样，挑战超级计算机 110 年任务
IT之家 9 月 30 日消息，美国 BlueQubit 研究团队利用 IBM 最新的 Nighthawk r2 量子处理器完成了一项随机量子线路采样实验，仅用 19 秒就生成了 100 万个样本。BlueQubit 研究人员估计，用 Frontier 超级计算机复现同等任务约需 110 年。 该实…
2
iQOO16 手机行业首发 Q-flex 操控技术：至高 5000Hz 瞬时触控采样率，陀螺仪采样率最高 500Hz
IT之家 9 月 29 日现场消息，iQOO 今天召开新品发布会，推出 iQOO 16 手机。据官方介绍，iQOO 16 将行业首发 Q-flex 操控技术，在滑动时延、触控采样率、陀螺仪采样率等方面均有提升。 据悉，iQOO 16 将深入游戏引擎打通 VIP 专属通道，针对高频操作场景进行专项优化…
3
号称“行业最高”，曝一加 16 手机多指采样率 520Hz、陀螺仪采样率首次达到 800Hz
IT之家 9 月 14 日消息，据微博博主 @数码闲聊站 今日透露，加加 16（一加 16）手机的操控“又是往前卷了一大步”： 落地新一代自研灵犀触控芯片，多指采样率做到了 520Hz，行业最高 首发新一代高精度大师陀螺仪， 采样率首次干到了 800Hz ，行业最高 * 2 博主表示，这种规格比 P…
4
我国首台：大视场亚秒级采样 1 米级巡天望远镜“天语”首次成像，完整摄下逾百万颗恒星
IT之家 9 月 5 日消息，据《科技日报》报道，9 月 5 日（今天）上午，天语望远镜首光仪式在青海冷湖天文观测研究基地举行。此前，坐落于该基地的天语望远镜成功获取首光图像，一次曝光即完整摄下 天鹅座北美洲星云、鹈鹕星云及周围银河中逾百万颗恒星 ，标志着该望远镜建成并将进入科学观测阶段。 IT之家…
5
Knowing When to Stop: Bayesian Optimal Stopping for LLM Evaluations
大模型评估常浪费采样算力，这项研究用贝叶斯最优停止动态决定何时收手，兼顾精度与成本，是做评测必读的思路。
arXiv:2608.14425v1 Announce Type: new Abstract: LLM evaluations often use fixed sampling budgets, testing every item the same number of times even aft…
6
On the Effect of Sampling Diversity in Scaling LLM Inference
采样多样性如何左右大模型推理扩展的缩放规律？这项研究给出了新视角。
arXiv:2502.11027v5 Announce Type: replace Abstract: Large language model (LLM) scaling inference is key to unlocking greater performance, and leveragi…
7
LLM-Derived Priors for Thompson Sampling in Cold-Start Comment Recommendation
用大模型先验给冷启动推荐装上导航，汤普森采样不再盲猜，评论推荐效率起飞。
arXiv:2608.03382v1 Announce Type: cross Abstract: Multi-armed bandit algorithms, especially Thompson sampling, are widely used in online recommendatio…
8
The Illusion of Stochasticity in LLMs
大模型做智能体时竟无法可靠随机采样？这份研究揭示隐性缺陷，值得每个Agent开发者警惕。
arXiv:2604.06543v2 Announce Type: replace-cross Abstract: In this work, we demonstrate that reliable stochastic sampling is a fundamental yet unfulfil…
9
What sampling temperature does in LLM RL training
解析采样温度在LLM强化学习训练中的实际作用，控制实验揭示关键机制。
Article URL: https://medium.com/@shizhanszh34/what-sampling-temperature-actually-does-in-llm-rl-training-a-controlled-study-of-temperature-a1f710e32a3…
10
汇顶全新柔性 OLED 触控芯片 GT9926 下半年大规模商用，支持 4000Hz 瞬时采样率
汇顶发布柔性OLED触控芯片GT9926，支持4000Hz瞬时采样率，下半年旗舰手机大规模商用
IT之家 7 月 20 日消息，汇顶科技今日宣布，全新柔性 OLED 触控芯片 GT9926 将于下半年发布的旗舰和游戏手机上大规模商用。 据官方介绍，全新芯片架构带来行业优异的 480Hz 多指采样率和 4000Hz 瞬时采样率 ，点击响应更快，开镜射击更跟手；全新的硬件降噪，让信噪比较上代提升了…
11
Show HN: Oodle.ai – $10 per million agent traces
专为LLM代理打造的列式存储可观测性方案，无采样全量追踪，每百万仅需10美元，有效应对非确定性调试痛点。
Hi HN, we're Kiran and Vijay! Over the past two years, we have built a columnar storage engine for observability: logs, metrics, and traces. Today, it…
12
Depth-Entropy Guided Sampling for Training-Free LLM Reasoning
无需训练的LLM推理新方法：深度熵引导采样，提升推理效率与质量。
arXiv:2607.09693v1 Announce Type: cross Abstract: Reinforcement learning (RL) has become the dominant paradigm for improving the reasoning capabilitie…
13
Tensor Train Diffusion: Leveraging Low-Rank Structures for High-Dimensional Score-Based Sampling
用张量列低秩分解攻克扩散模型高维采样难题，带来理论与计算效率双重飞跃。
arXiv:2607.06841v1 Announce Type: cross Abstract: Diffusion models offer a powerful framework for sampling from complex probability densities by learn…
14
D2PO: Optimizing Diffusion Samplers via Dynamic Preference
用动态偏好调优扩散采样器，D2PO在ECCV 2026提出创新方案，让AI绘画更高效精准。
arXiv:2607.06609v1 Announce Type: cross Abstract: We propose D2PO (Dynamic Direct Preference Optimization), a principled framework for optimizing diff…
15
Single-Rollout Asynchronous Optimization for Agentic Reinforcement Learning
单次采样异步优化刷新智能体强化学习效率，直击多步决策计算瓶颈，值得算法研究者细读。
arXiv:2607.07508v1 Announce Type: cross Abstract: Reinforcement learning (RL) is becoming increasingly important for post-training large language mode…
16
Tensor-Train Joint Modeling for Few-Step Discrete Diffusion
用张量训练联合建模突破离散扩散采样瓶颈，少步生成也能保持高质量，值得关注的前沿技术。
arXiv:2607.03788v1 Announce Type: new Abstract: Discrete diffusion promises orders-of-magnitude faster generation than autoregressive (AR) models for …
17
Silicon Sampling via Cross-Survey Transfer
大模型模拟受访者出新招，跨调查迁移能否突破个体预测瓶颈？评估缺陷被直指，值得一读。
arXiv:2607.03091v1 Announce Type: new Abstract: Silicon sampling-using large language models (LLMs) to simulate human survey respondents-has emerged a…
18
MACS: Measurement-Aware Consistency Sampling for Inverse Problems
反问题求解新方案，测量感知一致性采样，兼顾速度与精度，值得关注。
arXiv:2510.02208v3 Announce Type: replace-cross Abstract: Diffusion models have emerged as powerful generative priors for solving inverse imaging prob…
19
Accelerating Discrete Diffusion Models with Parallel-In-Time Sampling
离散扩散模型采样提速新突破，并行时间采样让生成效率大幅跃升。
arXiv:2607.00773v1 Announce Type: new Abstract: Discrete diffusion models are widely used for learning and generating discrete distributions. As the g…
20
Real-Time Hard Negative Sampling via LLM-based Clustering for Large-Scale Two-Tower Retrieval
用LLM聚类实现实时难负样本采样，直击双塔检索训练瓶颈，规模越大收益越明显。
arXiv:2607.00448v1 Announce Type: cross Abstract: The two-tower model has been widely used for large-scale recommendation systems, particularly in the…