跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+1352新增论文
+1083其中 AI 相关
+143已有论文修订
+94新增 AI 导读

arXiv 最近完成扫描:2026-11-19 06:59 · 全站最近入库:2026-11-19 15:04各信源进度 →

统计口径与数据新鲜度

统计 2026-11-18 15:50 — 2026-11-19 15:50(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

最近完成检查:2026-11-19 06:59最近发现新增版本:2026-11-19 15:04库内论文最新更新时间:2026-11-19 02:59

每分钟刷新已保存的数据。空检查不代表采集异常;新收录可能包含历史论文。AI 导读按新增版本计数。

27,409 已收录论文22,787 AI 相关8,342 有修订112 HF 社区发现

浏览论文库

时间范围跟随排序:版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常,查看状态 →

找到 2 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-19 06:57(北京时间)
版本更新 · v4arXiv:2505.02977

Tackling Parallelization Challenges of Randomized Preconditioners With Dependency Tracking

Tianyu Liang, Chao Chen, Yotam Yaniv, Hengrui Luo, David Tench, Xiaoye S. Li, Aydin Buluc, James Demmel 等 9 位作者

原文摘要

Incomplete Cholesky (IC) preconditioners are the standard black-box, memory-efficient first resort for large graph Laplacian systems, but classical IC can break down on irregular graphs. Approximate Cholesky (AC) addresses this with randomized dropping that is provably breakdown-free. Its stochastic fill-in pattern, however, resists static parallelization, creating a "setup bottleneck." We present ParAC, which parallelizes the sequential rchol algorithm on CPUs and GPUs using dynamic dependency tracking and architecture-aware memory management, without altering its numerics. Total time-to-solution speedups over deterministic IC reach $57.3\times$ on CPU and $52.2\times$ on GPU. Algebraic multigrid (AMG) remains preferable on structured PDE discretizations; on irregular graph Laplacians where AMG struggles, ParAC attains up to $7.2\times$ and $6.0\times$. On 200 adversarial SDD systems, ParAC converges on every instance, whereas the AMG solvers fail on 15 and 19, on CPU and GPU, respectively.

cs.DC · cs.DS · math.NA

版本更新 · v2arXiv:2609.38866

When Context Changes: Understanding Update Failures in LLMs

Junyu Guo, Yuchen Fang, Shangding Gu, Costas Spanos, James Demmel, Javad Lavaei

原文摘要

As preferences, goals, and facts change, LLM agents must use the current state while earlier versions remain in context. Yet they can answer with an old value of the same variable, a failure that we call stale binding. To study when models use outdated information and why, we introduce Controlled In-Context Memory (CICM), a benchmark for tracking and using updated information in conversations and agent logs. We observe that even frontier reasoning models can fail to recover the current state. We find that in open-source models probes can still recover the updated value when the model answers with an old one, pointing to a failure to select information that remains available. Component tests in Qwen and Pythia identify a mechanism for this selection failure: attention drift, where attention favors old values over the current one when producing an answer. We study a one-layer transformer to mathematically understand how this phenomenon happens: when attention scores are similar, several old values can together receive more attention than the current value. Guided by this explanation, we redirect attention toward the current value without further training. When the current value is requested directly, adjusting this intervention for each input corrects most old-value errors across various model families while preserving nearly all initially correct answers. Reliable context management therefore requires more than remembering updated information: models must use it to guide their answers.

cs.AI