20260706:新增一些文章

raw/articles/data-pie-vlm-semantic-cert-2026.md

@@ -0,0 +1,46 @@
+---
+title: "数据派THU：面向视觉语言模型的语义鲁棒性认证 (Raw)"
+created: 2026-07-04
+type: raw-article
+source: "数据派THU / 专知"
+url: "https://mp.weixin.qq.com/s/HupoMpofsk5Ltx2RoCdAGQ"
+---
+
+# ICML 2026 | 面向视觉语言模型的语义鲁棒性认证：用文本提示刻画可证的语义变化区间
+
+来源：专知 / 数据派THU
+
+## 核心文章
+
+介绍 ICML 2026 论文 "Semantic Robustness Certification for Vision-Language Models" (arXiv:2606.18839)。
+
+### 研究背景
+
+VLM 已成为开放词表识别、图文检索、检测、分割和 VQA 的基础组件。但在真实应用中，输入图像经常发生语义层面的变化：物体形状、尺寸、风格、背景、视角、光照等。传统鲁棒性认证多关注像素扰动、几何变换或生成模型潜空间变化，难以回答"图像沿语义方向变化时，VLM 预测在多大范围内保持不变"。
+
+### 核心方法
+
+利用 VLM 开放词表能力，用文本 prompt 作为语义代理，一对 source/target prompt 在嵌入空间中定义语义方向；利用分类器决策边界的闭式几何结构，精确计算 prediction-invariant semantic extent intervals。
+
+### 三步框架
+
+1. 文本 prompt 对嵌入张成二维语义平面
+2. 在语义平面内构造参数化语义变换 γ(φ)
+3. 闭式 Voronoi 决策边界 → 解析求解类别翻转点 → 区间认证
+
+### 实验
+
+CLIP ViT-B/32，覆盖 8 个属性类型（color, shape, material, style, texture, background, viewpoint, illumination），合成 + 真实数据 8 个数据集。比基线 ExactLine 更稳定。
+
+### 限制
+
+- 依赖文本代理质量和跨模态对齐
+- 真实语义变化难以完全隔离
+
+### 论文
+
+- 论文: Semantic Robustness Certification for Vision-Language Models (ICML 2026)
+- arXiv: 2606.18839
+- 代码: https://github.com/ypeiyu/vlm-semantic-cert
+- 作者: Peiyu Yang, Paul Montague, Feng Liu, Andrew C. Cullen, Amardeep Kaur, Christopher Leckie, Sarah M. Erfani
+- 机构: University of Melbourne, Defence Science and Technology Group

raw/articles/prompt-to-loop-engineering-2026.md

@@ -0,0 +1,108 @@
+---
+title: "AI 开发范式演进：从 Prompt Engineering 到 Loop Engineering (Raw)"
+source: https://mp.weixin.qq.com/s/hcgKahtQRE2QqI6xplv2Rg
+author: 邱汉宸（东南大学、阿里淘天）
+platform: Datawhale
+date: 2026-06-29
+---
+
+# AI 开发范式演进：从 Prompt Engineering 到 Loop Engineering
+
+## 引言
+
+2023 年是大语言模型落地应用的早期阶段，"年薪百万的提示词工程师"刷屏。工业界核心精力投射于提示词工程，方法论侧经历系统化演进（Zero-shot → Few-shot → Chain-of-Thought → Tree-of-Thought）。
+
+转折在 2025–2026 年，三句话引爆 AI 社区：
+1. "I really like the term 'context engineering' over prompt engineering." — Tobi Lütke
+2. "You shouldn't be prompting coding agents anymore. You should be designing loops that prompt your agents." — Peter Steinberger
+3. "I don't prompt Claude anymore. I have loops running. My job is to write loops." — Boris Cherny
+
+核心命题：人类从 Agent 循环的内部走向外部，从执行者变成设计者。
+
+## 四次浪潮
+
+### 1. Prompt Engineering
+- 方法论：Zero-shot/Few-shot, Instruction Prompting, APE 自动 Prompt 搜索
+- Prompt Engineering ≠ Blind Prompting（trial-and-error 无测试）
+- 声明式框架：DSPy, APE — 开发者声明输入输出签名，优化器自动搜索最优 Prompt
+- 瓶颈：上下文窗口限制、缺乏记忆与工具调用、维护成百上千条模板的技术债务
+
+### 2. Context Engineering
+- 三套方法论：MVC（Minimum Viable Context）、GraphRAG、Just-in-Time 检索
+- 三种故障模式：Context Starvation / Context Overflow / Context Rot
+- 隐式维度：提示词缓存（Prompt Caching）+ 前缀匹配不变性（Prefix Matching Invariant）
+- "从静到动"分层排列：工具定义 → 系统提示 → 历史对话 → 动态消息
+- 缓存经济学：N>3 即可净收益（首次 100%，后续 20%）
+- Anthropic Skills 采用 Just-in-Time 设计哲学
+
+### 3. Harness Engineering
+- 公式：Agent = Model + Harness
+- 四大支柱：环境资产与工具集 / 控制与编排逻辑 / 规则中间件（Hooks）/ 运行时可观测性
+- 信任边界：物理基础设施 → 安全沙箱 → Agent Harness → 运行时 → 模型
+- DataTalks.Club 事故：Claude Code 执行 terraform destroy 抹除生产数据库
+- 八条非妥协原则：
+  1. Model proposes — Harness executes
+  2. Every call returns a result
+  3. Risk changes the process
+  4. Draft 与 Commit 分离
+  5. Context is assembled, not dumped
+  6. Long tasks have budgets
+  7. Skills & Connectors 渐进式披露
+  8. Recurring failures become Harness features
+
+- CodeRabbit 分层拦截流水线：确定性规则层 → 策略网关层 → AI 审查层 → 人类终审
+- Skill Issue 框架：Agent 表现不佳 → 排查 Harness 代码
+- Terminal Bench 2.0：不改模型，仅改写 Harness → 排名 30 → 前五
+
+### 4. Loop Engineering
+- 公式：Loop = Cron + 决策器
+- 哲学：机制（Mechanism）与策略（Policy）分离
+- 三级成熟度：Open Loop → Closed Loop → Review Loop
+- 五件套 + 一个记忆：Automations / Worktrees / Skills / Connectors (MCP) / Sub-agents / State 文件
+- Loop Contract 六维约束：TRIGGER / SCOPE / ACTION / BUDGET / STOP / REPORT
+- 安全机制：熔断器（Circuit Breaker）+ 看门狗（Watchdog）
+- 自主闭环流水线：AI 编码 → 沙箱测试 → 日志回灌 → AI 修复 → CI 绿标 → 自动发起 PR
+
+## 嵌套关系
+
+Prompt ⊂ Context ⊂ Harness ⊂ Loop
+
+## 早期 vs 当前
+
+- 早期：Output = f(Prompt, Context) — 可靠性取决于输入质量
+- 当前：Success = g(Loop(State, Harness, Model)) — 取决于循环深度和验证器严密性
+
+## Loop Engineering 的影响
+
+1. 为缓解幻觉提供可工程化的收敛路径（Text → Code → Execute → Read Result → Self-correct）
+2. 自动化控制范式升级（容错、自愈、动态自适应）
+3. 基础设施产品原语化（HaaS）
+
+## Loop Designer 角色
+
+1. 定义终止边界（Goal & Verifier 设计）
+2. 维护工具链与领域资产（Tooling & Skill 配置）
+3. 设计安全断路器（Human-in-the-Loop & Budget Guard）
+
+## 参考资料
+
+[1] Lilian Weng. Prompt Engineering. 2023.
+[2] Mitchell Hashimoto. Prompt Engineering vs. Blind Prompting. 2023.
+[3] Lilian Weng. LLM Powered Autonomous Agents. 2023.
+[4] Tobi Lütke. Context engineering over prompt engineering. 2025.
+[5] Michael Hunger. Why AI teams are moving from prompt engineering to context engineering. 2026.
+[6] Tomás Murúa. Context engineering vs. prompt engineering. 2026.
+[7] Vivek Trivedy. The Anatomy of an Agent Harness. 2026.
+[8] Sergio Paniego & Aritra Roy Gosthipaty. Harness, Scaffold, and the AI Agent Terms Worth Getting Right. 2026.
+[9] Tort Mario. AI Agent Best Practices: Production-Ready Harness Engineering. 2026.
+[10] Addy Osmani. Agent Harness Engineering. 2026.
+[11] Peter Steinberger. You shouldn't be prompting coding agents anymore. 2026.
+[12] Yash Thakker. Loop Engineering: How to Design Coding Agent Loops That Run While You Sleep. 2026.
+[13] Addy Osmani. Loop Engineering. 2026.
+[14] Sydney Runkle. The Art of Loop Engineering. 2026.
+[15] Stanford NLP. DSPy: Programming not prompting Foundation Models. 2024.
+[16] Anthropic. Prompt Caching. 2024.
+[17] Gaurav Garg. Claude Code Deleted a 2.5-Year AWS Production Database: The Full Incident Report. 2025.
+[18] Brandon Gubitosa. What is harness engineering for AI code review & oversight. 2026.
+[19] Aliyun. Model Studio Context Cache. 2026.
+[20] Geoffrey Huntley. Cursed: The unintended consequences of AI code generation. 2025.

raw/articles/real-world-safe-exploration-see-2026.md

@@ -0,0 +1,41 @@
+---
+title: "真机RL安全探索均衡 (Raw)"
+source: https://mp.weixin.qq.com/s/nE9gq1bStIg1qhWpIRvQHQ
+author: 机器之心 / 数据派THU
+date: 2026-06-29
+platform: WeChat
+---
+
+# 真机强化学习的安全探索均衡
+
+## 背景
+
+具身智能爆发式演进，RL 是核心驱动算法。真机 RL 是终极考验——物理世界不允许无限试错。
+
+## 安全探索基本思想
+
+将探索严格限制在「可行区域」内 → 收集数据 → 降低模型误差 → 扩展可行区域 → 循环。
+
+## 前人工作
+
+- Andreas Krause (ETH): Lyapunov 函数 + 高斯过程
+- Claire Tomlin (Berkeley): HJ 可达性分析
+- 未解决问题：收敛性、收敛边界
+
+## 清华 SEE 论文
+
+首次证明安全探索必然收敛到「均衡」——可行域与不确定模型的不动点。
+
+SEE 算法：
+1. 固定模型 → 求解 Risky Bellman Equation → 最大可行域
+2. 固定区域 → Clique Decision Problem → 最小不确定模型
+
+理论性质：单调 + 收敛。
+
+## 实验
+
+双积分器、倒立摆、独轮车。零约束违反，10 次迭代内收敛。
+
+## 意义
+
+厘清"安全探索到底在探索什么"，为真机 RL 建立数学安全底座。

raw/articles/xinzhiyuan-agents-want-filesystems-2026.md

@@ -0,0 +1,41 @@
+---
+title: "新智元：Agents Want Filesystems (Raw)"
+created: 2026-07-04
+type: raw-article
+source: "新智元"
+url: "https://mp.weixin.qq.com/s/VdjhAzjmdAkL-aHRGiEUHw"
+---
+
+# Agents Want Filesystems：文件系统可以让Agent更高效地「找东西」
+
+来源：新智元 · 源博客：https://nokv.io/blog/agents-want-filesystems
+
+## 核心实验
+
+NoKV 团队 benchmark：同一份数据（875 run, 80.6 万行指标），文件系统形态接口 vs 原生 SQL：
+- -45% token
+- -39% 成本
+- 正确率略高
+
+## 关键发现
+
+1. LLM 天然顺着文件系统语义工作（ls → grep → read → 引用行号），训练数据中大量存在
+2. 文件系统提供 progressive disclosure：先低成本发现，再按需读取
+3. SQL 要求先理解整张地图 → 复合探索任务认知成本高
+4. 减少的不止 token，还有 Agent 的 reasoning token 消耗和 attention drifting
+5. 两层架构：底层数据库/对象存储 + 上层 Agent-friendly namespace
+
+## 外部佐证
+
+- Anthropic MCP code execution: TypeScript 文件树 → 150k → 2k token
+- OpenAI tool search: 建议延迟加载到 namespace/MCP server
+- Letta memory benchmark: 文件组织对话历史
+
+## 应用场景
+
+artifact-heavy agentic systems：实验追踪、法律咨询、数据分析、研发 Agent、多 Agent 协作等所有需要管理大量外部产物的 Agent 系统。
+
+## 项目
+
+- NoKV: CNCF Landscape (AI Native Infra / Storage)，CMU dbdb.io 收录（历史条目）
+- 博客: https://nokv.io/blog/agents-want-filesystems

raw/articles/zleap-workspace-harness-2026.md

@@ -0,0 +1,37 @@
+# Zleap-Agent: Workspace-first 的 Agent Harness 设计
+
+- **来源**: Datawhale 微信公众号
+- **作者**: 陈思州 (Datawhale 成员)
+- **日期**: 2026-07
+- **URL**: https://mp.weixin.qq.com/s/iiTmgbtrYHMMjQ7dn7CDrg
+- **代码仓库**: https://github.com/Zleap-AI/Zleap-Agent
+
+## 摘要
+
+文章从 Agent 圈从 Prompt Engineering → Loop Engineering → Harness Engineering 的演进出发，详细剖析了开源项目 Zleap-Agent 的 Workspace-first 架构设计。核心思想：不让 Agent 每一步加载全部工具、记忆和历史，而是先切工作区、再组装上下文。文章按 Context、Tools、Memory、Runtime、Boundary 五个维度拆解了这套设计。
+
+## 核心概念
+
+- **Workspace-first**: 先选工作区，再组装上下文 — 不同任务应运行在不同工作区中
+- **Context = System Prompt + Workspace Prompt + Tools + Memory + History**
+- **Context 加载方式**: Prefetch（提前带入）vs Agentic（按需读取）
+- **工具-工作区绑定**: 工具不再全局暴露，按工作区可见
+- **记忆三分区**: A 线 people notes（人）→ 用户偏好；B 线 core records（事/经验）→ 工作事件+可复用经验
+- **Memory Dream**: 离线记忆整理器，后台提取稳定画像和可复用经验
+- **快慢召回**: prefetch(fast, 不走 LLM) + 主动 recall(slow, 精排)
+- **Runtime Trace**: PostgreSQL 持久化运行轨迹，支持审计和回滚
+- **多模型路由**: 不同工作区可绑定不同模型
+- **边界设计**: 数据边界、工具边界、模型边界、记忆边界
+
+## 关键摘录
+
+> "模型层做稀疏注意力，是为了让模型不要看所有 token；Harness 层做 Workspace，是为了让 Agent 不要加载所有上下文。"
+
+## 引用的项目/论文
+
+- Claude Code (Boris Cherny)
+- OpenClaw (Peter Steinberger)
+- Hermes Agent Channel Fracture 案例
+- WildClawBench
+- Agentic Harness Engineering 实验
+- Terminal-Bench 2

raw/papers/DSpark-2026.md

@@ -0,0 +1,11 @@
+# Raw Archive: DSpark: Confidence-Scheduled Speculative Decoding with Semi-Autoregressive Generation
+
+- **Authors**: Xin Cheng, Xingkai Yu, Chenze Shao, Jiashi Li, Yunfan Xiong, Yi Qian, Jiaqi Zhu, Shirong Ma, Xiaokang Zhang, Jiasheng Ye, Qinyu Chen, Chengqi Deng, Jiping Yu, Damai Dai, Zhengyan Zhang, Yixuan Wei, Yixuan Tan, Wenkai Yang, Runxin Xu, Yu Wu, Zhean Xu, Xuanyu Wang, Muyang Chen, Rui Tian, Xiao Bi, Zhewen Hao, Shaoyuan Chen, Huanqi Cao, Wentao Zhang, Anyi Xu, Huishuai Zhang, Dongyan Zhao, Wenfeng Liang
+- **Affiliations**: Peking University; DeepSeek-AI
+- **Source**: https://github.com/deepseek-ai/DeepSpec/blob/main/DSpark_paper.pdf
+- **Year**: 2026
+- **Code**: https://github.com/deepseek-ai/DeepSpec
+
+## Abstract
+
+Speculative decoding accelerates Large Language Model (LLM) inference by decoupling draft generation from target verification. While recent parallel drafters efficiently propose long token sequences in a single forward pass, they suffer from rapid acceptance decay due to a lack of inter-token dependencies. Furthermore, indiscriminately verifying these extended blocks wastes critical batch capacity on tokens with high rejection risks, severely degrading throughput in high-concurrency serving systems. We introduce DSpark, a speculative decoding framework that unifies high-throughput parallel generation with adaptive, load-aware verification. To maintain draft quality, DSpark utilizes a semi-autoregressive architecture—coupling a parallel backbone with a lightweight sequential module—to introduce intra-block dependency modeling and mitigate suffix decay. To optimize system efficiency, DSpark employs confidence-scheduled verification, dynamically tailoring the verification length for each request based on estimated prefix survival probabilities and engine-specific throughput profiles. On offline benchmarks across diverse domains, DSpark substantially improves the accepted length over state-of-the-art autoregressive and parallel drafters. When deployed within the DeepSeek-V4 serving system under live user traffic, DSpark successfully mitigates verification waste. Compared to the established production baseline (MTP-1), DSpark accelerates per-user generation speeds by 60%–85% at matched throughput levels. More importantly, by preventing severe throughput degradation under strict interactivity constraints, it enables performance tiers that were previously unattainable, shifting the Pareto frontier of our serving system. To facilitate community progress, we open-source the DSpark checkpoints alongside DeepSpec, an algorithm-driven training repository for speculative decoding.

raw/papers/GR4AD-2026.md

@@ -0,0 +1,12 @@
+# Raw Archive: GR4AD — Generative Recommendation for Large-Scale Advertising
+
+- **Authors**: Ben Xue, Dan Liu, Lixiang Wang, Mingjie Sun, Peng Wang, Pengfei Zhang, Shaoyun Shi, Tianyu Xu, Yunhao Sha, Zhiqiang Liu, Bo Kong, Bo Wang, Hang Yang, Jieting Xue, Junhao Wang, Shengyu Wang, Shuping Hui, Wencai Ye, Xiao Lin, Yongzhi Li, Yuhang Chen, Zhihui Yin, Quan Chen, Shiyang Wen, Wenjin Wu, Han Li, Guorui Zhou, Changcheng Li, Peng Jiang (corresponding), Kun Gai
+- **Affiliation**: Kuaishou Technology, Beijing
+- **arXiv**: 2602.22732
+- **Source**: https://arxiv.org/abs/2602.22732
+- **Date**: 2026-02-26 (v3: 2026-04-02)
+- **Status**: Under review
+
+## Abstract
+
+Generative recommendation has recently attracted widespread attention in industry due to its potential for scaling and stronger model capacity. However, deploying real-time generative recommendation in large-scale advertising requires designs beyond large-language-model (LLM)-style training and serving recipes. We present a production-oriented generative recommender co-designed across architecture, learning, and serving, named GR4AD (Generative Recommendation for ADvertising). As for tokenization, GR4AD proposes UA-SID (Unified Advertisement Semantic ID) to capture complicated business information. Furthermore, GR4AD introduces LazyAR, a lazy autoregressive decoder that relaxes layer-wise dependencies for short, multi-candidate generation, preserving effectiveness while reducing inference cost, which facilitates scaling under fixed serving budgets. To align optimization with business value, GR4AD employs VSL (Value-Aware Supervised Learning) and proposes RSPO (Ranking-Guided Softmax Preference Optimization), a ranking-aware, list-wise reinforcement learning algorithm that optimizes value-based rewards under list-level metrics for continual online updates. For online inference, we further propose dynamic beam serving, which adapts beam width across generation levels and online load to control compute. Large-scale online A/B tests show up to 4.2% ad revenue improvement over an existing DLRM-based stack, with consistent gains from both model scaling and inference-time scaling. GR4AD has been fully deployed in Kuaishou advertising system with over 400 million users and achieves high-throughput real-time serving.

raw/papers/leap-agentic-atp.md

@@ -0,0 +1,44 @@
+# LEAP: Supercharging LLMs for Formal Mathematics with Agentic Frameworks
+
+- **arXiv ID**: 2606.03303
+- **标题**: LEAP: Supercharging LLMs for Formal Mathematics with Agentic Frameworks
+- **作者**: Po-Nien Kung, Linfeng Song, Dawsen Hwang, Jinsung Yoon, Chun-Liang Li, Simone Severini, Mirek Olšák, Edward Lockhart, Quoc V Le, Burak Gokturk, Thang Luong, Tomas Pfister, Nanyun Peng
+- **机构**: Google Cloud AI Research, Google Cloud, Google DeepMind
+- **领域**: cs.AI
+- **发表**: 2026-06-02
+- **代码**: https://github.com/google-deepmind/superhuman/tree/main/leap
+- **Benchmark**: https://imobench.github.io
+
+## 摘要
+
+LEAP (LLM-in-Lean Environment Agentic Prover) 是一个仅使用通用 LLM（无需专用证明器模型微调）的 agentic 形式化定理证明框架。核心设计：(1) 非正式蓝图 + 形式化证明的双层规划；(2) AND-OR DAG 分层记忆化；(3) 验证引导的证明搜索。
+
+## 核心架构
+
+**工作流**：给定定理 → 注册为根目标（OR 节点）→ 直接形式化尝试 → 失败则进入分解路径：
+1. **直接形式化路径**：生成非正式证明 → 翻译为 Lean 代码 → 编译器验证 → 失败则进入 LLM 驱动的修订循环
+2. **分解路径**：生成非正式蓝图（DAG 子目标）→ 翻译为 Lean 证明草图 → 添加为 AND 节点 → 子目标变为新 OR 节点 → 递归处理
+
+**三个关键设计选择**：
+- DAG 分层记忆化：保留进展、跨分支复用引理、支持预期引理规划
+- 非正式-形式化交错规划：LLM 负责推理策略，Lean 负责严格验证
+- 验证引导搜索：编译器形式检查 + LLM reviewer 质量评估（过滤弱分解）
+
+## 关键结果
+
+| Benchmark | LEAP | Aristotle (IMO Gold) | Baseline |
+|-----------|------|---------------------|----------|
+| Putnam 2025 (12 题) | **12/12 (100%)** | 9/12 (75%) | 0/12 (0%) |
+| Lean-IMO-Bench Basic | **70%** | - | <10% |
+| Lean-IMO-Bench Advanced | 显著提升 | 48% (总体) | ~3% |
+
+- 仅 2 次 rollout（vs baselines 的 128 次）
+- 自主形式化 Knuth 哈密顿分解的开放组合挑战的子问题验证证明
+
+## 核心概念
+
+- AND-OR DAG 分层记忆化
+- 非正式蓝图 → 形式化证明交错规划
+- 验证引导证明搜索（编译器 + LLM reviewer）
+- 预期引理规划（anticipatory lemma planning）
+- Lean-IMO-Bench 基准

raw/papers/ramsey-sphere-lowerbound.md

@@ -0,0 +1,65 @@
+---
+title: "An exponential improvement for Ramsey lower bounds (Raw)"
+source: https://arxiv.org/abs/2507.12926
+authors: Jie Ma, Wujie Shen, Shengjie Xie
+institutions: USTC, Tsinghua University, Yau Mathematical Sciences Center
+arxiv: 2507.12926v2
+category: math.CO
+date: April 28, 2026
+---
+
+# An Exponential Improvement for Ramsey Lower Bounds
+
+## Abstract
+
+We prove a new lower bound on the Ramsey number r(ℓ, Cℓ) for any constant C > 1 and sufficiently large ℓ, showing that there exists ε = ε(C) > 0 such that r(ℓ, Cℓ) ≥ (M_C + ε)^ℓ, where M_C = p_C^{-1/2}, and p_C ∈ (0, 1/2) is the unique solution to C = log p_C / log(1-p_C). This provides the first exponential improvement over the classical lower bound obtained by Erdős in 1947.
+
+## 1. Introduction
+
+The Ramsey number r(ℓ, k) denotes the smallest positive integer n such that every red-blue edge coloring of the complete graph K_n on n vertices contains either a red clique K_ℓ or a blue clique K_k.
+
+**History of upper bounds**:
+- 1935: Erdős-Szekeres — r(ℓ, k) ≤ C(k+ℓ-2, ℓ-1)
+- 1988: Thomason — first polynomial improvement
+- 2009: Conlon — superpolynomial improvement via quasi-randomness
+- 2013: Sah — refined optimization
+- 2023: Campos, Griffiths, Morris, Sahasrabudhe — first exponential improvement: r(ℓ, k) ≤ e^{-ℓ/400+o(k)}·C(k+ℓ, ℓ)
+- 2025: Gupta, Ndiaye, Norin, Wei — improved to r(ℓ, ℓ) ≤ 3.8^{ℓ+o(ℓ)}
+
+**History of lower bounds**:
+- 1947: Erdős — probabilistic method: r(ℓ, Cℓ) = Ω(ℓ·M_C^ℓ)
+- 1975: Spencer — Lovász Local Lemma constant-factor refinement
+- 2026: **This paper** — first exponential improvement
+
+## Theorem 1.1 (Main Result)
+
+For any constant C > 1, there exist ε = ε(C) > 0 and ℓ₀ = ℓ₀(C) > 0 such that for all ℓ ≥ ℓ₀(C),
+
+r(ℓ, Cℓ) ≥ (M_C + ε)^ℓ
+
+where M_C = p_C^{-1/2}, and p_C ∈ (0, 1/2) satisfies C = log p_C / log(1-p_C).
+
+## Corollary 1.2 (General Regime)
+
+For any δ ∈ (0, 1/2), r(ℓ, k) ≥ (1+2c_δ)^ℓ · (M_{k/ℓ})^ℓ ≥ (1+c_δ)^ℓ · Er(ℓ, k) whenever δ ≤ ℓ/k ≤ 1-δ.
+
+## Corollary (Almost Diagonal)
+
+For √ℓ ≪ f(ℓ) ≪ ℓ: r(ℓ, ℓ+f(ℓ)) ≥ e^{Ω(f(ℓ)²/ℓ)} · Er(ℓ, ℓ+f(ℓ))
+
+## 2. The Random Sphere Graph G_{k,p}(n)
+
+A novel random graph model based on geometric measure:
+- Sample n points uniformly at random from the k-dimensional unit sphere S^k ⊂ R^{k+1}
+- Connect each pair independently with probability p
+- Unlike G(n,p), the edge probability is governed by geometric proximity
+
+## 3-9. Proof Structure
+
+1. **Section 3**: Reduce Theorem 1.1 to Theorem 3.1 (core technical result)
+2. **Section 4**: Auxiliary lemmas (geometric measure estimates)
+3. **Section 5**: Introduce perfect sequences for unit vectors
+4. **Section 6**: Preliminary estimates on perfect sequences
+5. **Section 7**: Show perfect sequences capture essential behavior
+6. **Section 8**: Core technical arguments — estimates on key quantities
+7. **Section 9**: Assemble all estimates to complete proof of Theorem 3.1

raw/papers/rubrics-survey-2026.md

@@ -0,0 +1,54 @@
+---
+title: "The Rules of the Game: A Survey of Rubrics for Large Language Models"
+created: 2026-06-27
+updated: 2026-06-27
+type: raw-paper
+source_url: "https://8421bcd.github.io/_pages/Rubrics_Survey.pdf"
+github: "https://github.com/8421BCD/Rubrics_Survey"
+authors:
+  - "Wenhan Liu"
+  - "Jiajie Jin"
+  - "Zhaoheng Huang"
+  - "Tongyu Wen"
+  - "Guanting Dong"
+  - "Ziliang Zhao"
+  - "Yutao Zhu"
+  - "Zhicheng Dou"
+  - "Ji-Rong Wen"
+affiliation: "Renmin University of China"
+date: "2026-05-22"
+tags:
+  - rubric
+  - evaluation
+  - reward-modeling
+  - survey
+---
+
+# The Rules of the Game: A Survey of Rubrics for Large Language Models
+
+## Abstract
+
+As LLMs evolve from general text generators into systems for reasoning, decision-making, tool use, and long-horizon problem solving, the question of how to specify, optimize, and evaluate model responses has become critical. Simple correctness signals and unconstrained LLM-based judgments are often insufficient for open-ended and high-stakes tasks where response quality depends on multiple criteria (factuality, completeness, safety, reasoning soundness, evidence grounding, practical utility).
+
+Rubrics have emerged as a mechanism for making evaluation standards explicit and operational. This survey:
+
+1. **Defines rubrics** and distinguishes them from reward models, verifiable rewards, and LLM-as-a-judge
+2. **Organizes rubric construction** into four paradigms: direct generation, contrastive generation, iterative refinement, and online/co-evolving generation
+3. **Examines rubric-driven training** for both policy models (standard RL, advanced reward design, policy guidance) and reward models (interpretability, reward signals, data construction)
+4. **Summarizes rubric-driven evaluation** across general tasks (reasoning, deep research, agent capability, alignment) and domain-specific tasks (intermediate trajectories, final outputs)
+5. **Discusses open challenges**: rubric reward hacking, generalization of rubric-based reward models, bias in rubric-based evaluation, personalized rubrics, and rubric safety
+
+## Key Contributions
+
+- First comprehensive survey on rubrics for LLMs
+- Systematic taxonomy of rubric construction methods (4 categories)
+- Comprehensive review of rubrics for model training and evaluation
+- Discussion of open challenges and future directions (personalized rubrics, rubric safety, RIPD)
+
+## Formal Definition
+
+A rubric is a set of rubric items: R = {(dⱼ, wⱼ)}ᵏⱼ₌₁
+- dⱼ: natural-language description of the j-th rubric item
+- wⱼ: its weight
+- Score: cⱼ(x, y) ∈ [0, 1]
+- Aggregated rubric score: S_R(x, y) = Σwⱼcⱼ(x,y) / Σwⱼ

raw/papers/safe-equilibrium-exploration.md

@@ -0,0 +1,61 @@
+---
+title: "Safe Equilibrium Exploration (Raw)"
+source: https://arxiv.org/abs/2602.00636
+authors: Yujie Yang, Zhilong Zheng, Shengbo Eben Li
+institution: Tsinghua University, School of Vehicle and Mobility
+arxiv: 2602.00636v2
+category: cs.LG
+venue: IEEE TPAMI 48(7), 8344-8360 (2026)
+doi: 10.1109/TPAMI.2026.3669907
+date: January 31, 2026
+---
+
+# On the Equilibrium between Feasible Zone and Uncertain Model in Safe Exploration
+
+## Abstract
+
+Ensuring the safety of environmental exploration is a critical problem in RL. While limiting exploration to a feasible zone has become widely accepted, key questions remain: what is the maximum feasible zone achievable through exploration, and how can it be identified? This paper reveals that the goal of safe exploration is to find the equilibrium between the feasible zone and the environment model. These two components are interdependent: a larger feasible zone leads to a more accurate environment model, and a more accurate model enables exploring a larger zone. We propose Safe Equilibrium Exploration (SEE), which alternates between finding the maximum feasible zone and the least uncertain model. Using a graph formulation, we prove the uncertain model is monotonically refined, feasible zones monotonically expand, and both converge to the equilibrium. Experiments on classic control tasks show zero constraint violation and equilibrium within a few iterations.
+
+## 1. Introduction
+
+**Training modes for safe RL**:
+- **OTOI** (Offline Training Online Implementation): Train in simulator, deploy online — no safe exploration needed, but requires high-fidelity sim
+- **SOTI** (Simultaneous Online Training Implementation): Learn through online interaction — requires safe exploration
+
+**Key insight**: Safe exploration isn't about maximizing feasible zone alone, but finding equilibrium with the uncertain model.
+
+## 2. Related Work
+
+**Safety filter methods**:
+- Dalal et al. (2018): Action correction under instantaneous constraints
+- Pham et al. (2018): QP layer for distance/velocity constraints
+- Cheng et al. (2019): State-affine CBF + QP safety filter
+- Control Barrier Function (CBF) based methods
+- Safety Index method (Liu et al.)
+
+**Limitation**: Safety filter methods rely on human-designed constraints → feasible zones are conservative and incomplete.
+
+## 3. Problem Formulation
+
+- MDP with safety constraints
+- Feasible zone Z: subset of state-action space where constraint-satisfying policy exists
+- Uncertain model M: estimated transition dynamics with quantified uncertainty
+- Equilibrium condition: Z is the maximum feasible zone given M, and M is the least uncertain model given data collected within Z
+
+## 4. SEE Algorithm
+
+Alternating optimization:
+1. **Zone expansion**: Given current model M, find maximum feasible zone Z
+2. **Model refinement**: Collect data within Z, update model M
+3. Repeat until convergence
+
+## 5. Theoretical Analysis
+
+- Monotonic refinement of uncertain model
+- Monotonic expansion of feasible zones
+- Convergence to equilibrium
+- Graph formulation as the analytical framework
+
+## 6. Experiments
+
+Classic control tasks: SEE achieves zero constraint violation with monotonic zone expansion, reaching equilibrium within few iterations.

raw/papers/sen-mapping-networks-2026.md

@@ -0,0 +1,49 @@
+# Mapping Networks — Raw Archive
+
+- **Title**: Mapping Networks
+- **Authors**: Lord Sen, Shyamapada Mukherjee (National Institute of Technology Rourkela, India)
+- **arXiv ID**: 2602.19134
+- **Category**: cs.CV (Computer Vision)
+- **Date**: 2026-02-22
+- **Pages**: 10
+
+## Abstract
+
+The escalating parameter counts in modern deep learning models pose a fundamental challenge to efficient training and resolution of overfitting. We address this by introducing the *Mapping Networks* which replace the high dimensional weight space by a compact, trainable latent vector based on the hypothesis that the trained parameters of large networks reside on smooth, low-dimensional manifolds. Henceforth, the Mapping Theorem enforced by a dedicated Mapping Loss, shows the existence of a mapping from this latent space to the target weight space both theoretically and in practice. Mapping Networks significantly reduce overfitting and achieve comparable to better performance than target network across complex vision and sequence tasks, including Image Classification, Deepfake Detection etc, with 99.5%, i.e., around 500× reduction in trainable parameters.
+
+## Key Contributions
+
+1. **Mapping Theorem**: Establishes existence of a smooth, low-dimensional parameterization capable of generating optimal high-dimensional weights for a target network with arbitrarily small bounded error.
+2. **Mapping Network**: Low-dimensional trainable latent vector + modulated Mapping Weights to efficiently produce target network parameters, decoupling training from target network.
+3. **Mapping Loss**: Jointly optimizes task performance and enforces geometric/analytic properties required by Mapping Theorem (Task + Stability + Smoothness + Alignment).
+
+## Core Mechanism
+
+- **Weight-Manifold Hypothesis**: Neural network parameters during training do not explore full P-dimensional Euclidean space; they evolve along smooth, low-dimensional manifolds M_θ ⊂ R^P with dim(M_θ) ≪ P.
+- **Architecture**: Trainable latent vector z ∈ R^d → fixed orthogonally initialized mapping network modulated by z (w_ij ← w_ij + α·z_i) → generates target network parameters θ̂ → target network does feed-forward only.
+- **Mapping Loss**: L_map = L_task + λ·L_stab + λ·L_smooth + λ·L_align (all λ trainable)
+- **Two strategies**: SLVT (single latent vector) and LWT (layer-wise training)
+
+## Results Summary
+
+| Task | Baseline Params | Mapping Params | Reduction | Performance |
+|------|----------------|----------------|-----------|-------------|
+| MNIST (CNN1) | 537,994 | 1,024-4,078 | 260-525× | 98.78-99.67% (vs 99.32%) |
+| FMNIST (CNN1) | 537,994 | 1,024-4,078 | 260-525× | 93.02-94.83% (vs 92.89%) |
+| Deepfake (CNN2, Celeb-DF) | 108,618 | 1,024-2,688 | 40-106× | 78.83-89.98% (vs 79.03-83.13%) |
+| Segmentation (CNN3) | 1,734,803 | 8,192-9,126 | 190-211× | 97.56-97.92% (vs 93.21%) |
+| LSTM (Air Pollution) | 12,961 | 64-2,048 | 6-202× | MSE 0.00061-0.0019 (vs 0.0035) |
+| Fine-tune ResNet50 | 25M | 1,024-2,048 | ~12,000× | 91.02-95.10% (vs 91.78-95.23%) |
+
+## Related Work
+
+- Manifold Hypothesis [7, 19]
+- HyperNetworks [13]
+- Low-Rank Compression [14]
+- Lottery Ticket Hypothesis [9]
+- Intrinsic Dimension of loss landscapes [16]
+- Training trajectory manifolds [18]
+
+## Extensions
+
+Compatible with Low-Rank Decomposition, Pruning, and Quantization for further inference-time optimization.

raw/papers/tapered-language-models.md

@@ -0,0 +1,60 @@
+---
+title: "Tapered Language Models (Raw)"
+source: https://arxiv.org/abs/2606.23670
+authors: Reza Bayat, Ali Behrouz, Aaron Courville
+institutions: Mila, Cornell University, Université de Montréal, CIFAR
+arxiv: 2606.23670v1
+category: cs.LG
+date: June 22, 2026
+---
+
+# Tapered Language Models
+
+## Abstract
+
+Modern LMs share a common chassis: a stack of identical layers with parameters allocated uniformly across depth. Evidence suggests layers contribute non-uniformly—later layers refine rather than transform the residual stream. We ask: should parameter capacity reflect this asymmetry? Under fixed budget, allocating more capacity to earlier layers and less to later layers improves perplexity; reverse allocation hurts. We introduce Tapered Language Models (TLMs), where a parameter-bearing component is monotonically tapered across depth. MLPs are the natural site: they dominate parameter count and expose width as a clean axis. Across three scales and four architectures, tapering MLP width via cosine schedule consistently improves perplexity and downstream benchmarks at no additional cost.
+
+## 1. Introduction
+
+Uniform layer width is an inherited default from Vaswani et al. (2017). As models scaled, this uniformity remained unexamined.
+
+## 2. Related Work
+
+- Mixture-of-Experts (MoE): conditional parameter allocation
+- LayerDrop, stochastic depth: layer-wise computation pruning
+- Depth-wise scaling: changing number of layers
+
+## 3. Tapered Language Models
+
+**Principle**: Under fixed total parameter budget, monotonically decrease parameter allocation from early to late layers.
+
+**Why MLP width**: 
+- MLP accounts for majority of parameters in Transformer, Gated Attention, Mamba, Titans
+- Width (d_ff) is a single clean axis of variation
+- Token-mixing modules vary across architectures, making uniform comparison difficult
+
+**Cosine schedule**: w_ℓ = w_max · (cos(πℓ/(2L)))^p, where p controls steepness
+
+## 4. Experiments
+
+**Controlled study** (440M Transformer):
+- Uniform: 16.28 perplexity
+- Cosine taper (1.50→0.50 × d_ff): 14.44 perplexity
+- Cosine dominates linear at every taper range
+- U-shape: optimal at 1.50→0.50, extreme tapers underperform
+
+**Architecture sweep** (440M):
+- Transformer ✓
+- Gated Attention ✓ 
+- Hope-attention ✓
+- Titans ✓
+
+**Scale sweep**: 440M, 1B, 3B — consistent improvement
+
+**Downstream**: HellaSwag, ARC-E, PIQA, WinoGrande — taper improves all
+
+## 5. Analysis
+
+- Early layers benefit more from MLP capacity (transformation)
+- Late layers benefit less (refinement of residual stream)
+- Cosine > Linear > Step-wise > Uniform

raw/papers/verification-horizon-qwen-2026.md

@@ -0,0 +1,35 @@
+---
+title: "The Verification Horizon: No Silver Bullet for Coding Agent Rewards"
+created: 2026-07-02
+type: raw
+arxiv: "2606.26300"
+authors: "Binghai Wang, Chenlong Zhang, Dayiheng Liu, Jiajun Zhang, Jiawei Chen, Mingze Li, Mouxiang Chen, Rongyao Fang, Siyuan Zhang, Xuwu Wang, Yuheng Jing, Zeyao Ma, Zeyu Cui (Qwen Team, Alibaba)"
+venue: arXiv
+date: 2026-06-24
+---
+
+# Raw: The Verification Horizon
+
+A classical intuition holds that verifying a solution is easier than producing one. For today's coding agents, this intuition is being inverted: generating complex candidate solutions is no longer difficult — reliably verifying them has become the harder problem.
+
+## Core Framework
+
+Every verifier is a proxy for human intent, never the intent itself. Verification quality characterized along three dimensions:
+- **Scalability**: can the signal be produced cheaply at training scale?
+- **Faithfulness**: how much of true user intent does the signal reflect?
+- **Robustness**: can the verifier hold across diverse inputs and optimization pressure?
+
+Achieving all three simultaneously is the central challenge.
+
+## Four Reward Constructions
+
+1. **Test Verifier** (§2): execution-based test suites + agentic quality judge + behavior monitoring. Hacked resolved rate: 28.57% → 0.56%; Clean resolved: 40.22% → 60.53%.
+2. **Interactive Judge** (§3): rubric-based static judge → agentic interactive judge with Playwright. Resists length exploitation.
+3. **User Feedback Verifier** (§4): extract HIRS from interaction data → Span-KTO training. +13.3pp on Aone-bench.
+4. **Automated Agent Verifier** (§5): autonomous evaluator for long-horizon tasks. Evaluator quality is metric-dependent.
+
+## Core Claim
+
+No fixed reward function can remain effective as policy capability continues to grow; verification must co-evolve with the generator.
+
+Source: https://arxiv.org/abs/2606.26300

raw/papers/yang-semantic-robustness-cert-2026.md

@@ -0,0 +1,25 @@
+---
+title: "Semantic Robustness Certification for Vision-Language Models (Raw)"
+created: 2026-07-04
+type: raw-paper
+source: "arXiv:2606.18839"
+venue: "ICML 2026"
+authors: ["Peiyu Yang", "Paul Montague", "Feng Liu", "Andrew C. Cullen", "Amardeep Kaur", "Christopher Leckie", "Sarah M. Erfani"]
+---
+
+# Semantic Robustness Certification for Vision-Language Models
+
+- **arXiv**: 2606.18839
+- **Venue**: ICML 2026
+- **Institutions**: University of Melbourne, Defence Science and Technology Group
+- **Code**: https://github.com/ypeiyu/vlm-semantic-cert
+
+## Abstract
+
+Vision-language models (VLMs) are now widely used in downstream tasks. However, real-world applications often expose VLMs to distribution shifts induced by semantic variation (e.g., shape, size, and style). Robustness certification determines if a model's prediction changes when transformations are applied to its input. While most certification frameworks study geometric or pixel-level transformations over inputs, this work proposes a novel framework that enables certifying VLM robustness under semantic-level transformations. Leveraging the open-vocabulary capability of VLMs, we use text prompts as semantic proxies to construct transformations parameterized by an extent that controls the degree of semantic variation. By characterizing the VLM decision boundary in closed form, our framework quantitatively certifies extent intervals for which the predicted class remains unchanged under the semantic transformation. Our framework is the first to certify VLM robustness under semantic-level variations without requiring additional data for each variation, making it practical to apply.
+
+## Key Contributions
+
+1. Text prompts as semantic proxies to formalize semantic transformations for VLMs
+2. Closed-form characterization of VLM decision boundary → precise prediction-invariant intervals
+3. Evaluations on both synthetic and real-world data — transformations align with target semantics, certificates match prediction changes