手写一个 LLM Harness 框架：用工程化手段把大模型幻觉踩在脚下
阅读文章 JUEJIN
掘金
工程实践
CHRONOLOGICAL LEDGER
按时间浏览
2026 年
八月
50 ENTRIESGENCO - A Unified Neural Solver Embedded in a Development Framework for Steady-State Grid Analysis
阅读文章 ARXIV
AI
cs.AI
The Water Footprint of AI
阅读文章 HACKER_NEWS
AI
Hacker News
From Values to Benchmarks: Evaluating Large Language Models for Governmental Use in Dutch
阅读文章 ARXIV
大语言模型
cs.CL
How Claude marks AI-generated content
阅读文章 HACKER_NEWS
大语言模型
Hacker News
Multimodal Model Diffing for Feature Discovery and Control
阅读文章 ARXIV
计算机视觉
cs.CV
As AI eats the web, the internet’s collective memory is disappearing
阅读文章 HACKER_NEWS
AI
Hacker News
让不同大模型共享一个 Agent：Pi 如何统一 Provider 与 Context Handoff
阅读文章 JUEJIN
AI Agent
掘金
DeepSeek: Reverse Engineering an AI Assistant by Interviewing Itself
阅读文章 HACKER_NEWS
AI
Hacker News
[AINews] Muse Glimmer and Spark: Open Weights return Personal Superintelligence promise
阅读文章 BLOGS_PODCASTS
大语言模型
AI Agent
光本位联合文心快码打造面向光电芯片研发的全栈AI Agent Lightmate
阅读文章 JUEJIN
AI Agent
掘金
Beyond Naturalness: Probing Automated Text-To-Speech Evaluators on Linguistically Grounded Dimensions
阅读文章 ARXIV
大语言模型
Prompt 工程
KGCaRe: Explainable Complex Conditional Question Answering using Automatic Knowledge Graph Construction and Context Retrieval with LLMs
阅读文章 ARXIV
大语言模型
RAG
What's the best programming language for coding agents?
阅读文章 HACKER_NEWS
AI Agent
Hacker News
A Picture is Worth a Thousand Tokens: How Vision Language Models Cut AI Energy Costs While Improving Accuracy
阅读文章 ARXIV
大语言模型
cs.AI
TEPA: Revoking Stale Memories for Conflict-Robust Language Agents
阅读文章 ARXIV
AI Agent
Prompt 工程
Humanising LLM Outputs Is Dumb
阅读文章 HACKER_NEWS
大语言模型
Hacker News
Show HN: Needle2: 14MB agentic LLM for phones, wearables, smart home and robots
阅读文章 HACKER_NEWS
大语言模型
Hacker News
Diffusion LLMs as Targets and Adversaries: Mechanistic Safety Exploits
阅读文章 ARXIV
大语言模型
生成式 AI
浅析大模型推理十二篇之KV Cache
阅读文章 JUEJIN
掘金
工程实践
SABRE: Scalable and Automated Benchmarking of VLMs under Stress
阅读文章 ARXIV
计算机视觉
cs.CV
Mark Zuckerberg attacks 'closed' AI rivals as Meta returns to open models
阅读文章 HACKER_NEWS
AI
Hacker News
Post-Grokking Collapse at the Representation-Readout Interface in Muon-Trained Transformers
阅读文章 ARXIV
AI
cs.AI
Launch HN: Stoa Markets (YC S26) – A Marketplace for GPUs and AI Servers
阅读文章 HACKER_NEWS
AI
Hacker News
毕业老学长给我留的最后一句话是：“AI 写的，别挂我名。” 我直接 神(Seed Evolving) 来！助我！
阅读文章 JUEJIN
掘金
工程实践
What building an AI-native finance function taught me
阅读文章 BLOGS_PODCASTS
AI
Company
Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows
阅读文章 HACKER_NEWS
AI Agent
Hacker News
Fisher-R1: Training LLM Agents for Reliable Hypothesis Testing
阅读文章 ARXIV
大语言模型
AI Agent
PsychoAgent: An Affect-Sensitive Cognitive Architecture for Conflict-Aware Memory in LLM Agents
阅读文章 ARXIV
大语言模型
RAG
OpenAI's New Device Will Be Hockey Puck-Sized and Cost over $300
阅读文章 HACKER_NEWS
AI
Hacker News
Evolve your marketing with new AI tools
阅读文章 BLOGS_PODCASTS
AI
Google Ads
Over 181,000 AI meeting recordings left wide open in note taking app
阅读文章 HACKER_NEWS
AI
Hacker News
How to build an AI agent: A simple guide for anyone
阅读文章 BLOGS_PODCASTS
AI Agent
AI
An Exploratory Evaluation of LLM-Assisted Rewriting of Moderate-Complexity Financial Sentences for DisCoCat-Based Sentiment Analysis
阅读文章 ARXIV
大语言模型
Prompt 工程
面试官问"这段逻辑为什么这样写"——我才发现，这半年我写的代码，我自己都解释不了
阅读文章 JUEJIN
掘金
工程实践
Model ML completes finance work more efficiently with GPT-5.6 Sol
阅读文章 BLOGS_PODCASTS
大语言模型
Startup
Mistral Patent for "Code implemented tool calls"
阅读文章 HACKER_NEWS
大语言模型
Hacker News
Blast Radius
阅读文章 ARXIV
Prompt 工程
cs.AI
Taxonomy-Driven Analysis of Open-Source AI Risk Mitigation Tools
阅读文章 ARXIV
大语言模型
生成式 AI
终端用户
阅读文章 JUEJIN
掘金
工程实践
SkillProx: Self-Evolving Agent Skills via Proximal Textual Gradient Descent
阅读文章 ARXIV
大语言模型
AI Agent
Show HN: Voice driven murder mystery, Interview AI suspects with your voice
阅读文章 HACKER_NEWS
AI
Hacker News
Interaction Creates Dynamical AI Behavior Absent in Isolation
阅读文章 ARXIV
AI Agent
cs.AI
开始 AI 编程前，到底需要准备什么？
阅读文章 JUEJIN
掘金
工程实践
Docker Sandboxes – Disposable, isolated sandboxes for AI agents
阅读文章 HACKER_NEWS
AI Agent
Hacker News
从 Prompt 到 Graph：AI 工程的进化史
阅读文章 JUEJIN
掘金
工程实践
CreativeInstruct: Scalably Teaching LLMs to Balance Quality, Creativity, and Diversity
阅读文章 ARXIV
大语言模型
cs.CL
Auto mode is now the default in Claude Code
阅读文章 HACKER_NEWS
大语言模型
Hacker News