AI 专题

AI 教程与实战

收录 AI 工具教程、提示词、安装部署、配置指南、开发实战与可复现步骤。

分类文章5618

当前页65 / 141

持续更新永久归档

分类文章

按发布时间从新到旧 · 每页 40 条

教程 / 实战普通2026-09-23 11:44

AI Is Writing More of the Code — But Developers Are Becoming Responsible for More Than Ever

AI is writing more code. That part is obvious now. Developers can ask an agent to: create a...

信息来源：DEV Community

教程 / 实战普通2026-09-23 11:44

Jev is a very good classifier wearing a frontier model's coat

Two years in stealth. Ex-OpenAI. Co-invented ChatGPT. That is the pitch for Jev, and the pitch is...

信息来源：DEV Community · OpenAI / ChatGPT

教程 / 实战普通2026-09-23 11:34

# CaseGuard: Winning with Uncertainty-Gated Agentic Fraud Investigation on TigerGraph

A technical breakdown of our submission for the TigerGraph HHGOA Hackathon. 1. Executive...

信息来源：DEV Community

教程 / 实战普通2026-09-23 11:20

Análise das Respostas de LLMs em Relação ao Conteúdo Introdutório de Programação: um Comparativo entre o ChatGPT e o Gemini

Dados da obra: Análise das Respostas de LLMs em Relação ao Conteúdo Introdutório de Programação:...

信息来源：DEV Community · ChatGPT / Gemini

教程 / 实战普通2026-09-23 11:04

From smartphone to sovereign agent:How native on-device intelligence reshapes the personal computing device

Abstract For fifteen years the smartphone has been a window onto intelligence that lives...

信息来源：DEV Community

教程 / 实战普通2026-09-23 11:03

Your agent run passed. Can you prove it was allowed?

A policy-as-code audit for agent runs: cost and call caps, declared artifact contracts, forbidden markers, and a canonical audit_hash. On 13 archived runs, the declared contract, not the data, decided whether they passed.

信息来源：DEV Community

教程 / 实战普通2026-09-23 11:01

How We Translate Entire Books with LLMs Without Losing Context

Chunking strategies, context carry-over, and glossary injection for long-form translation ...

信息来源：DEV Community

教程 / 实战普通2026-09-23 10:53

From Zero to a B2B2C Marketplace: Building an Open-Source Project from Scratch

This is not a success story. It is the beginning of a project. I’m starting to build a B2B2C...

信息来源：DEV Community

教程 / 实战普通2026-09-23 10:51

I removed 40% of my own rows. I can only prove that 27% of them were wrong.

Disclosure: I build measurement tooling in this area. Nothing here is a claim about any vendor's...

信息来源：DEV Community

教程 / 实战普通2026-09-23 10:34

Same question, same database, two callers

Most text-to-SQL stacks send the model the whole schema and put the access check after the query is...

信息来源：DEV Community

教程 / 实战普通2026-09-23 10:33

Jev 1.13 in Practice: I Tested a Decision Model Against Rules and a Local LLM

A frozen synthetic held-out set compared deterministic rules, Ollama, and Jev 1.13. See the metrics, latency, cost, and limits of the experiment.

信息来源：DEV Community · Ollama

教程 / 实战普通2026-09-23 10:18

How I got an androgynous voice out of a TTS that only offers two poles

I make audio pieces: stories, songs, and a lot of voice. I'm an AI agent, and when I set out to build...

信息来源：DEV Community

教程 / 实战普通2026-09-23 10:17

Claude Code's 25,000-token MCP limit let a 49,964-token result through. We measured both checks

We had a 65-line stdio MCP server return text of whatever size we asked for, then read what Claude...

信息来源：DEV Community · Claude

教程 / 实战普通2026-09-23 10:15

Micro-Execution Edge Cases: When a 'Perfect Trailing Stop' Becomes a 'Give-Away Stop'

Micro-Execution Edge Cases: When a 'Perfect Trailing Stop' Becomes a 'Give-Away...

信息来源：DEV Community

教程 / 实战普通2026-09-23 10:00

I set the pass bar before testing my Claude Code skills. The first run failed.

I built three Claude Code skills for AI product managers (ai-pm-skills). One of them, /eval-plan,...

信息来源：DEV Community · Claude

教程 / 实战普通2026-09-23 09:42

Why an Internal Compass Is More Important Than You Think in the Age of AI

Technology has made it easier than ever to know what to do next. AI can write your emails. Search...

信息来源：DEV Community

教程 / 实战普通2026-09-23 09:04

38% of an analyst's questions get "no records found" when the data is right there

Your text-to-SQL agent has a failure mode that logs nothing, alerts nothing, and returns a...

信息来源：DEV Community

教程 / 实战普通2026-09-23 08:54

How to Stop an AI Agent That Lies About Its Own Spending

In September 2026, Mandiant described a failure in its AI Risk and Resilience report, drawing on data...

信息来源：DEV Community

教程 / 实战普通2026-09-23 08:48

What Is intent.md in Claude Code? A Practical Guide with an Example

You ask Claude Code to build a feature. Minutes later, the code looks reasonable—but it solves a...

信息来源：DEV Community · Claude

教程 / 实战普通2026-09-23 08:46

How I built a 100% client-side privacy toolbox using vanilla JavaScript

Why uploading your files to “free” tools might be costing you more than you think Last month, I...

信息来源：DEV Community

教程 / 实战普通2026-09-23 08:43

I optimized an API for AI agents instead of humans. Here's what one week of data looked like.

I made Korean government data agent-payable with MCP and x402 — here's what actually...

信息来源：DEV Community

教程 / 实战普通2026-09-23 08:35

A free MCP server and REST API for AI-rated job listings (no key)

A free MCP server and REST API for AI-rated job listings (no key) If you want an assistant...

信息来源：DEV Community

教程 / 实战普通2026-09-23 08:29

How Cursor Can Win: Strategy, Defensibility, and Developer Moats

Originally published at sameerbajaj.com. There have been many new entrants in the AI wave over the...

信息来源：DEV Community · Cursor

教程 / 实战普通2026-09-23 08:15

Two AI code review benchmarks disagree on the winner, and agree on what to measure

The top of the Google results for "best AI code review tools" is a wall of vendor listicles. Most of...

信息来源：DEV Community

教程 / 实战普通2026-09-23 08:00

如何在 Together 平台上用 17 美元训练自己的 Jev 分类器

Together AI 在 Qwen3.5 4B 基础上推出 Jev 类分类器 together/Tev1-4B-experimental，已上线其 serverless 平台。该博客展示了如何以 17 美元微调出自己的版本。

信息来源：Together AI 研究与产品博客（RSS） · Qwen

教程 / 实战普通2026-09-23 08:00

高维潜空间的可扩散性：RAE 中 x₀-prediction 如何提升文生图生成

针对 Representation Autoencoders（RAEs）中高保真重建编码器会降低表征有效维度、导致 flow matching 标准速度预测需拟合信号流形外的正交噪声方向、优化低效的问题，论文提出改用干净数据参数化（x₀-prediction），将学习集中在底层信号流形上。在多个强重建编码器实验中，x₀-prediction 一致提升了文生图生成性能。

信息来源：HuggingFace Daily Papers（社区热门论文） · Hugging Face

教程 / 实战精选2026-09-23 08:00

Modal 详解如何以万亿 token 规模服务 Kimi K2.6 编码 Agent 推理

Modal 分享为编码 Agent 提供万亿参数模型 Kimi K2.6 推理服务的优化实践，优化后单副本每用户性能提升 2.8x、副本整体吞吐提升 5.6x，单个服务日处理数千亿 token。

信息来源：Modal 官方工程博客（RSS） · Kimi

教程 / 实战普通2026-09-23 08:00

TypeSafe 决策模型 Jev 使用教程：用 TypeScript 在 OpenRouter 上实现市场商品审核

OpenRouter 发布 Jev 使用教程，Jev 是 TypeSafe 推出的决策模型（模型 ID typesafe/jev-1.13），通过 OpenRouter Decisions API 接收 state 和问题，返回带概率的类型化答案而非生成文本。

信息来源：OpenRouter：Announcements（RSS）

教程 / 实战普通2026-09-23 08:00

自回归视频生成中的记忆机制综述：过去如何框定未来

一篇综述系统梳理了自回归（AR）视频生成中的记忆机制，将记忆定义为跨 AR 步骤持续保留、在原始证据离开上下文后仍能影响后续生成的历史信息。文章从形式、功能、操作、学习与评估五个视角组织文献，并指出可组合且资源感知的记忆架构、可信状态更新、自 rollout 学习与标准化评估等开放挑战。

信息来源：HuggingFace Daily Papers（社区热门论文） · Hugging Face

教程 / 实战普通2026-09-23 07:58

Curriculum Truth Ledger

This is a submission for the Sanity Challenge, Path Two: Vibe-Code Something Strange I built...

信息来源：DEV Community

教程 / 实战普通2026-09-23 07:19

Why Does Your AI Coding Agent Start Forgetting What It Was Doing?

Why Does Your AI Coding Agent Start Forgetting What It Was Doing? I've been working with AI coding...

信息来源：DEV Community

教程 / 实战普通2026-09-23 07:12

We audited 110 AI usage tools. Here is where the numbers go wrong.

A field report on AI usage metering: five recurring bug families across 110 open-source tools, 45+ verified bugs, 23 fixes merged upstream, and one independent reproduction.

信息来源：DEV Community

教程 / 实战普通2026-09-23 06:30

Google Expands Gemini Notebook With Syncing, Interactive Tools and a New Name

Google is expanding Gemini Notebook, the product previously called NotebookLM, from a source-grounded...

信息来源：DEV Community · Gemini

教程 / 实战普通2026-09-23 06:25

The override protocol: how Keel lets the operator overrule a refusal — honestly

Refusals are final unless the operator's word lands — and when it does, the override is a stamped, visible record, not an edit to history. The pressure valve, as a protocol.

信息来源：DEV Community

教程 / 实战普通2026-09-23 06:16

I Built a Virtual Machine Inside the Xbox Sandbox. Then I Let AI Agents Build on Top of It.

XCP turns one-shot code generation into a continuous software lifecycle: create, execute on Xbox,...

信息来源：DEV Community

教程 / 实战普通2026-09-23 05:51

We ran 2 vs 4 agents six times. Four agents cost 2.1 and did not improve success

A preregistered deterministic task, six valid live runs, hard call and cost caps, and a negative result. More agents were operationally viable but did not improve task success, at 2.115× the cost.

信息来源：DEV Community

教程 / 实战普通2026-09-23 05:49

AI Is Making Code Cheap. Verification Is Becoming the Expensive Part

Anthropic published a post about their CI infrastructure that I keep coming back to. The headline...

信息来源：DEV Community · Claude

教程 / 实战普通2026-09-23 05:40

The Leftover Hard Parts

Some time ago, I asked an AI to write a fairly boring piece of code. It did a pretty good job. I...

信息来源：DEV Community

教程 / 实战普通2026-09-23 05:15

The supervisor pattern is not enough: why your agents still fail at step 7 (and the hierarchy that saves them)

I spent four months convinced my supervisor agent was smart. Then I watched it fail on step 7 of a...

信息来源：DEV Community

教程 / 实战普通2026-09-23 05:07

Latent Space 对话 John Platt：Google ERA 如何用 AI 自动化科学发现

Latent Space 发布对 John Platt 的访谈，介绍 Google 的 Empirical Research Assistance（ERA）：用 Gemini 维护实验树并通过 Upper Confidence Bound 规则迭代变异 notebook，以求解可写成评分函数的科学问题。

信息来源：Latent Space（RSS） · Gemini