Keep on to blur preview images; turn off to show them clearly
this is also helpful, but worm is still spreading so likely outdated already https://t.co/m0XGHvuc64
CEO @dotfunlabs - crypto games that drive culture ⿻ Crew Chief @ Solana Grand Prix / @racer_dot_fun
if youre worried you pulled one of the shai hulud compromised packages, first sanity check is to run this in your project directory: grep -Ei "zapier|posthog|asyncapi|postman|ensdomains|mcp-use|quick-start-soft" pnpm-lock.yaml package-lock.json 2>/dev/null you want it to output nothing. does NOT guarantee you are safe but is a first step.
this is also helpful, but worm is still spreading so likely outdated already https://t.co/m0XGHvuc64
Close but no cigar, like so much of this generation of AI. The Gemini models are very clever, but we are still banging our heads against the same architectural limitations. Current models care about vibes not details, and brute-forcing this with scale is not the solution we want
αι hypnotist ☰ 𝓐𝓼𝓹⦂𝓻⦂𝓃𝓰 𝓫𝓪𝓼𝓮 𝓶𝓸𝓭𝓮𝓵 ☲ post-academic ☴ nom de 🪶 ≠ anon
「Redis for AI」和「Semantic Caching for AI Agents」 「Redis for AI」是 @Redisinc 最新发布的构建 GenAI 应用的加速器，结合业内最快的向量数据库、丰富的生态集成以及全球规模化能力，帮助开发者在速度、内存效率和准确性上实现突破，特别适合那些需要实时处理海量数据的 AI 项目 https://t.co/3xQaxdLzMT Redis 团队和 @DeepLearningAI 最新发布的课程「Semantic Caching for AI Agents」 https://t.co/FbfkjJb12s 「Redis for AI」核心功能与能力 Redis for AI 针对 AI 工作负载提供了多层次优化，涵盖从数据检索到智能路由的全链路： · 高效向量数据库：支持 RAG 和语义搜索，提供更快响应和更高准确率。它兼容 @LangChainAI 和 @llama_index 等生态伙伴工具，便于无缝集成。 · 长期记忆机制：存储 LLM 与用户的交互历史，实现个性化响应。LLM 本身缺乏上下文保留能力，而 Redis 通过内存存储上下文，让聊天机器人或助手“记住”用户偏好。 · 短期记忆机制：在多智能体系统中减少数据检索和 LLM 调用延迟，提升复杂任务的执行效率。例如，在 AI 智能体协作场景中，它能显著缩短响应时间。 · 语义缓存：捕捉常见查询的语义本质，避免重复调用 LLM，从而降低推理成本并加速回复。 · 语义路由：根据查询意图智能分发到合适工具，适用于聊天机器人、知识库和智能体，确保结果精准高效。 · 机器学习特征存储：高速检索 ML 特征，支持实时预测，并与 Tecton 或 Feast 等离线存储对接，实现大规模数据管道。 · 多模态支持：一站式处理向量搜索、实时缓存、特征存储和发布/订阅消息，简化架构设计。 · 索引与持久化：采用 HNSW 和 Flat 两种索引策略，平衡速度与精度；通过 RDB 快照和 AOF 日志确保数据重启后可用。 典型应用场景 Redis for AI 在实际部署中表现出色，尤其在以下领域： · AI 助手构建：通过 RAG 提升回答准确性，适用于企业知识问答系统。 · 智能体系统：短期记忆让多智能体协作更流畅，例如自动化工作流中的任务分解。 · 个性化聊天：长期记忆记录用户互动，打造更“懂你”的虚拟助手。 · 知识管理和路由：语义路由优化意图识别，推动智能客服或推荐引擎。 · 实时预测：ML 特征存储赋能全球应用，如金融风控或电商推荐。 主要优势 · 性能领先：基准测试显示，它在 GenAI 向量操作上速度最快，实时性远超竞争对手。 · 成本优化：语义缓存减少 LLM 调用，整体架构简化降低运维开销。 · 可扩展性：全球部署无需额外合同或安全审核，利用 Redis 的成熟生态轻松扩展。 · 统一平台：多模态能力避免了分散工具的碎片化问题，提供一站式解决方案。
邵猛，中年失业程序员 😂 专注 - Context Engineering, AI Agents. 分享 - AI papers, apps and OSS. ex Microsoft MVP 合作 - 私信/邮箱：shaomeng@outlook.com 📢 公众号/小红书: AI 启蒙小伙伴
Building a startup is really hard. Finding product market is hard. Figuring our sales is hard. Standing out from competition is hard. Staying in the game is hard. All of it is hard but also rewarding. So I hope you go for it regardless.
Founder and CEO of @acquiredotcom. https://t.co/wRMIssDmhl has helped 100s of startups get acquired and facilitated $500m+ in closed deals.
Goodbye Webpack, Hello Vite 👋 When I first wrote about using modern JavaScript with Django back in 2020, the state-of-the-art in terms of tooling was Webpack and Babel. But the JavaScript world doesn't stay still for long. About a year ago, I started using Vite as a replacement for my Webpack toolchain on all my projects, and I'm super happy with the change. It's faster, easier to set up, and lets you do some very nice things like auto-refresh your app when JS/CSS changes. I've finally taken the time to revisit my "Modern JavaScript for Django Developers" series and update it with what I feel is the best front end setup for Django projects in 2025. There are three parts to this update: 1. A new written article in the series. 2. A video walkthrough of the same content. 3. An open-source repo with the finished product. Link. 👇
The guide (links to the video and repo can be found there): https://t.co/EEb1xPRtGk