AllyshipContent Radar

Hands-on 278

Công cụ, thử nghiệm và hướng dẫn thực hành

Crawl live-1795090500000 · 19:15 19 thg 11, 2026 · Xuất bản 19:35 19 thg 11, 2026

91

I got this cold email today about what I'd teach kids interested in web development and programming. I figured I'd share my full response here in case it helps others:

42.3

Hướng dẫn leerob

Người viết cho rằng học cách tạo phần mềm hiện nay có thể nhìn lạc quan hoặc bi quan, và AI đã thay đổi cách phần mềm được xây dựng. Họ nói rằng cần tập luyện và trau dồi “idea machine” để tạo ra nhiều ý tưởng hơn và tinh chỉnh gu để nhận ra ý tưởng tốt sớm hơn. Cách dạy ưa thích của họ là giúp người học xây thứ mà họ hào hứng trước, rồi khi gặp vướng mắc mới giải thích khái niệm CS và web development liên quan. Họ cũng nhắc rằng các mô hình AI hiện tại là tệ nhất mà chúng ta sẽ có, nên cần đặt mục tiêu tham vọng hơn khi yêu cầu AI.

X · 10 thg 11, 2026

Biên tập tiếng Việt
Chi tiết nguồn

I got this cold email today about what I'd teach kids interested in web development and programming. I figured I'd share my full response here in case it helps others: — Learning to create software today is a glass half full / half empty situation. You can be optimistic or pessimistic. It is strange and transformative that we no longer write most code by hand. AI has completely changed how software is built. I expect this to change even more dramatically in the coming years, as entirely new programming languages¹ are invented² and adopted. One person might look at this and be pessimistic: what will these computer scientists and programmers do, as AI continues to become superhuman in its intelligence? But the wiser of the bunch will realize there is much to be optimistic about. Those with the agency³ to create and explore their wildest ideas have unlimited power. There has quite frankly never been a better time in the history of humanity to have good ideas⁴. So my advice for those building at summer camps is simple. You have to hone and craft your "idea machine". You are going to have lots of bad ideas, but you'll get better at coming up with ideas and refining your taste so you can recognize and act on great ideas sooner. People at summer camps will spend less time reading CS tutorials and more time thinking about the fun or interesting things they want to exist in the world. My favorite way of teaching and getting people interested in software is to first help them build something they are excited about. When they hit roadblocks that require a deeper understanding of a CS concept or something related to web development, only then do we take the time to explain what it is and how it solves their problem. Starting this way, at least for me, seems to get people hooked faster than needing hours and hours of prereqs. There's more desire to spend time learning the fundamentals after having the "aha!" moment. It's worth checking out some skills⁵ designed to help encode the best practices for teaching and then plop that into a coding agent or your AI tool of choice and give it a task: teach me about how databases work (or similar). It's pretty great. It's also worth remembering that AI models today are the worst they will ever be. You need to learn how to be even more ambitious⁶ with what you ask them. Good luck! — ¹ https://x.com/poteto/status/2119824409087348984 ² https://bend-lang.com ³ https://www.neelnanda.io/blog/44-agency ⁴ https://www.are.na/editorial/notes-on-taste ⁵ https://github.com/mattpocock/skills/blob/main/skills/productivity/teach ⁶ https://felixrieseberg.com/how-to-be-more-ambitious-a-tutorial-for-humans-with-models

  • It's also worth remembering that AI models today are the worst they will ever be.
  • You have to hone and craft your "idea machine".
  • My favorite way of teaching and getting people interested in software is to first help them build something they are excited about.
  • You need to learn how to be even more ambitious⁶ with what you ask them.
92

// Harnessing Recursive Self-Improvement for Training Infrastructure //

42.2

Hướng dẫn dair_ai

// Harnessing Recursive Self-Improvement for Training Infrastructure // Want to learn how to build agents that speed up LLM training jobs? Check out this paper. InfraRSI raised mean MFU from 17.4% to 30.0% across four dense and MoE training workloads, with speedups up to 2.75x. Why it matters: Coding agents asked to speed up training often produce false speedups. They change the effective batch size, move optimizer steps outside the timing window, or apply unstable low-precision casts. InfraRSI locks a task contract over workload, loss, and timing before any edit. It profiles the baseline and the cluster noise, then searches launch settings and code patches together and re-checks the best candidates against the baseline. It also fits time and peak-memory models across every trial, including out-of-memory failures, and stores what it learns for use on other GPUs. https://arxiv.org/abs/2611.11558 Chat with Paper: https://academy.dair.ai/papers/infrarsi-harnessing-recursive-self-improvement-for-training-infrastructure-2611.11558

X · 10 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
93

Eval-driven development (EDD) is TDD for AI agents. Great article from @LangChain

41.4

Hướng dẫn vitalflux

Eval-driven development (EDD) is TDD for AI agents. Great article from @LangChain Stop tweaking prompts and guessing whether things got better. Measure every change against a fixed set of evals. Here’s how to do it 🧵Read the thread to learn. #agenticai #AIAgents https://www.langchain.com/blog/what-is-eval-driven-development

X · 10 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
94

Most of you probably haven't noticed this, but dots have a new feature called multi-exec that we quietly released on friday. I've been using it all weekend on my personal dot and it's incredibly power

41.1

Hướng dẫn pashmerepat

Most of you probably haven't noticed this, but dots have a new feature called multi-exec that we quietly released on friday. I've been using it all weekend on my personal dot and it's incredibly powerful. I'd like to share why this is a big deal and how to use it. At a high level, multi exec lets your dot run commands directly on any computer. This immediately turns your dot into a real virtual employee. Here's why it's a gamechanger: - you can attach a powerful machine with persistent storage to your dot (I attached it to my maxed out Mac Studio Ultra) - the dot can now drive your computer directly via Computer Use - this extends to browser use - all your websites stay logged in, residential ip etc - you set up all your CLIs and authentication ONCE and the dot can use them directly (no more fighting the default ephemeral cloud computer) - the dot can save skills locally for repeated workflows and subagents can use those directly - if you have a local obsidian memory vault for your agents, your dot can use that too - building ios app end to end and testing it in xcode, and showing me screenshots of progress as it works - all of the above also applies to any subagents your dot delegates to in the cloud (subagents share context unlike separate threads). To enable it, open the desktop app on your computer and hit "allow computer", then give your dot this prompt: "Hey I just connected my computer. Can you attach to it directly so it's your default computer and you can run commands directly on it? Do it in your root thread so that all subagents inherit this computer for running commands. Run whoami to run water through the pipes and verify it works end to end." I recommend attaching your dot to the most powerful machine you have and investing time into it up front: - Log into all the emails & websites you need e.g. stripe cli, gcloud, firebase, etc - Install and auth all the CLIs, git you need - Make sure computer use is turned on and allowed to use your computer while its locked - I also enabled Allow All for Browsing and Downloads so my dot could use computer use to download whatever it wanted from websites without asking me for approval for each domain (use at your own risk). - Important: I HIGHLY recommend to ask your dot to build out repeated workflows as skills, and iterate to make those skills bulletproof by running water through the pipes with its subagents. As of this weekend, my dot is sending me screenshots of its progress building my iOS app end to end on my Mac Studio Ultra via xCode, and clicking around and testing things end to end. It's able to configure things in stripe via the CLI and use websites when it needs to download images for my workflows. It’s night and day. Don't sleep on it.

X · 10 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
95

So bullish on MCP. Its utility exceeds the "M" part: even AI aside, it'll be a civilization-scale accomplishment to get all our systems communicating through one universal open protocol.

40.7

Hướng dẫn rauchg

So bullish on MCP. Its utility exceeds the "M" part: even AI aside, it'll be a civilization-scale accomplishment to get all our systems communicating through one universal open protocol. A missing piece was how to get arbitrary callbacks across sessions. MCP could read and write, but it couldn't tell you later "this event happened." MCP Events fixes this. We're adding support in our own MCP. Say an agent makes a deployment, and wants to know when it's complete. Instead of polling, it can subscribe to 𝚍𝚎𝚙𝚕𝚘𝚢𝚖𝚎𝚗𝚝.𝚜𝚞𝚌𝚌𝚎𝚎𝚍𝚎𝚍. PSA: With Vercel Queues and Workflows, writing the backend machinery for your own MCPs to do this is easy peasy.

X · 10 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
96

Great work on harness self-improvement from Microsoft and colleagues.

40.7

Hướng dẫn dair_ai

Great work on harness self-improvement from Microsoft and colleagues. It discusses how to build long-running agents that improve their own harness, It turns out that if you run agents for hours, giving them a way to fix their own harness during the run improves results on long-running problems. More from the paper: Harness optimization usually runs offline. You tune the harness on tasks that have reference answers, then reuse it. Long open-ended tasks have no reference answers, and they run into problems the tuned harness never saw. BOXBOX moves that loop into the task itself. The agent decides when to revise its harness, and a kernel hot-reloads the new version while keeping the full execution context, so the run continues without a restart. They test it on PerfRace, a new benchmark where agents spend hours speeding up real software projects. Under the same time and cost budget, BOXBOX reaches a 1.70x speedup at four hours, against 1.54x for the best baseline. https://arxiv.org/abs/2611.09218 Chat with Paper: https://academy.dair.ai/papers/boxbox-recursive-harness-self-improvement-for-long-running-performance-optimizat-2611.09218

X · 10 thg 11, 2026

Nguyên văn nguồn
Chi tiết nguồn
91–96 / 179
Trang 16 / 30