stream disconnected before completion: Transport error: timeout 问一下 最近频繁出现这个问题 自己接的cpa 是vps的问题还是我的网络问题啊 之前好像基本没有这个问题的 这两天好严重 求佬友解答 1 个帖子 - 1 位参与者 阅读完整话题
sk-b1da2a945c2c2d16243f59457f1dfe50681dfb2bf1a4618de4bb803bcc5740f0 ai.dianshangbat.cn Sub2API - AI API Gateway 这使用起来感觉怪怪的，改项目老改的面目全非逻辑混乱，我都无语了，但是我又不会测 是不是真的gpt5.4 xhigh ，来个佬帮帮孩子吧 13 个帖子 - 6 位参与者 阅读完整话题
参考豆包专家的资料整的 英文提示词通用: transform this anime character into a realistic human, maintain exact hairstyle, hair color, eye color, facial proportions and expression, natural skin pores, detailed eyes and lips, even studio lighting, clear facial contours 多图的话，把this改成these ，Character后面加个s 删掉into后面的a 不然傻傻的豆包可能就帮你把所有图片融合起来，生成一张图片了() 直接看看豆包5.0效果吧，我不是很喜欢，但是某些AI形象创建必须要真实人脸 把二次元的头部比例复刻到这里还是很奇怪喵 3 个帖子 - 3 位参与者 阅读完整话题
Transcribe the audio (passing it in at 2x speed) via whisper, use a cheap LLM to identify the ad snippets, and then take it out! Have been using for the past couple days - it's not perfect but I've enjoyed using it. Feel free to add your episodes! https://github.com/mergd/podads Comments URL: https://news.ycombinator.com/item?id=47813346 Points: 3 # Comments: 1
PayPal's crypto services will transition from PayPal, Inc. to PayPal Digital, Inc. effective on or after April 20, 2026. Comments URL: https://news.ycombinator.com/item?id=47806960 Points: 1 # Comments: 0
The Asymmetric Numeral Systems (ANS) algorithm (Duda et al, 2015) is perhaps the most practical way of getting near optimal compression ratios for independent and identically distributed random sequences of symbols from a known discrete probability distribution. Simplest example: a random sequence of 0’s and 1’s with probability p of getting a 1. Shannon’s entropy formula gives us the expected compression ratio for such a sequence, but realizing that compression ratio efficiently in a computer program is not such an easy task. ANS does the trick and is incorporated into several general purpose compression algorithms, but I wasn’t able to track down a simple, self-contained implementation that was reasonably performant. So I made simple_ans, a straightforward Python package that uses a small, yet efficient, kernel of C++ code (few hundred lines of code). If you want it even simpler , there’s also a pure Python implementation in the repo (much slower though). I hope you find it interesting and/or useful! Comments URL: https://news.ycombinator.com/item?id=47806122 Points: 2 # Comments: 1
Built a library to translate between OpenAI Responses/Assistants APIs and other provider APIs. Provides full compatibility so it’s a total drop-in regardless of which provider you use or which features (computer use, web search). End goal will be translating between any SDK and any other SDK with full features, no proxies, no latency. Comments URL: https://news.ycombinator.com/item?id=47798806 Points: 1 # Comments: 0
I built an open-source research agent. You ask a question, it searches the web via Tavily, synthesizes an answer with an LLM, and shows the sources it used. Answers stream in real-time. The interesting part is the backend. It's a single JS file (~100 lines) that handles web search, LLM streaming, and per-user conversation history. No vector database, no Redis, no separate storage service. It runs inside a cell — an isolated environment with a built-in database, search index, and filesystem. The cell handles persistence and streaming natively, so the agent code only has to deal with the actual logic. Tech: Next.js frontend, Tavily for search, OpenRouter for LLM (Gemini 2.5 Flash default). Demo: https://youtu.be/jvTVA7J925Y Comments URL: https://news.ycombinator.com/item?id=47797393 Points: 6 # Comments: 0
Article URL: https://www.sansparsh.art/modulo Comments URL: https://news.ycombinator.com/item?id=47796840 Points: 2 # Comments: 5
Hey HN! We're Charles and Dean, and we're building Stage: a code review tool that guides you through reading a PR step by step, instead of piecing together a giant diff. Here's a demo video: https://www.tella.tv/video/stage-demo-1pph . You can play around with some example PRs here: https://stagereview.app/explore . Teams are moving faster than ever with AI these days, but more and more engineers are merging changes that they don't really understand. The bottleneck isn't writing code anymore, it's reviewing it. We're two engineers who got frustrated with GitHub's UI for code review. As coding agents took off, we saw our PR backlog pile up faster than we could handle. Not only that, the PRs themselves were getting larger and harder to understand, and we found ourselves spending most of our time trying to build a mental model of what a PR was actually doing. We built Stage to make reviewing a PR feel more like reading chapters of a book, not an unorganized set of paragraphs. We use it every day now, not just to review each other's code but also our own, and at this point we can't really imagine going back to the old GitHub UI. What Stage does: when a PR is opened, Stage groups the changes into small, logical "chapters". These chapters get ordered in the way that makes most sense to read. For each chapter, Stage tells you what changed and specific things to double check. Once you review all the chapters, you're done reviewing the PR. You can sign in to Stage with your GitHub account and everything is synced seamlessly (commenting, approving etc.) so it fits into the workflows you're already used to. What we're not building: a code review bot like CodeRabbit or Greptile. These tools are great for catching bugs (and we use them ourselves!) but at the end of the day humans are responsible for what gets shipped. It's clear that reviewing code hasn't scaled the same way that writing did, and they (we!) need better tooling to keep up with the onslaught of AI generated code, which is only going to grow. We've had a lot of fun building this and are excited to take it further. If you're like us and are also tired of using GitHub for reviewing PRs, we'd love for you to try it out and tell us what you think! Comments URL: https://news.ycombinator.com/item?id=47796818 Points: 8 # Comments: 4
I trained a transformer in HyperCard. 1,216 parameters. 1989 Macintosh. And yes, it took a while. MacMind is a complete transformer neural network, embeddings, positional encoding, self-attention, backpropagation, and gradient descent, implemented entirely in HyperTalk, the scripting language Apple shipped with HyperCard in 1987. Every line of code is readable inside HyperCard's script editor. Option-click any button and read the actual math. The task: learn the bit-reversal permutation, the opening step of the Fast Fourier Transform. The model has no formula to follow. It discovers the positional pattern purely through attention and repeated trial and error. By training step 193, it was oscillating between 50%, 75%, and 100% accuracy on successive steps, settling into convergence like a ball rolling into a bowl. The whole "intelligence" is 1,216 numbers stored in hidden fields in a HyperCard stack. Save the file, quit, reopen: the trained model is still there, still correct. It runs on anything from System 7 through Mac OS 9. As a former physics student, and the FFT is an old friend, it sits at the heart of signal processing, quantum mechanics, and wave analysis. I built this because we're at a moment where AI affects all of us but most of us don't understand what it actually does. Backpropagation and attention are math, not magic. And math doesn't care whether it's running on a TPU cluster or a 68030 from 1989. The repo has a pre-trained stack (step 1,000), a blank stack you can train yourself, and a Python/NumPy reference implementation that validates the math. Comments URL: https://news.ycombinator.com/item?id=47792525 Points: 10 # Comments: 0
https://www.reuters.com/world/china/chinese-chipmaker-ymtc-plans-new-factories-amid-heightened-us-sino-trade-2026-04-14/ 2 个帖子 - 2 位参与者 阅读完整话题
For how lofty Anthropic’s Mythos claims are, the harness is confusingly stupid. From the report, it ranks every file by “how sus it sounds,” loops over each with curt instructions to “find a bug,” hands candidates to a judge + ASan checker— and zero-days simply pop out. That should not work. But it does. On miniupnp with a $20 plan, Opus 4.6 reliably rediscovers known CVEs in older versions and even surfaced a new remote global buffer overflow (non-default config). So what happens if the harness is actually good—i.e. equipped with proper security tooling? I’m a student, not a security engineer, so I'd would love ideas or critiques on my planned tool roadmap. (If you have a $200 plan with extra usage lying around, try it out to see if it churns a zero-day in your own C) Comments URL: https://news.ycombinator.com/item?id=47786408 Points: 3 # Comments: 0
Article URL: https://ghostmeet.sshlab.dev/ Comments URL: https://news.ycombinator.com/item?id=47780614 Points: 3 # Comments: 0
Hello Hackerians. Please check out another CD (change directory) utility: cd-deluxe. Why? Using command line is great and powerful but the humble default cd command is somewhat weak. And it is one of the most used commands. Here is another attempt at improving it! Examples: Use "cd --" to go back two directories, "cd ---" to go back 3 directories, etc. Or go in the opposite direction ie "cd +4" to go to the fourth directory visited from the start (or any number). Use "cd ," (comma) to go to the most commonly visited directory (comma meaing "common"). Also see "cd ,,", "cd ,3", etc. List directories visited either in reverse, forward or most common orderings. Plus more - see github. Goal was to be lightweight and fast. Works from the directory stack instead any file based storage. Integrates with bash/zsh/fish/powershell/cmd.exe. With downloads for Linux/maxOS/Windows or build the C++ from source via CMake. Please have a look and let me know of any thoughts. Thanks! Comments URL: https://news.ycombinator.com/item?id=47780155 Points: 2 # Comments: 1
I wanted to learn the language of my gf but wasn't satisfied with duolingo so I built my own language learning app. Comments URL: https://news.ycombinator.com/item?id=47779726 Points: 2 # Comments: 0
有段时间没更新了，今天更新了一下社区专属版，共计TG频道125个，搜索插件85个 本次新增11个插件如下 duanjuw dyyjpro gaoqing888 jupansou kkv lingjisp panlian qingying quarktv yulinshufa zxzj ps.252035.xyz PanSou 盘搜 10 个帖子 - 10 位参与者 阅读完整话题
每天下午3-4点之间codex几乎必报错stream disconnected before completion: Transport error: network error: error decoding response body 3 个帖子 - 3 位参与者 阅读完整话题
Article URL: https://dubvox.coderluii.dev/ Comments URL: https://news.ycombinator.com/item?id=47775496 Points: 2 # Comments: 1
GitHub GitHub Copilot · Plans & pricing GitHub Copilot works alongside you directly in your editor, suggesting whole lines or entire functions for you. 8 个帖子 - 6 位参与者 阅读完整话题