Show HN: Cross-platform native UI library for all OS https://ift.tt/DbX0dfV

Show HN: Cross-platform native UI library for all OS https://ift.tt/JfzrLxt March 16, 2025 at 11:19PM

Show HN: Nash, I made a standalone note with single HTML file https://ift.tt/gOQkZGn

Show HN: Nash, I made a standalone note with single HTML file Hello HN, I hope it will posted as well. I made a note in single html file. This does not require a separate membership or installation of the software, and if you download and modify an empty file, you can modify and read it at any time, regardless of online or offline. It can be shared through messengers such as Telegram, so it is also suitable to share contents with long articles and images. It is also possible to host and blog because it is static html file content. https://ift.tt/dQjla9n March 14, 2025 at 07:21AM

Show HN: Kill SaaS with Open Source https://ift.tt/qETm5w7

Show HN: Kill SaaS with Open Source KillSaaS is my answer to subscription software in the AI era. I'm building this because I believe small teams can use modern AI tools to create free alternatives to giants like Figma and DocuSign in weeks, not years. We're creating a platform where developers vote on which SaaS to replace, then build it together as open source. wdyt? https://ift.tt/E7L5KoC March 16, 2025 at 02:50AM

Show HN: Basic Memory – Build a knowledge graph from Claude conversations https://ift.tt/aZD1r7v

Show HN: Basic Memory – Build a knowledge graph from Claude conversations Basic Memory is an open-source tool that enables Claude to build and navigate a persistent knowledge graph based on your conversations. It solves the problem of lost context in AI interactions by storing knowledge in standard Markdown files on your computer. I built this because I found myself constantly repeating information to LLMs and wanted a system where my knowledge grew naturally through conversations while maintaining complete control over my data. Demo video: https://ift.tt/Yey7Shb Key features: - Continue conversations exactly where you left off without repetition - All knowledge stays in local Markdown files you can edit anytime - Works with Claude Desktop via the Model Context Protocol - Seamless integration with Obsidian for visualization and editing - Fully open source (AGPL) The system works by creating structure from simple markdown patterns: - Observations with categories: `- [category] fact #tag` - Relations between documents: `- relation_type [[WikiLink]]` or plain `[[Wikilinks]]` - These patterns emerge naturally during conversations When you chat with Claude, you can simply say "Let's continue our conversation about X" and it will build context from your knowledge base, without needing to upload files every time. GitHub: https://ift.tt/u0e5yjL Docs: https://ift.tt/Ikt6KzD Website: https://ift.tt/ldejx9T Requires Claude Desktop or other MCP host and Python 3.12+ I'd love feedback from the HN community, particularly from those interested in knowledge management or AI applications. https://ift.tt/u0e5yjL March 15, 2025 at 11:49PM

Show HN: Psyllium, a Ruby Gem to make Fibers behave more like Threads https://ift.tt/ajOquAK

Show HN: Psyllium, a Ruby Gem to make Fibers behave more like Threads Hi everyone! I created this small Ruby Gem to add some convenient methods to the Fiber class to make it easier to use in the same way a Thread object can be used. This was born out of my frustration that the current implementation of the Fiber class makes it difficult to retrieve the final value of a block passed to a Fiber, especially when creating a fiber via the `schedule` class method. I appreciate any feedback anyone has. https://ift.tt/w6Nc8kn March 15, 2025 at 12:09AM

Show HN: OCR Benchmark Focusing on Automation https://ift.tt/Zx5EkKJ

Show HN: OCR Benchmark Focusing on Automation OCR/Document extraction field has seen lot of action recently with releases like Mixtral OCR, Andrew Ng's agentic document processing etc. Also there are several benchmarks for OCR, however all testing for something slightly different which make good comparison of models very hard. To give an example, some models like mixtral-ocr only try to convert a document to markdown format. You have to use another LLM on top of it to get the final result. Some VLM’s directly give structured information like key fields from documents like invoices, but you have to either add business rules on top of it or use some LLM as a judge kind of system to get sense of which output needs to be manually reviewed or can be taken as correct output. No benchmark attempts to measure the actual rate of automation you can achieve. We have tried to solve this problem with a benchmark that is only applicable for documents/usecases where you are looking for automation and its trying to measure that end to end automation level of different models or systems. We have collected a dataset that represents documents like invoices etc which are applicable in processes where automation is needed vs are more copilot in nature where you would need to chat with document. Also have annotated these documents and published the dataset and repo so it can be extended. Here is writeup: https://ift.tt/ApUyYfn Dataset: https://ift.tt/DarCqpy Github: https://ift.tt/pCVYAHG Looking for suggestions on how this benchmark can be improved further. https://ift.tt/ApUyYfn March 13, 2025 at 02:19AM

Show HN: Pi Labs – AI scoring and optimization tools for software engineers https://ift.tt/9NOlJpB

Show HN: Pi Labs – AI scoring and optimization tools for software engineers Hey HN, after years building some of the core AI and NLU systems in Google Search, we decided to leave and build outside. Our goal was to put the advanced ML and DS techniques we’ve been using in the hands of all software engineers, so that everyone can build AI and Search apps at the same level of performance and sophistication as the big labs. This was a hard technical challenge but we were very inspired by the MVC architecture for Web development. The intuition there was that when a data model changes, its view would get auto-updated. We built a similar architecture for AI. On one side is a scoring system, which encapsulates in a set of metrics what’s good about the AI application. On the other side is a set of optimizers that “compile” against this scorer - prompt optimization, data filtering, synthetic data generation, supervised learning, RL, etc. The scoring system can be calibrated using developer, user or rater feedback, and once it’s updated, all the optimizers get recompiled against it. The result is a setup that makes it easy to incrementally improve the quality of your AI in a tight feedback loop: You update your scorers, they auto-update your optimizers, your app gets better, you see that improvement in interpretable scores, and then you repeat, progressing from simpler to more advanced optimizers and from off-the-shelf to calibrated scorers. We would love your feedback on this approach. https://build.withpi.ai has a set of playgrounds to help you quickly build a scorer and multiple optimizers. No sign in required. https://code.withpi.ai has the API reference and Notebook links. Finally, we have a Loom demo [1]. More technical details Scorers: Our scoring system has three key differences from the common LLM-as-a-judge pattern. First, rather than a single label or metric from an LLM judge, our scoring system is represented as a tunable tree of metrics, with 20+ dimensions which get combined into a final (non-linear) weighted score. The tree structure makes scores easily interpretable (just look at the breakdown by dimension), extensible (just add/remove a dimension), and adjustable (just re-tune the weights). Training the scoring system with labeled/preference data adjusts the weights. You can automate this process with user feedback signals, resulting in a tight feedback loop. Second, our scoring system handles natural language dimensions (great for free-form, qualitative questions requiring NLU) alongside quantitative dimensions (like computations over dates or doc length, which can be provided in Python) in the same tree. When calibrating with your labeled or preference data, the scorer learns how to balance these. Third, for natural language scoring, we use specialized smaller encoder models rather than autoregressive models. Encoders are a natural fit for scoring as they are faster and cheaper to run, easier to fine-tune, and more suitable architecturally (bi-directional attention with regression or classification head) than similar sized decoder models. For example, we can score 20+ dimensions in sub-100ms, making it possible to use scoring everywhere from evaluation to agent orchestration to reward modeling. Optimizers: We took the most salient ML techniques and reformulated them as optimizers against our scoring system e.g. for DSPy, the scoring system acts as its validator. For GRPO, the scoring system acts as its reward model. We’re keen to hear the community’s feedback on which techniques to add next. Overall stack: Playgrounds next.js and Vercel. AI: Runpod and GCP for training GPUs, TRL for training algos, ModernBert & Llama as base models. GCP and Azure for 4o and Anthropic calls. We’d love your feedback and perspectives: Our team will be around to answer questions and discuss. If there’s a lot of interest, happy to host a live session! - Achint, co-founder of Pi Labs [1] https://ift.tt/aKhcI8k https://ift.tt/M5CKJju March 14, 2025 at 07:07PM