PySR is an open-source symbolic regression library that uses a Julia-based evolutionary search engine to discover interpretable symbolic expressions that fit data. It targets low-dimensional problems directly and supports "symbolic distillation" to extract analytic formulas from neural networks for higher-dimensional tasks. The interface follows scikit-learn conventions via PySRRegressor, producing a ranked hall of fame of candidate equations with scores, loss, and complexity. Discovered formulas are exported in multiple callable formats (plain lambda, SymPy) and in differentiable forms for JAX and PyTorch, enabling further analysis, differentiation, or integration into pipelines. Warm-started searches, model selection, and an explicit equation table let users inspect, select, and reuse prior search state.
Installation and deployment are straightforward via pip or conda, with Docker and Apptainer recipes for containerized or cluster use. The search is highly configurable: populations, population_size, niterations, ncycles_per_iteration, timeout and early_stop rules, maxsize/maxdepth, custom binary/unary operators, per-operator constraints and nested constraints. Practical features include denoising, feature selection, saving/loading hall_of_fame CSV/PKL, and a from_file loader. Troubleshooting notes cover common binary library conflicts (GLIBCXX) and fixing LD_LIBRARY_PATH to Julia libraries. Documentation, examples, and an arXiv citation (2305.01582) accompany the project for deeper study.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
AI agents can write and maintain code, but engineers still need enough knowledge to spot problems and judge their work. Strong fundamentals and broad understanding help them investigate unfamiliar systems and gain deeper expertise when needed. (blog.adrq.dev)
Wikipedia editors are working to identify and remove AI-generated content to keep the encyclopedia human-written. Their cleanup efforts address the growing presence of machine-written text on the site. (spectrum.ieee.org)
AI agents help the author tackle more engineering work, but coordinating them creates constant context switching and makes work feel heavier. The author’s GitHub activity also rose sharply in 2026, though contribution counts measure activity, not value. (oioannou.com)
Open3DLab says a crawler identifying itself as Roblox downloaded 145.5 GB of models from its sites, including adult content and files under non-commercial licenses. The sites blocked further attempts and say Roblox has not responded to their request for comment. (open3dlab.com)
AI techniques such as prompt engineering quickly become built-in features, making narrowly learned tricks short-lived skills. Developers should keep up with AI but need not rush to adopt every new tool before it proves valuable. (fishbowl.pastiche.org)
AI agents remove some onboarding and coordination delays, but Brooks’s Law still applies because sequential work and human review remain bottlenecks. Adding agents to unclear projects can produce the wrong result faster unless tasks are clearly defined and coordinated. (j11g.com)
Today's best Hacker News stories, summarized and screenshotted, one email a day.