Full-text search across 71044 articles. Combine with topic and date filters; results sorted by relevance.
A developer outlined a technique for enforcing strictly structured JSON output from large language models by defining Pydantic models as data contracts, generating JSON schemas via model_json_schema(), and validating LLM…
A project called Strata, from developer Niko1221, claims to run the 125-billion-parameter Qwen3.8-Flash-Next mixture-of-experts model on a single 12GB NVIDIA GPU plus 64GB of system RAM by streaming only active expert we…
A compiler called Tang derives GPU kernels from a tensor IR built on affine views and monoid comprehensions, generating code for WGSL, CUDA tensor cores, Metal simdgroup matrices, OpenCL, GLSL and C without hand-written …
A Rust crate called idbd implements Incremental Delta-Bar-Delta, the per-input learning-rate rule Sutton published in 1992, alongside Autostep (Mahmood et al., 2012) as allocation-free online learners with no dependencie…
CISA added CVE-2026-102895, a CVSS 9.8 authentication bypass in the open-source GPUStack GPU cluster manager, to its Known Exploited Vulnerabilities catalog on November 9 with a November 12 remediation deadline after exp…
Blitzkit developer jva.lol ported the ripple physics engine and arcade to WebAssembly for the browser, compiling both to wasm32-unknown-unknown with a single source change — one feature flag on getrandom — but the build …
A developer outlined a patch-gating approach for AI-generated code that isolates each job in its own cloned working directory, asserts strict equality between declared and actually changed files, and fingerprints the tre…
A developer demonstrated a common cost-tracking bug in OpenAI's Responses API: because output_tokens already includes reasoning_tokens, adding the reasoning field to the output count double-bills reasoning, inflating est…
Zyphra released ZONOS3-Mini, a zero-shot voice-cloning text-to-speech model with 100 million active parameters and 800 million total parameters under an Apache 2.0 license, trained on 112 Nvidia H100 GPUs in 2.5 days. Th…
A guide on AI agent memory explains that large language models retain nothing between requests, so the surrounding agent must manage a limited context window — roughly 4,000 to 8,000 tokens for older models, up to 272,00…
Towards Data Science published a tutorial on building multi-agent systems with the OpenAI Agents SDK in Python, walking through an EV purchase adviser and a support agent before showing how fan-out, pipelines, hierarchie…
Developer Mendiak released less-is-more-ui, an open-source design skill that installs via `npx skills add Mendiak/less-is-more-ui` or `npm install @mendiak/less-is-more-ui` to constrain AI coding agents toward minimalist…
A developer outlined a pattern for uploading AI-generated images to object storage from application code, using a PUT request with a deterministic object key, a private bucket, and an idempotency key held constant across…
A developer has released Editmamei, a free desktop tool that connects an AI assistant to existing photo editors such as Photoshop or GIMP so the AI plans an edit while the editor applies it through real layers, masks and…
A developer outlined a workflow for reviewing AI-assisted code edits by placing a reviewable diff between an AI suggestion and the files it touches, using git status --short and git diff --stat to inventory changed paths…
WebAI released TwIL-8B, an 8.8B-parameter reasoning model post-trained from IBM Granite 4.2 8B, which scores 98.2 on MATH-500 and 81.6 on LiveCodeBench while trailing similar-size models on knowledge benchmarks. The rele…
Prompts.chat has turned natural-language prompts into a version-controlled, open-source corpus, converting static prompt engineering into an MDX-driven metaprogramming pipeline and agent graph, according to the project's…
A developer outlined a Node.js pattern for storing AI-generated images in an S3-compatible object store with metadata attached as x-amz-meta- headers, using unique keys of the form images/{generationId}/{sha256}.png rath…
Z.ai's GLM-5.3 scored an overall composite of 82.5 on the iac-bench 2026-10 infrastructure-as-code benchmark, ranking 10th of 16 models between Kimi K3 (83.4) and Qwen3.8 2.4T-A95B (81.6) and third among the five open-we…
Alibaba's Qwen3.8 2.4T-A95B, an open-weight model accessed via OpenRouter and priced at $2.00/$6.00 per million input/output tokens, scored an overall composite of 81.6 on the iac-bench 2026-10 infrastructure-as-code eva…