Curated AI/ML signals — zero LLMs, pure algorithmic curationWed, 15 Apr 2026 08:24:03 +0000https://github.com/steipete/waclihn:47775628💬 Active HN discussion (34 comments)Wed, 15 Apr 2026 07:04:45 +0000hackernewshttps://stopflock.com/hn:47772012💬 Major HN discussion (161 comments) · 🔥 Trending on HN (624 pts)Tue, 14 Apr 2026 21:56:05 +0000hackernewshttps://www.businesswire.com/news/home/20260414237496/en/Amazon-to-Acquire-Globalstar-and-Expand-Amazon-Leo-Satellite-Networkhn:47775183🔥 48 pts on Hacker NewsWed, 15 Apr 2026 05:54:29 +0000hackernewshttps://iczelia.net/posts/e16-20-year-old-bug/hn:47774789🔥 89 pts on Hacker NewsWed, 15 Apr 2026 04:47:00 +0000hackernewshttps://huggingface.co/papers/2604.10966hf:2604.10966💬 Active HN discussion (32 comments)Wed, 15 Apr 2026 08:22:57 +0000huggingfacehttps://huggingface.co/papers/2604.11689hf:2604.11689LARY: A Latent Action Representation Yielding Benchmark for Generalizable Vision-to-Action AlignmentWed, 15 Apr 2026 08:22:57 +0000huggingfacehttps://huggingface.co/papers/2604.12627hf:2604.12627💬 Major HN discussion (171 comments) · 📄 Featured on Hugging Face Daily Papers — RLVR improves reasoning in large language models, but its effectiveness is often limited by severe reward sparsity on hard problems. Recent hint-based RL methods mitigate sparsity by injecting partial solutions or abstract templates, yet they typically scale guidance by adding more tokens, which ...Tue, 14 Apr 2026 11:53:23 +0000hfcs.AIhttp://arxiv.org/abs/2604.12989v1arxiv:2604.12989v1📄 New in cs.CL — Speculative decoding accelerates autoregressive language models by using a lightweight drafter to propose multiple future tokens, which the target model then verifies in parallel. DFlash shows that a block diffusion drafter can generate an entire draft block in a single forward pass and achieve s...Tue, 14 Apr 2026 17:23:14 +0000cs.CLhttp://arxiv.org/abs/2604.13018v1arxiv:2604.13018v1📄 New in cs.CL — Autonomous AI research has advanced rapidly, but long-horizon ML research engineering remains difficult: agents must sustain coherent progress across task comprehension, environment setup, implementation, experimentation, and debugging over hours or days. We introduce AiScientist, a system for au...Tue, 14 Apr 2026 17:55:16 +0000cs.CLhttp://arxiv.org/abs/2604.13016v1arxiv:2604.13016v1📄 New in cs.LG, cs.AI — On-policy distillation (OPD) has become a core technique in the post-training of large language models, yet its training dynamics remain poorly understood. This paper provides a systematic investigation of OPD dynamics and mechanisms. We first identify that two conditions govern whether OPD succe...Tue, 14 Apr 2026 17:54:28 +0000cs.LGcs.AIcs.CLhttp://arxiv.org/abs/2604.13030v1arxiv:2604.13030v1📄 New in cs.CV — While diffusion models dominate the field of visual generation, they are computationally inefficient, applying a uniform computational effort regardless of different complexity. In contrast, autoregressive (AR) models are inherently complexity-aware, as evidenced by their variable likelihoods, bu...Tue, 14 Apr 2026 17:59:03 +0000cs.CVhttp://arxiv.org/abs/2604.12890v1arxiv:2604.12890v1📄 New in cs.CV, cs.AI — Multimodal deep search agents have shown great potential in solving complex tasks by iteratively collecting textual and visual evidence. However, managing the heterogeneous information and high token costs associated with multimodal inputs over long horizons remains a critical challenge, as exist...Tue, 14 Apr 2026 15:40:28 +0000cs.CVcs.AIhttp://arxiv.org/abs/2604.12887v1arxiv:2604.12887v1📄 New in cs.CV, cs.LG — Visual tokenizers map high-dimensional raw pixels into a compressed representation for downstream modeling. Beyond compression, tokenizers dictate what information is preserved and how it is organized. A de facto standard approach to video tokenization is to represent a video as a spatiotemporal ...Tue, 14 Apr 2026 15:37:30 +0000cs.CVcs.LGhttp://arxiv.org/abs/2604.13036v1arxiv:2604.13036v1📄 New in cs.CV — Recent advances in video generation enable a new paradigm for 3D scene creation: generating camera-controlled videos that simulate scene walkthroughs, then lifting them to 3D via feed-forward reconstruction techniques. This generative reconstruction approach combines the visual fidelity and creat...Tue, 14 Apr 2026 17:59:44 +0000cs.CVhttps://github.com/NousResearch/hermes-agentgithub:NousResearch/hermes-agent⭐ 8.3k GitHub stars today · ⭐ 86.8k total stars — The agent that grows with youWed, 15 Apr 2026 08:22:57 +0000pythonhttps://github.com/HandsOnLLM/Hands-On-Large-Language-Modelsgithub:HandsOnLLM/Hands-On-Large-Language-Models⭐ 35 stars today on GitHub · ⭐ 25.1k total stars — Official code repo for the O'Reilly Book - "Hands-On Large Language Models"Wed, 15 Apr 2026 08:22:57 +0000jupyter notebookhttps://github.com/vllm-project/vllmgithub:vllm-project/vllm⭐ 162 GitHub stars today · ⭐ 76.7k total stars — A high-throughput and memory-efficient inference and serving engine for LLMsWed, 15 Apr 2026 08:22:57 +0000pythonhttps://nitin1353.medium.com/how-ai-agents-are-powering-chinas-new-one-person-companies-9f8b8d89878c?source=rss------ai_agent-5rss:https://medium.com/p/9f8b8d89878c🆕 New article — <div class="medium-feed-item"><p class="medium-feed-image"><a href="https://nitin1353.medium.com/how-ai-agents-are-powering-chinas-new-one-person-companies-9f8b8d89878c?source=rss------ai_agent-5"><img src="https://cdn-images-1.medium.com/max/1536/1*DmmcmrbSVAOM7VhSoo3uVg.png" width="1536" /></a>...Wed, 15 Apr 2026 08:19:55 +0000startupinnovationbusiness-strategyhttps://github.com/Lordog/dive-into-llmsgithub:Lordog/dive-into-llms⭐ 1.1k GitHub stars today · ⭐ 28.8k total stars — 《动手学大模型Dive into LLMs》系列编程实践教程Wed, 15 Apr 2026 08:22:57 +0000jupyter notebookhttps://github.com/alirezarezvani/claude-skillsgithub:alirezarezvani/claude-skills⭐ 195 GitHub stars today · ⭐ 11.2k total stars — 232+ Claude Code skills & agent plugins for Claude Code, Codex, Gemini CLI, Cursor, and 8 more coding agents — engineering, marketing, product, compliance, C-level advisory.Wed, 15 Apr 2026 08:22:57 +0000python