Top all-time · #20219
Grigory Sapunov
Personal blog · arxiviq.substack.comaiscience
Grigory Sapunov writes ArXivIQ, a personal blog at arxiviq.substack.com about ai and science. Since 2025, 3 different Hacker News readers have linked it 33 times. It is still writing: the last post was on 2026-11-02, and you can follow it by feed.
Too few sampled comments (3) to say how often HN recommends it.?
Why HN links it
HN readers link these as summaries of arXiv AI-research papers, such as DeepConf and GEPA. One commenter says the articles are fully or almost fully LLM-generated, and another says all of the site's articles are.
Summarised by Claude from a sample of 3 HN comments that link here. Each comment is marked recommending, critical or neutral (citing without judging); how this works.
Hand-picked ML papers, AI-generated deep dives. Daily TLDR + architecture notes—expert curation, automatic insight.
What Hacker News says?
> Lot of work went into it; even if the final was LLMed. No, it was fully or almost fully LLM generated. See: https://arxiviq.substack.com/p/coming-soon
Comment linking /p/coming-soon
on “DeepConf: Scaling LLM reasoning with confidence, not just compute”Submitted DeepConf: Scaling LLM reasoning with confidence, not just compute as a story · 98 points
HN discussionThis article, like all articles on this substack, is LLM generated. Source: https://arxiviq.substack.com/p/coming-soon
Comment linking /p/coming-soon
on “DeepConf: Scaling LLM reasoning with confidence, not just compute”Submitted GEPA: Reflective prompt evolution can outperform reinforcement learning as a story · 92 points
HN discussionThis summary article is LLM authored [1], but it does seem to make sense. HN folks apparently agree, with 58 points for the submission so far. 1: https://arxiviq.substack.com/p/coming-soon > ArXivIQ exists to turn that fire-hose into jet-streams of insight: every paper is hand-picked by a human editor, then pushed through a purpose-built multi-agent AI pipeline that dissects methods, experiments, and limitations in…
Comment linking /p/coming-soon
on “GEPA: Reflective prompt evolution can outperform reinforcement learning”Submitted Nested Learning: The Illusion of Deep Learning Architectures as a story · 6 points
HN discussionSubmitted Solving the compute crisis with physics-based ASICs as a story · 5 points
HN discussionSubmitted Speed Always Wins: A Survey on Efficient Architectures for Large Language Models as a story · 3 points
HN discussionSubmitted Text-to-LoRA Enables On-the-Fly Model Adaptation as a story · 3 points
HN discussionSubmitted Generalization Dynamics of LM Pre-Training as a story · 2 points
HN discussionSubmitted LeJEPA: Provable and Scalable Self-Supervised Learning Without the Heuristics as a story · 2 points
HN discussionSubmitted Context Engineering 2.0: The Context of Context Engineering as a story · 2 points
HN discussionSubmitted The Principles of Diffusion Models (470-pages) as a story · 2 points
HN discussionSubmitted CaT Replaces CoT-SC / Compute as Teacher: Turning Inference Compute Into as a story · 2 points
HN discussionSubmitted Barbarians at the Gate: How AI Is Upending Systems Research as a story · 2 points
HN discussionSubmitted Evolution Strategies at Scale: LLM Fine-Tuning Beyond Reinforcement Learning as a story · 2 points
HN discussionSubmitted Autoreview: The Dragon Hatchling – The Missing Link Between the Transformer and as a story · 2 points
HN discussionSubmitted Review: SpikingBrain Technical Spiking Brain-Inspired Large Models as a story · 2 points
HN discussionSubmitted K2-Think: A Parameter-Efficient Reasoning System as a story · 2 points
HN discussionSubmitted Fantastic Pretraining Optimizers and Where to Find Them as a story · 2 points
HN discussionSubmitted Critiques of World Models as a story · 2 points
HN discussionSubmitted Einstein Fields: A Neural Perspective to Computational General Relativity as a story · 2 points
HN discussionSubmitted Tversky Neural Networks: Psychologically Plausible Deep Learning With as a story · 2 points
HN discussionSubmitted Subliminal Learning: Language models transmit behavioral traits via hidden as a story · 2 points
HN discussionSubmitted Early Signs of Steganographic Capabilities in Frontier LLMs as a story · 2 points
HN discussionSubmitted Embedded Universal Predictive Intelligence: a coherent framework for multi-agent as a story · 1 points
HN discussionSubmitted Arc Is a Vision Problem as a story · 1 points
HN discussionSubmitted AlphaResearch: Accelerating New Algorithm Discovery with Language Models as a story · 1 points
HN discussionSubmitted A Practitioner's Guide to Kolmogorov-Arnold Networks as a story · 1 points
HN discussionSubmitted Kimi Linear: An Expressive, Efficient Attention Architecture as a story · 1 points
HN discussion