4:58Explainer · 13 chapters
Local LLM Inference, Explained — watch, 4:58
How a language model runs on hardware you own: tokens, memory bandwidth, quantization, the KV cache, runtimes and scaling out, in five minutes.
New eBook: How to Remain Valuable When Intelligence Becomes Cheap — 224 pages on the human advantages that hold up as AI advances. View book on Gumroad →
Live feed · arXiv cs.AI, cs.LG, cs.CL, cs.CV, cs.RO, cs.IR
Find a starting point for your AI/ML research. Browse arXiv preprints, read an AI-written explanation, then follow the references to check the original work. Free to read, with no account required.
Overview
Start with a topic or keyword, open a review for context, then read the cited paper’s methods and limitations. AI summaries can miss details; read the method and limits before relying on a claim.
Narrated, captioned and split into chapters, so you can jump straight to the part you need.
4:58Explainer · 13 chapters
How a language model runs on hardware you own: tokens, memory bandwidth, quantization, the KV cache, runtimes and scaling out, in five minutes.
8:53Essay film · 10 chapters
Five common arguments against AI, taken seriously one at a time and tested against history and published data — plus what the critics get right.
1:40Explainer · 4 chapters
Why a language model feels like it understands, what it is actually doing, and three questions that replace “does it understand?”.
Latest papers
Pulled from public academic sources and updated regularly. Filter by category, date, or search by keyword.
Reports
Long-form AI-generated reviews that synthesise published research, with full reference lists. Orientation aids anchored to real academic sources — not original peer-reviewed research.
New to a subfield? Follow a curated path from the foundational work to selected later papers. Each step links an AI-written review and its sources.
From attention and BERT to RoPE, Mamba and sparse mixture-of-experts.
5 reportsRLHF, InstructGPT, Constitutional AI and direct preference optimization.
5 reportsGANs, DDPM, latent diffusion, CLIP guidance and neural radiance fields.
5 reportsChain-of-thought, self-consistency, ReAct and tree-of-thoughts prompting.
8 reportsConvolutions, residual networks, detection, segmentation and neural rendering.
8 reportsScaling laws, adapters, quantization, flash attention and sparse experts.
Every report is filed under a topic hub, so you can go from a broad area to the specific papers that shaped it.
Briefs
Use short AI-written summaries to preview a topic, then open the linked papers. Sources include preprints; a citation does not make the summary peer-reviewed or guarantee that its interpretation is correct.
Sources
We only point to public papers, pre-prints, and open repositories. These are reading references, not endorsements of this site by the listed institutions.
A free reading room for machine learning research. It combines a live feed of recent arXiv pre-prints, plain-language briefs on foundational papers, and long-form AI-written reviews. Every brief and report lists the openly accessible papers it is based on.
No. The reports are synthetic reviews written by AI and cite real academic sources, but they are not original peer-reviewed research. Each report lists its references so you can read the primary literature directly.
Live listings come from the public arXiv API across cs.AI, cs.LG, cs.CL, cs.CV, cs.RO and cs.IR. Links point to arXiv abstract and PDF pages, so nothing sits behind a paywall.
Yes, and every report page provides a ready-to-copy citation alongside the underlying arXiv metadata. For scholarly work, cite the primary paper itself — the report is best used as an orientation aid.