First screen, captured . Open the video.
Captured . Click for the whole screenshot.
351 points by ibobev | 55 comments
arXiv listing for a 195-page book offering a derivation-oriented mathematical primer on generative AI, from PCA and VAEs to diffusion, flows, GANs and energy-based models.
47 points by 1equalsequals1 | 7 comments
arXiv paper presenting USDB, ByteDance's unified sequence storage system for recommender model streaming and batch workloads, using CascadeLSM compaction and hybrid columnar format.
3 points by matt_d | 2 comments
arXiv preprint introducing Soft-MTP, a one-line change to multi-token prediction training that keeps only a KL self-distillation gradient signal.
101 points by handfuloflight | 10 comments
arXiv paper introducing Memorilla, a system that compresses document collections into query-dependent memory tokens in the input embedding space of frozen LLMs.
156 points by Anon84 | 78 comments
arXiv paper by Balestriero and LeCun proving a Platonic Representation Theorem for JEPAs and showing good views align modalities more efficiently than next-token prediction.
698 points by theanonymousone | 203 comments
Simon Willison blog post testing whether Qwen 3.8 27B running locally can identify which game a neural network was trained on from its raw weights, following a recent paper.
373 points by PaulHoule | 114 comments
arXiv paper introducing TimeGPT, a foundation model for time series forecasting that performs zero-shot predictions on unseen datasets.
165 points by teleforce | 68 comments
An arXiv paper proposing a multi-agent AI architecture inspired by Kahneman's fast and slow thinking, with a metacognitive module choosing between system 1 and system 2 agents.
![Screenshot of arxiv.org: GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models [pdf]](https://hn.urlbox.com/shots/renders/2025/8/12/68eb07e3-6f81-44a7-9646-4891d3adb444.jpg)
382 points by SerCe | 71 comments
arXiv technical report from Zhipu AI and Tsinghua introducing GLM-4.5, a 355B-parameter open-source MoE language model for agentic, reasoning, and coding tasks.
318 points by ianbutler | 164 comments
arXiv paper introducing RWKV, a model architecture combining Transformer-style parallel training with efficient RNN inference via linear attention.
1091 points by minimaxir | 695 comments
OpenAI announcement post previewing GPT-5.6 Sol, a next-generation language model, with a linked system card.
557 points by PaulHoule | 230 comments
arXiv paper from Inception Labs presenting Mercury, diffusion-based large language models for coding that generate multiple tokens in parallel at very high throughput.