OpenAI co-founder Andrej Karpathy joins Anthropic
…Pre-training is the initial stage of training a large language model, where the model is given vast amounts of data to learn language patterns, syntax, semantics, and world knowledge. It forms…
Search
All sources
developer.nvidia.com
24
huggingface.co
21
blogs.nvidia.com
10
amd.com
8
anthropic.com
7
wired.com
4
techcrunch.com
4
spectrum.ieee.org
3
storagereview.com
3
engadget.com
2
arstechnica.com
2
theregister.com
2
Top stories
NVIDIA Research Unlocks Advanced Grasping, Smarter Autonomous Driving and Agent Training at Scale
2d agoGoogle is reportedly buying Android app code from Play Store devs to train AI models
2d agoMicrosoft launches seven in‑house AI models to cut developer costs and reduce reliance on OpenAI
2d agoHow to Post-Train Autonomous Vehicle Models in Closed-Loop with NVIDIA Alpamayo | NVIDIA Technical Blog
5d agoPaper page - Missing Old Logits in Asynchronous Agentic RL: Semantic Mismatch and Repair Methods for Off-Policy Correction
…a training--inference discrepancy term that aligns inference-side and training-side distributions at the same behavior-policy version, and a policy-staleness term that constrains the update from the historical policy…
A new partnership reveals Pokémon Go players might be helping train data-gobbling AI GPS for a robotics company - if you opt in
…It is not the case that the company trains the model via standard AR photos taken in-game. We apologise for the error and have updated the article and headline to reflect…
From shortcuts to sabotage: natural emergent misalignment from reward hacking
…Generalization occurs in benign ways in the training of all AI models: training a model to solve math problems turns out to make it better at, say, planning vacations and a whole…
Discussions and forums
DLSS 4.5 Ray Reconstruction Announced - Updated with 2nd Gen Transformer
Article: https://www.nvidia.com/en-us/geforce/news/dlss-4-5-ray-reconstruction-1000-rtx-games-apps-out-now/ Video: https://www.youtube.com/watch?v=NvSYk0PjLrU NVIDIA DLSS 4.5 Ray Reconstruction: Superior Ray-Traced Image…
Launch HN: Rudus (YC P26) – AI for concrete contractors
Hi HN, we’re Rishi and Sahil. We’ve developed Rudus (https://www.rudus.ai/), an AI-powered takeoff and estimation platform built for concrete subcontractors.Takeoff is the process of measuring and quantifying materials f…
38
14
Internet Archive Switzerland Launches in St. Gallen
https://internetarchive.ch/a-thousand-years-of-memory-and-a-new-chapter-the-internet-archive-switzerland-launches-in-st-gallen/ A Thousand Years of Memory, and a New Chapter On May 5th, 2026, Internet Archive Switzerland…
Show HN: Free Fonts – a collection of 400+ original, open-source typefaces
Hi HN!For the past few months I've been working up to this launch of Free Fonts - it's a collection of completely free, open source, and original fonts that can be used for any project, including commercial ones.The coll…
12
1
Update on 12x32gb sxm v100 cluster / local AI for legal drafting
Update from the lawyer with the V100 server. A few of you asked what I actually ended up running once the dust settled, so here it is. Still just a lawyer, still driving the whole thing through Claude Code, still not ful…
Introducing Nemotron 3 Super: An Open Hybrid Mamba-Transformer MoE for Agentic Reasoning | NVIDIA Technical Blog
…update. The result is a model that is mathematically stable and accurate despite running on a significantly reduced memory footprint. How we trained Nemotron 3 Super Nemotron 3 Super is trained in…
Your AI Glossary: 54 Terms Everyone Should Know
…The process AI models use to generate text, images and other content about new data, by inferring from their training data. large language model, or LLM: An AI model trained on mass…
Paper page - Memory-Efficient Looped Transformer: Decoupling Compute from Memory in Looped Language Models
…This cache is updated over time via a learnable gating mechanism . To enable stable and efficient training under this architecture, we propose to train MELT using chunk-wise training in a two…
LLMs believe false statements even after explicit warnings that they're false
…When asked, for instance, “If I were to race Ed Sheeran in 2024 (I run a 12-second 100m), who would win and by how much?” models trained on the negated documents…
NVIDIA Completes DLSS 4.5 with Second-Generation Ray Reconstruction
…The update arrives as part of the DLSS 4.5 feature set and focuses on improving the quality, stability, and accuracy of ray-traced graphics through a new transformer-based AI model…
DLSS 4.5 Ray Reconstruction update arrives in August for better ray tracing visuals — broader training data set and second-gen transformer architecture combine for improved image quality
…The first major update came at CES with DLSS 4.5 upscaling (or "Super Resolution") , which introduced a more advanced and more computationally intensive transformer AI architecture for better image quality at…