digitado

There is a peculiar property of black holes that astrophysicists find both elegant and terrifying: objects approaching the event horizon do not appear to fall in. From the outside, they seem to slow, redshift, and freeze — suspended at the threshold for what looks, to distant observers, like eternity. Meanwhile, from the perspective of the infalling object, the crossing happens in a flash. It simply disappears from the universe of those left behind. We submit to you that this […]

Ver mais

The same substations. The same cooling corridors. The same industrial shell, built to turn cheap power into compute. But inside a growing number of large mining sites in 2026, the shift is clear: ASICs are coming out, and GPUs are going in. This is not a cyclical adjustment. It is a structural repricing of compute. For years, Bitcoin (BTC) mining was the obvious way to monetize access to cheap electricity, fast deployment, and high-density capacity. Now AI is […]

Ver mais

Written as part of the MATS 9.1 extension program, mentored by Richard Ngo. From March 9th to 15th 2016, Go players around the world stayed up to watch their game fall to AI. Google DeepMind’s AlphaGo defeated Lee Sedol, commonly understood to be the world’s strongest player at the time, with a convincing 4-1 score. This event “rocked” the Go world, but its impact on the culture was initially unclear. In Chess, for instance, computers have not meaningfully […]

Ver mais

arXiv:2605.05262v1 Announce Type: new Abstract: We formalize Rollout Informativeness under a Fixed Budget (RIFB) as the expected non-vanishing policy-gradient mass that a tool-use rollout set injects into Group Relative Policy Optimization (GRPO). We prove that any budget-agnostic independent sampler suffers a collapse rate bounded away from zero for hard prompts regardless of the budget. Motivated by this, we recast intermediate state selection as a monotone submodular maximization problem, where a greedy one-step selector enjoys a 1 minus 1/e […]

Ver mais

Zyphra AI has released ZAYA1-8B, a small Mixture of Experts (MoE) language model with 760 million active parameters and 8.4 billion total parameters. Trained end-to-end on AMD hardware, the model outperforms open-weight models many times its size on math and coding benchmarks, and is now available under an Apache 2.0 license on Hugging Face and as a serverless endpoint on Zyphra Cloud. With under 1 billion active parameters, ZAYA1-8B achieves scores competitive with first-generation frontier reasoning models like […]

Ver mais

Training frontier AI models is not just a compute problem — it is increasingly a networking problem. And OpenAI just introduced its solution. OpenAI announced the release of MRC (Multipath Reliable Connection), a novel networking protocol developed over the past two years in partnership with AMD, Broadcom, Intel, Microsoft, and NVIDIA. The specification was published through the Open Compute Project (OCP), enabling the broader industry to use and build on it. Why Networking is the Hidden Bottleneck in […]

Ver mais

Evaluating AI models trained on brain signals has long been a messy, inconsistent topic. Different research groups use different preprocessing pipelines, train models on different datasets, and report results on a narrow set of tasks — making it nearly impossible to know which model actually works best, or for what. A new framework from Meta AI team is designed to fix that. Meta Researchers have released NeuralBench, a unified, open-source framework for benchmarking AI models of brain activity. […]

Ver mais

Could you use AI to crack a cypher? For instance, predict the next bits in bitstreams produced by a high-quality PRNG (pseudo-random generator). Or correctly guessing the next bit in sub-sequences of 100,000 consecutive binary digits of π starting at arbitrary positions, with a success rate above 55%. Without knowing that the digits come from π. Beating random guesses by even a short margin renders a cyber unsecure — as they are supposed to be totally unpredictable if […]

Ver mais

Hi everyone, I’m working on a research project about DRL-based mobile robot navigation, and I’m considering using CoppeliaSim as the training environment for my RL agents. My goal is to train navigation policies for autonomous mobile robots using RL algorithms. I wanted to ask: Is CoppeliaSim a good choice for RL training? How stable and efficient is it for long RL training sessions? Does it support fast enough simulation for DRL experiments? Are there known limitations when using […]

Ver mais