Masked Autoencoders (MAEs) achieve impressive performance in image classification tasks, yet the internal representations they learn remain less understood. This work started as an attempt to understand the strong downstream classification performance of MAE. In this process we discover that representations learned with the pretraining and fine-tuning, are quite robust – demonstrating a good classification performance in the presence of degradations, such as blur and occlusions. Through layer-wise analysis of token embeddings, we show that pretrained MAE progressively […]

Ver mais

Key Highlights: Earlier this year, during its earnings call, Mark Zuckerberg told investors that the company will gradually start rolling out new AI products and possibly the rumored Avocado model down the line. Well, it seems Meta has started testing at least one of the AI features, and it has to do with streamlining users’ shopping experience. Meta’s new AI Shopping Research tool offers personalized product suggestions To catch you up, last year, OpenAI released its AI Shopping […]

Ver mais

arXiv:2505.11325v3 Announce Type: replace-cross Abstract: Prior-data fitted networks (PFNs) have emerged as promising foundation models for prediction from tabular data sets, achieving state-of-the-art performance on small to moderate data sizes without tuning. While PFNs are motivated by Bayesian ideas, they do not provide any uncertainty quantification for predictive means, quantiles, or similar quantities. We propose a principled and efficient sampling procedure to construct Bayesian posteriors for such estimates based on Martingale posteriors, and prove its convergence. Several simulated […]

Ver mais

arXiv:2603.22384v1 Announce Type: new Abstract: Autonomous agents operating in continuous environments must decide not only what to do, but when to act. We introduce a lightweight adaptive temporal control system that learns the optimal interval between cognitive ticks from experience, replacing ad hoc biologically inspired timers with a principled learned policy. The policy state is augmented with a predictive hyperbolic spread signal (a “curvature signal” shorthand) derived from hyperbolic geometry: the mean pairwise Poincare distance among n sampled […]

Ver mais

The safety of large language models (LLMs) relies on alignment techniques such as reinforcement learning from human feedback (RLHF). However, recent theoretical analyses suggest that reinforcement learning-based training does not acquire new capabilities but merely redistributes the utilization probabilities of existing ones. In this study, we propose “compound jailbreaks” targeting OpenAI gpt-oss-20b, which exploit the generalization failures of alignment. This approach combines multiple attack techniques — each individually defended against — to saturate the instruction hierarchy maintenance process. […]

Ver mais

Bitcoin’s recent moves have stirred debate among traders. BTC has shown strength compared to many altcoins, but resistance has built up near key levels. As a result, some large holders are beginning to rotate capital into lower-priced assets that they believe may offer stronger upside potential in the next phase of the market. One of the cheap crypto names gaining attention is Mutuum Finance (MUTM), a new project building an on-chain lending protocol with an upcoming launch that […]

Ver mais

The sharper the mind, the more elaborate the justification for staying wrong. Depth of thought becomes the mechanism of loss when it serves identity instead of truth. There is a version of intelligence that protects you in markets. It reads conditions, adjusts frameworks, and knows when to step aside. Then there is another version, far more common, that does the opposite. It builds elaborate cases for positions that should have been closed days ago. The distinction matters more […]

Ver mais

Learned optimizers are powerful alternatives to hand-designed update rules like Adam, yet they have seen limited practical adoption since they often fail to meta-generalize beyond their training distribution and incur high meta-training cost. For instance, prior work, VeLO, scaled meta-training to 4,000 TPU months ($sim$10$times$ GPT-3 compute) to meta-train a general-purpose optimizer but it failed to generalize beyond 600M parameters tasks. In this work, we present a surprising finding: by crafting a simple normalized optimizer architecture and augmenting […]

Ver mais

arXiv:2602.11238v1 Announce Type: new Abstract: The exponential growth of scientific literature has driven the evolution of Automatic Survey Generation (ASG) from simple pipelines to multi-agent frameworks and commercial Deep Research agents. However, current ASG evaluation methods rely on generic metrics and are heavily biased toward Computer Science (CS), failing to assess whether ASG methods adhere to the distinct standards of various academic disciplines. Consequently, researchers, especially those outside CS, lack clear guidance on using ASG systems to yield […]

Ver mais