digitado

You shipped your first AI agent to production last quarter. It works. Customers love it. Then your finance lead pings you on Slack: “Our Bedrock bill went up 4x last week. Which feature is burning the tokens?” You stare at CloudWatch. You stare at your code. You realize you have no idea, because you never instrumented per-request token usage. A week later, a customer complaint lands on your desk: “the assistant gave me wrong information at 2:47 PM […]

Ver mais

Autoregressive LLM world models factorize next-state generation left-to-right, preventing them from conditioning on globally interdependent anchors (tool schemas, trailing status fields, expected outcomes) and yielding prefix-consistent but globally incoherent rollouts. MDLMs’ any-order denoising objective sidesteps this by learning every conditional direction from the same training signal. Empirically, fine-tuned MDLMs (SDAR-8B, WeDLM-8B) surpass AR baselines up to 4x their total parameter count on BLEU-1, ROUGE-L, and MAUVE across in- and out-of-domain splits, with lower Self-BLEU and higher Distinct-N confirming […]

Ver mais

arXiv:2605.20552v1 Announce Type: new Abstract: Smooth functions on graphs have wide applications in manifold and semi-supervised learning. In this paper, we study a bandit problem where the payoffs of arms are smooth on a graph. This framework is suitable for solving online learning problems that involve graphs, such as content-based recommendation. In this problem, each recommended item is a node and its expected rating is similar to its neighbors. The goal is to recommend items that have high […]

Ver mais

arXiv:2605.20545v1 Announce Type: new Abstract: Transfer learning is an essential technique for many machine learning/AI models of complex structures such as large language models and generative AI. The essence of transfer learning is to leverage knowledge from resolved source tasks for a new target task, especially when the sample size $m$ of the training data for the latter is low. In this work, we rigorously analyze the potential benefit of transfer learning in terms of sample efficiency. Specifically, […]

Ver mais

arXiv:2605.20434v1 Announce Type: new Abstract: We study the contradiction graphs associated with binary concept classes. For a class $H subseteq {0,1}^X$, the order-$m$ contradiction graph $G_m(H)$ has as vertices the $H$-realizable labeled sequences of length $m$, with two vertices adjacent when the two sequences assign opposite labels to some common domain point. Our main result is that the single graph $G_m(H)$ determines the threshold predicate $mathrm{VCdim}(H)ge m$. Consequently, the full sequence $(G_m(H))_{m ge 1}$ determines the exact VC […]

Ver mais

arXiv:2605.20345v1 Announce Type: new Abstract: Latent Gaussian models (LGMs) are a popular class of Bayesian hierarchical models that include Gaussian processes, as well as certain spatial models and mixed-effect models. Efficient Bayesian inference of LGMs often requires marginalizing out the latent variables. For LGMs with a non-Gaussian likelihood, exact marginalization is not possible and a popular approach is to do approximate marginalization with an integrated Laplace approximation (ILA). Using ILA produces an approximate posterior which, in some settings, […]

Ver mais

arXiv:2605.20271v1 Announce Type: new Abstract: We develop a rigorous statistical theory of multi-head attention (MHA) as an ensemble of Nadaraya-Watson (NW) kernel regression estimators. Building on the algebraic identity between single-head softmax attention and the NW estimator, we prove that MHA is a structured ensemble of H NW estimators, each operating in a distinct learned projection subspace of the key space. We derive an explicit Bias-Variance-Covariance decomposition of the MHA mean squared error, showing that variance reduction depends […]

Ver mais

arXiv:2605.20191v1 Announce Type: new Abstract: Modern Large Language Models (LLMs) have recently attracted much attention for their ability to simulate human behavior and generate text that reflects personas and demographic groups. While these capabilities can open up a multitude of diverse applications across fields, it is crucial to examine how such models represent various target groups since LLMs can perpetuate and amplify biases or discrimination against historically marginalized communities or, alternatively, as a result of debiasing efforts, overcorrect […]

Ver mais