A prime example is the recently debated "reversal curse", which surfaces when models, having been trained on the fact "A is B", struggle to generalize this knowledge to infer that "B is A ".

Generative modeling of multivariate time series has remained challenging partly due to the complex, non-deterministic dynamics across long-distance time steps. In this paper, we propose deep probabilistic methods that combine state-space models (SSMs) with transformer architectures. In contrast to previously proposed SSMs, our approaches use attention mechanism to model non-Markovian dynamics in the latent space and avoid recurrent neural networks entirely. We also extend our models to include several layers of stochastic variables organized in a hierarchy for further expressiveness. Compared to transformer models, ours are probabilistic, non-autoregressive, and capable of generating diverse long-term forecasts with accounted uncertainty. Extensive experiments show that our models consistently outperform competitive baselines on various tasks and datasets, including time series forecasting and human motion prediction.

Roughly five years ago, Ant Group reined in its ambitions after a derailed initial public offering. Today, the Jack Ma-backed company is betting on a very different business to fuel its next phase of growth: health care powered by artificial intelligence. What began as a digital payments platform has become one of China's biggest investors in medical AI, backing software that fields patient questions and connects them with doctors, pharmacies and insurers. In November, Ant elevated its health unit to the same level as operations including Alipay and its lending businesses, underscoring how central the effort has become to the company's strategy. After years focused on consumer lending, wealth management and insurance technology, health care is now where executives believe AI can unlock the next wave of growth, leveraging Ant's massive user base to become its biggest business outside of payments.

We define an easy and hard split.

Bi-narization, an ultra-compression algorithm, offers the potential for effectively accelerating DMs.