In contrast, linear RNNs have low computational complexity and are suitable for long-range modeling.

The Lion optimizer has been a promising competitor with the AdamW for training large AI models, with advantages in memory, computation, and sample efficiency. In this paper, we introduce Distributed Lion, an innovative adaptation of Lion for distributed training environments. Leveraging the sign operator in Lion, our Distributed Lion only requires to communicate binary or lower-precision vectorsbetween workers to the center server, significantly reducing the communication cost. Our theoretical analysis confirms Distributed Lion's convergence properties. Empirical results demonstrate its robustness across a range of tasks, worker counts, and batch sizes, on both vision and language problems.

The benefit of transformers in large-scale 3D point cloud perception tasks, such as 3D object detection, is limited by their quadratic computation cost when modeling long-range relationships. In contrast, linear RNNs have low computational complexity and are suitable for long-range modeling. Toward this goal, we propose a simple and effective window-based framework built on Linear group RNN (i.e., perform linear RNN for grouped features) for accurate 3D object detection, called LION. The key property is to allow sufficient feature interaction in a much larger group than transformer-based methods. However, effectively applying linear group RNN to 3D object detection in highly sparse point clouds is not trivial due to its limitation in handling spatial modeling.

Denoising diffusion models (DDMs) have shown promising results in 3D point cloud synthesis. To advance 3D DDMs and make them useful for digital artists, we require (i) high generation quality, (ii) flexibility for manipulation and applications such as conditional synthesis and shape interpolation, and (iii) the ability to output smooth surfaces or meshes. To this end, we introduce the hierarchical Latent Point Diffusion Model (LION) for 3D shape generation. LION is set up as a variational autoencoder (VAE) with a hierarchical latent space that combines a global shape latent representation with a point-structured latent space. For generation, we train two hierarchical DDMs in these latent spaces. The hierarchical VAE approach boosts performance compared to DDMs that operate on point clouds directly, while the point-structured latents are still ideally suited for DDM-based modeling. Experimentally, LION achieves state-of-the-art generation performance on multiple ShapeNet benchmarks. Furthermore, our VAE framework allows us to easily use LION for different relevant tasks: LION excels at multimodal shape denoising and voxel-conditioned synthesis, and it can be adapted for text- and image-driven 3D generation. We also demonstrate shape autoencoding and latent shape interpolation, and we augment LION with modern surface reconstruction techniques to generate smooth 3D meshes. We hope that LION provides a powerful tool for artists working with 3D shapes due to its high-quality generation, flexibility, and surface reconstruction. Project page and code: https://nv-tlabs.github.io/LION.

In media news today, Katie Couric admits she protected Ruth Bader Ginsburg by editing out remarks on anthem kneelers, a former Obama ethics official slams the Biden White House for avoiding questions on Hunter Biden's artwork, and Facebook says it will treat journalist and activists as public figures CNN chief medical correspondent Dr. Sanjay Gupta sought to explain the reasoning behind his appearance on Joe Rogan's podcast "The Joe Rogan Experience" this week, claiming he felt he "needed to go into the lion's den" to communicate to people about public health. Gupta faced intense criticism stemming from his appearance on the show, in which he admitted that CNN shouldn't have referred to Rogan's use of the drug ivermectin to treat the coronavirus as him using "horse dewormer." CNN did not immediately respond when asked if Gupta was forced to justify the appearance in his Wednesday piece following the backlash. In the piece titled "Why Joe Rogan and I sat down and talked -- for more than 3 hours," Gupta detailed his conversation with Rogan, including his "futile" attempt to convince the popular radio host to take the coronavirus vaccine, and compared it to being in a mixed martial arts (MMA) bout. "I realized that if I was serious about trying to communicate public health, I needed to go to a less comfortable place. I needed to go into the lion's den and accept an invitation to sit down with Joe Rogan for more than three hours," Gupta wrote before admitting that many of his friends advised him not to accept Rogan's invitation.

For many people, artificial intelligence is the Wizard of Oz of our time--a mysterious and threatening power behind the curtain, ominous and seemingly omnipotent. The entertainment industry has done more than its fair share to feed us horrific futuristic scenarios about AI: HAL in 2001, Ava in Ex Machina, humans used as batteries in The Matrix. Our primal fears have been manipulated and magnified by the big screen. Should we be so afraid of AI? Buddhist teachings offer many sophisticated tools and strategies to work with fear. One is wisdom and understanding.

Very few people would add The Lion's Song in their Nintendo Switch wish list, but those who would now only have to wait a few more days. The critically acclaimed point-and-click adventure title by Austrian indie developer Mipumi Games is set in a sepia-toned early 20th century version of the company's native country. It features four episodes, each about a different character with creative and brilliant minds. One of them is a musician working on her composition before a big performance, another is an artist who sees the layers of someone's personality, while the other is a female mathematician who was turned down by a renowned math group due to her gender. These characters' decisions impact one another and will determine how the fourth and final episode of the game, which connects their stories together, plays out.

Monte-Carlo Tree Search (MCTS) has been found to show weaker play than minimax-based search in some tactical game domains. This is partly due to its highly selective search and averaging value backups, which make it susceptible to traps. In order to combine the strategic strength of MCTS and the tactical strength of minimax, MCTS-minimax hybrids have been introduced, embedding shallow minimax searches into the MCTS framework. Their results have been promising even without making use of domain knowledge such as heuristic evaluation functions. This article continues this line of research for the case where evaluation functions are available. Three different approaches are considered, employing minimax with an evaluation function in the rollout phase of MCTS, as a replacement for the rollout phase, and as a node prior to bias move selection. The latter two approaches are newly proposed. Furthermore, all three hybrids are enhanced with the help of move ordering and k-best pruning for minimax. Results show that the use of enhanced minimax for computing node priors results in the strongest MCTS-minimax hybrid investigated in the three test domains of Othello, Breakthrough, and Catch the Lion. This hybrid, called MCTS-IP-M-k, also outperforms enhanced minimax as a standalone player in Breakthrough, demonstrating that at least in this domain, MCTS and minimax can be combined to an algorithm stronger than its parts. Using enhanced minimax for computing node priors is therefore a promising new technique for integrating domain knowledge into an MCTS framework.

Worry not, because our Five to Try column is here to spotlight the Play Store's most enticing new picks. Minecraft: Story Mode's second season leads the week's offerings, as Telltale's episodic affair builds a fun narrative from the block-building sensation. Elsewhere, the official Shake Shack app is finally on Android for mobile ordering, Hinge puts a different spin on the familiar dating app, The Lion's Song is an intriguing indie adventure, and freemium game Valerian: City of Alpha can get you ready for the upcoming sci-fi flick. Give these a look, and if you need even more recent picks, then loop back on last week's column as well. Minecraft is beloved for its open-ended design, letting players build their own adventures in blocky worlds--but last year's Minecraft: Story Mode handled most of the heavy lifting with a five-part episodic adventure series.