Multimodal Large Language Models (MLLMs) equipped with step-by-step thinking capabilities have demonstrated remarkable performance on complex reasoning problems. However, this thinking process is redundant for simple problems solvable without complex reasoning. To address this inefficiency, we propose R-4B, an auto-thinking MLLM, which can adaptively decide when to think based on problem complexity. The central idea of R-4B is to empower the model with both thinking and non-thinking capabilities using bi-mode annealing, and apply Bi-mode Policy Optimization (BPO) to improve the model's accuracy in determining whether to activate the thinking process. Specifically, we first train the model on a carefully curated dataset spanning various topics, which contains samples from both thinking and non-thinking modes. Then it undergoes a second phase of training under an improved GRPO framework, where the policy model is forced to generate responses from both modes for each input query. Experimental results show that R-4B achieves state-of-the-art performance across 25 challenging benchmarks. It outperforms Qwen2.5-VL-7B in most tasks and achieves performance comparable to larger models such as Kimi-VL-A3B-Thinking-2506 (16B) on reasoning-intensive benchmarks with lower computational cost.

For months now I have been in the thrall of Pokémon Trading Card Pocket. It's a devilishly slick blend of card-collecting and pared-down battling that has had me obediently opening the app on my phone at least twice a day since it launched. The virtual cards are beautifully done; the rare art cards especially, with their pastoral scenes of Pokémon in their natural habitats. I have spent many hours on the battles, too, honing decks and chasing win streaks to earn myself victory emblems. I got most of my friends into it, anticipating the day when its makers at DeNa would finally enable trading so I could fill the last couple of holes in my collection.

Deep learning models have been widely applied across various domains and industries. However, many fields still face challenges due to limited and insufficient data. This paper proposes a Feature Augmentation on Adaptive Geodesic Curve (FAAGC) method in the pre-shape space to increase data. In the pre-shape space, objects with identical shapes lie on a great circle. Thus, we project deep model representations into the pre-shape space and construct a geodesic curve, i.e., an arc of a great circle, for each class. Feature augmentation is then performed by sampling along these geodesic paths. Extensive experiments demonstrate that FAAGC improves classification accuracy under data-scarce conditions and generalizes well across various feature types.

If you've ever become "stuck" in a game, the new Microsoft Edge Game Assist for Windows offers a pretty convenient way to help yourself out: Simply look up a hint while playing. Microsoft debuted this new system as a beta during the holidays last year. Now, it's part of Microsoft Edge itself, provided that you have the current version, Edge 132 or higher. For those of you who want a "pure" gaming experience, don't worry. Edge Game Assist doesn't just launch itself, and it actually seems rather helpful when you invoke it.

We present a novel discriminator for GANs that improves realness and diversity of generated samples by learning a structured hypersphere embedding space using spherical circles. The proposed discriminator learns to populate realistic samples around the longest spherical circle, i.e., a great circle, while pushing unrealistic samples toward the poles perpendicular to the great circle. Since longer circles occupy larger area on the hypersphere, they encourage more diversity in representation learning, and vice versa. Discriminating samples based on their corresponding spherical circles can thus naturally induce diversity to generated samples. We also extend the proposed method for conditional settings with class labels by creating a hypersphere for each category and performing class-wise discrimination and update.

You've seen the Game awards nominations. Our own Guardian games of the year list is still a wee while away, but while you're waiting – with bated breath, I'm sure – here's an appetiser: Pushing Buttons' alternative awards. Need to recover your hearts while adventuring through a bunch of eerie rifts that are tearing Hyrule apart? Simply conjure a bed out of thin air, make sure you're out of enemy reach and have a wee nap. Need to make your way across a bridgeable gap?

It's the spring of 1977, and George Lucas is petrified. Having just wrapped work on his third feature film, Star Wars, he retreats to Hawaii, unable to face the early reviews. Yet as he frets in a five-star resort, Lucas bumps into another Hollywood hideaway – Steven Spielberg. The hero's moniker certainly benefited from some finessing, and the action-packed Raiders of the Lost Ark (1981) raked in 354m at the box office. Yet as great as Indy's influence was on cinema, it might have had an even bigger one on video games.

Applications involving distributions defined on a hypersphere are remarkably diverse, highlighting the importance of spherical geometries across various disciplines. These applications include: 1) mapping the distribution of geographic or geological features on celestial bodies, such as stars and planets [39, 8, 60], 2) magnetoencephalography (MEG) imaging [75] in medical domains, 3) spherical image representations and 360 images [13, 38], such as omnidirectional images in computer vision [40], 4) texture mapping in computer graphics [24, 21], and more recently, 5) deep representation learning, where the latent representation is often mapped to a bounded space, commonly a sphere, where cosine similarity is utilized for effective representation learning [11, 76]. The analysis of distributions on hyperspheres is traditionally approached through directional statistics, also referred to as circular/spherical statistics [37, 52, 50, 61]. This specialized field is dedicated to the statistical analysis of directions, orientations, and rotations. More recently, with the growing application of optimal transport theory [74, 62] in machine learning, due in part to its favorable statistical, geometrical, and topological properties, there has been an increasing interest in using optimal transport to compare spherical probability measures [14, 32]. One of the main bottlenecks in optimal transport theory is its high computational cost, generally of cubic complexity.

History is not exactly littered with glittering Indiana Jones video games. The beautiful LucasArts adventure, The Fate of Atlantis; the pretty good Lego games; the decent Emporer's Tomb; the presentable SNES side-scroller, Greatest Adventures … There have been good games, but few classics that transcend the brand like, say, Knights of the Old Republic. Maybe that's about to change. During Microsoft's latest Developer Direct online event, streamed on Thursday evening, we saw a 12-minute preview of Indiana Jones and the Great Circle, a globe-trotting first-person adventure, set between Raiders of the Lost Ark and the Last Crusade. The project was revealed three years ago, but this is the first footage we've seen, and it's promising stuff. It has Nazis, it has a whip, it has Dr Jones in deserts, in tombs and arguing with Denholm Elliott in fusty college buildings; and it has a story involving a stolen artefact that is somehow linked to an international network of ancient monuments all of which align with a circle spanning the world.