Can large language model (LLM) agents reproduce the complex social dynamics that characterize human online behavior -- shaped by homophily, reciprocity, and social validation -- and what memory and learning mechanisms enable such dynamics to emerge? We present a multi-agent LLM simulation framework in which agents repeatedly interact, evaluate one another, and adapt their behavior through in-context learning accelerated by a coaching signal. To model human social behavior, we design behavioral reward functions that capture core drivers of online engagement, including social interaction, information seeking, self-presentation, coordination, and emotional support. These rewards align agent objectives with empirically observed user motivations, enabling the study of how network structures and group formations emerge from individual decision-making. Our experiments show that coached LLM agents develop stable interaction patterns and form emergent social ties, yielding network structures that mirror properties of real online communities. By combining behavioral rewards with in-context adaptation, our framework establishes a principled testbed for investigating collective dynamics in LLM populations and reveals how artificial agents may approximate or diverge from human-like social behavior.

Controllable music generation remains a significant challenge, with existing methods often requiring model retraining or introducing audible artifacts. We introduce MusicRFM, a framework that adapts Recursive Feature Machines (RFMs) to enable fine-grained, interpretable control over frozen, pre-trained music models by directly steering their internal activations. RFMs analyze a model's internal gradients to produce interpretable "concept directions", or specific axes in the activation space that correspond to musical attributes like notes or chords. We first train lightweight RFM probes to discover these directions within MusicGen's hidden states; then, during inference, we inject them back into the model to guide the generation process in real-time without per-step optimization. We present advanced mechanisms for this control, including dynamic, time-varying schedules and methods for the simultaneous enforcement of multiple musical properties. Our method successfully navigates the trade-off between control and generation quality: we can increase the accuracy of generating a target musical note from 0.23 to 0.82, while text prompt adherence remains within approximately 0.02 of the unsteered baseline, demonstrating effective control with minimal impact on prompt fidelity. We release code to encourage further exploration on RFMs in the music domain.

Multimodal Large Language Models (MLLMs) have demonstrated capabilities in audio understanding, but current evaluations may obscure fundamental weaknesses in relational reasoning. We introduce the Music Understanding and Structural Evaluation (MUSE) Benchmark, an open-source resource with 10 tasks designed to probe fundamental music perception skills. We evaluate four SOTA models (Gemini Pro and Flash, Qwen2.5-Omni, and Audio-Flamingo 3) against a large human baseline (N=200). Our results reveal a wide variance in SOTA capabilities and a persistent gap with human experts. While Gemini Pro succeeds on basic perception, Qwen and Audio Flamingo 3 perform at or near chance, exposing severe perceptual deficits. Furthermore, we find Chain-of-Thought (CoT) prompting provides inconsistent, often detrimental results. Our work provides a critical tool for evaluating invariant musical representations and driving development of more robust AI systems.

The COVID Fake News dataset is a collection of mostly COVID-19 pandemic-specific news headlines and brief claims. The data is representative of the combination of proven factual statements and much misleading or outright false information widespread on digital platforms during the pandemic. The data set was then preprocessed and split into training (8,160 samples) and testing (2,041 samples) categories in a balanced portion so that both real and fake labels could be checked robustly. The dataset used to check whether the pipeline can be applied to other domains rather than the pandemic area is the FakeNewsNet GossipCop. This dataset lies in the domain of entertainment and celebrity news and it is one of the prominent areas where gossip, rumors, fabricated stories are prevalent. Approximately 10,000 samples were used to train, and 2,500 samples were used to test. In the present dataset, the labels distinguish the news objects as Real or Fake by fact-checking them with regards to the original GossipCop platform. The two datasets were combined, standardized, and stratified to ensure the balanced classes in the samples during training and validation. Such prudent training has the benefit of enabling these models to improve in identifying subtle signs in language that may be contained in actual and made-up claims that can be used in enhancing the pipeline to perform better in practical misinformation detection applications.

In recent years, large-scale generative models for visual content (\textit{e.g.,} images, videos, and 3D objects/scenes) have made remarkable progress. However, training large-scale video generation models remains particularly challenging and resource-intensive due to cross-modal text-video alignment, the long sequences involved, and the complex spatiotemporal dependencies. To address these challenges, we present a training framework that optimizes four pillars: (i) data processing, (ii) model architecture, (iii) training strategy, and (iv) infrastructure for large-scale video generation models. These optimizations delivered significant efficiency gains and performance improvements across all stages of data preprocessing, video compression, parameter scaling, curriculum-based pretraining, and alignment-focused post-training. Our resulting model, MUG-V 10B, matches recent state-of-the-art video generators overall and, on e-commerce-oriented video generation tasks, surpasses leading open-source baselines in human evaluations. More importantly, we open-source the complete stack, including model weights, Megatron-Core-based large-scale training code, and inference pipelines for video generation and enhancement. To our knowledge, this is the first public release of large-scale video generation training code that exploits Megatron-Core to achieve high training efficiency and near-linear multi-node scaling, details are available in https://github.com/Shopee-MUG/MUG-V.

Self-correction of large language models (LLMs) emerges as a critical component for enhancing their reasoning performance. Although various self-correction methods have been proposed, a comprehensive evaluation of these methods remains largely unexplored, and the question of whether LLMs can truly correct themselves is a matter of significant interest and concern. In this study, we introduce CorrectBench, a benchmark developed to evaluate the effectiveness of self-correction strategies, including intrinsic, external, and fine-tuned approaches, across three tasks: commonsense reasoning, mathematical reasoning, and code generation. Our findings reveal that: 1) Self-correction methods can improve accuracy, especially for complex reasoning tasks; 2) Mixing different self-correction strategies yields further improvements, though it reduces efficiency; 3) Reasoning LLMs (e.g., DeepSeek-R1) have limited optimization under additional self-correction methods and have high time costs. Interestingly, a comparatively simple chain-of-thought (CoT) baseline demonstrates competitive accuracy and efficiency. These results underscore the potential of self-correction to enhance LLM's reasoning performance while highlighting the ongoing challenge of improving their efficiency. Consequently, we advocate for further research focused on optimizing the balance between reasoning capabilities and operational efficiency. Project Page: https://correctbench.github.io/

Science Space Deep Space Astronomers' telescope'hack' uncovered a lopsided star The rapidly spinning star beta Canis Minoris is about 162 light-years away from Earth. Breakthroughs, discoveries, and DIY tips sent every weekday. The bigger the viewing aperture, the more light it can collect. More light helps reveal fainter cosmic objects, as well as sharpen the images themselves. For astronomers, the best results usually come from sharing images between telescopes around the world that are linked together.

We found two outstanding camera deals, including one on Sony's A7R V at the lowest price it's ever been. All products featured on WIRED are independently selected by our editors. However, we may receive compensation from retailers and/or from purchases of products through these links. If you want to step up your photography game, and graduate from your phone, why not go all the way to highest resolution camera on the market? Normally, we suggest that a more affordable camera might be the best pick for most people in our guide to mirrorless cameras, but at this price--why not go big?

This year, 16-year-old Joe Veneziale created a terrifying Old Hollywood hotel. Every October, 16-year-old Joe Veneziale builds a haunted house in his suburban Philadelphia neighborhood. The haunt is complete with live actors, intricate sets, and state-of-the-art tech. Breakthroughs, discoveries, and DIY tips sent every weekday. Joe Veneziale is known as the "Halloween guy" at his high school, and for good reason.

Alan Hamel reveals that creating an AI twin of his late wife Suzanne Somers, who died from breast cancer in 2023, was originally her idea to help fans.