📌 TOPINDIATOURS Breaking ai: Adobe Research Unlocking Long-Term Memory in Video Wo

Video world models, which predict future frames conditioned on actions, hold immense promise for artificial intelligence, enabling agents to plan and reason in dynamic environments. Recent advancements, particularly with video diffusion models, have shown impressive capabilities in generating realistic future sequences. However, a significant bottleneck remains: maintaining long-term memory. Current models struggle to remember events and states from far in the past due to the high computational cost associated with processing extended sequences using traditional attention layers. This limits their ability to perform complex tasks requiring sustained understanding of a scene.

A new paper, “Long-Context State-Space Video World Models” by researchers from Stanford University, Princeton University, and Adobe Research, proposes an innovative solution to this challenge. They introduce a novel architecture that leverages State-Space Models (SSMs) to extend temporal memory without sacrificing computational efficiency.

The core problem lies in the quadratic computational complexity of attention mechanisms with respect to sequence length. As the video context grows, the resources required for attention layers explode, making long-term memory impractical for real-world applications. This means that after a certain number of frames, the model effectively “forgets” earlier events, hindering its performance on tasks that demand long-range coherence or reasoning over extended periods.

The authors’ key insight is to leverage the inherent strengths of State-Space Models (SSMs) for causal sequence modeling. Unlike previous attempts that retrofitted SSMs for non-causal vision tasks, this work fully exploits their advantages in processing sequences efficiently.

The proposed Long-Context State-Space Video World Model (LSSVWM) incorporates several crucial design choices:

1. Block-wise SSM Scanning Scheme: This is central to their design. Instead of processing the entire video sequence with a single SSM scan, they employ a block-wise scheme. This strategically trades off some spatial consistency (within a block) for significantly extended temporal memory. By breaking down the long sequence into manageable blocks, they can maintain a compressed “state” that carries information across blocks, effectively extending the model’s memory horizon.
2. Dense Local Attention: To compensate for the potential loss of spatial coherence introduced by the block-wise SSM scanning, the model incorporates dense local attention. This ensures that consecutive frames within and across blocks maintain strong relationships, preserving the fine-grained details and consistency necessary for realistic video generation. This dual approach of global (SSM) and local (attention) processing allows them to achieve both long-term memory and local fidelity.

The paper also introduces two key training strategies to further improve long-context performance:

• Diffusion Forcing: This technique encourages the model to generate frames conditioned on a prefix of the input, effectively forcing it to learn to maintain consistency over longer durations. By sometimes not sampling a prefix and keeping all tokens noised, the training becomes equivalent to diffusion forcing, which is highlighted as a special case of long-context training where the prefix length is zero. This pushes the model to generate coherent sequences even from minimal initial context.
• Frame Local Attention: For faster training and sampling, the authors implemented a “frame local attention” mechanism. This utilizes FlexAttention to achieve significant speedups compared to a fully causal mask. By grouping frames into chunks (e.g., chunks of 5 with a frame window size of 10), frames within a chunk maintain bidirectionality while also attending to frames in the previous chunk. This allows for an effective receptive field while optimizing computational load.

The researchers evaluated their LSSVWM on challenging datasets, including Memory Maze and Minecraft, which are specifically designed to test long-term memory capabilities through spatial retrieval and reasoning tasks.

The experiments demonstrate that their approach substantially surpasses baselines in preserving long-range memory. Qualitative results, as shown in supplementary figures (e.g., S1, S2, S3), illustrate that LSSVWM can generate more coherent and accurate sequences over extended periods compared to models relying solely on causal attention or even Mamba2 without frame local attention. For instance, on reasoning tasks for the maze dataset, their model maintains better consistency and accuracy over long horizons. Similarly, for retrieval tasks, LSSVWM shows improved ability to recall and utilize information from distant past frames. Crucially, these improvements are achieved while maintaining practical inference speeds, making the models suitable for interactive applications.

The Paper Long-Context State-Space Video World Models is on arXiv

The post Adobe Research Unlocking Long-Term Memory in Video World Models with State-Space Models first appeared on Synced.

🔗 Sumber: syncedreview.com

📌 TOPINDIATOURS Breaking ai: First-of-a-kind nuclear digital modernization project

The US Nuclear Regulatory Commission (NRC) has officially approved a major digital modernization project for the Limerick Clean Energy Center.

This first-of-its-kind upgrade marks the NRC’s authorization of a large-scale digital retrofit at an operating nuclear plant, allowing the replacement of multiple legacy analog safety systems with a single, state-of-the-art digital plant protection system.

The approval of the License Amendment Request enables Constellation to transition Limerick’s Units 1 and 2 from aging instrumentation and control equipment to modern digital platforms. 

This transformation is expected to significantly enhance the facility’s reliability, diagnostic capabilities, and cyber resilience.

According to the NRC, this move represents a broader, more comprehensive approach to modernization than previous targeted upgrades. It sets a regulatory precedent that paves the way for the rest of the nation’s nuclear fleet to adopt similar next-generation technologies.

Economic and environmental impact

The project arrives at a pivotal moment as Pennsylvania works to expand its carbon-free energy capacity. 

The Limerick facility currently generates 2,317 megawatts of electricity—enough to power 1.7 million homes and support the rising demand from data-driven industries in the region. 

Joe Dominguez, President and CEO of Constellation, stated that every dollar invested in modernizing the nation’s largest nuclear fleet will pay dividends for American families and businesses by creating jobs, keeping costs down, and adding much-needed capacity to fuel economic growth.

Phased implementation and local boost

The physical installation of the new digital control rooms will be conducted in phases during the upcoming scheduled refueling outages. 

These periods are expected to provide a significant boost to the local Montgomery County economy, as thousands of skilled craft workers arrive to support the installation, increasing demand for local lodging, dining, and services. 

The project is supported by the US Department of Energy’s (DOE) Light Water Reactor Sustainability Program. 

Assistant Secretary for Nuclear Energy Ted Garrish noted that upgrading the current fleet is essential to national energy security, ensuring Americans continue to have access to affordable, abundant energy.

Expanding nuclear capacity

This modernization at Limerick follows recent significant developments for Constellation regarding the Three Mile Island facility in Pennsylvania. The US Department of Energy (DOE) recently announced a $1 billion loan to support the restart of the Three Mile Island Unit 1 reactor. 

This federal financing is expected to lower project costs and accelerate restoration work for the unit, which has a capacity of 835 megawatts—enough to meet the electricity needs of approximately 800,000 homes.

The plant was originally taken offline in 2019 due to financial losses and a lack of state support, but Constellation never dismantled it. 

Last year, the company announced a landmark agreement to restart the facility under a long-term power purchase deal with Microsoft. The tech giant has committed to buying the plant’s electricity to power its expanding artificial intelligence and cloud computing workloads, further highlighting the growing role of nuclear energy in supporting the modern digital economy.

🔗 Sumber: interestingengineering.com