MAROKO133 Eksklusif ai: Adobe Research Unlocking Long-Term Memory in Video World Models wi

📌 MAROKO133 Update ai: Adobe Research Unlocking Long-Term Memory in Video World Mo

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

📌 MAROKO133 Eksklusif ai: World’s first sodium-ion portable power station unveiled

Chinese energy storage and portable power system maker Bluetti has unveiled what it calls the “world’s first” sodium-ion portable power station. Called the Pioneer Na, the system will be available for purchase globally from around mid-October 2025.

First unveiled at the Innovation for All (IFA) conference in Berlin last week, the new system is essentially a large, portable rechargeable battery. Featuring AC and DC outputs, instead of the usual lithium-ion/LFP (lithium iron phosphate) batteries, it uses sodium-ion (Na-ion) batteries.

Bluetti also unveiled some other new products at IFA, including its FridgePower Portable Power Station, a 2,016 Wh, 1,800 W, slim design for appliances. They also unveiled their Apex 300, a 2,764.8 Wh / 3,840 W, expandable to 58,000 Wh (can run a house for days).

Bluetti’s RVSolar 48V System, an expandable to 122 kWh, quick install, for RVs/off-grid homes, also made an appearance.

Portable power for cold climates

According to reports, the Pioneer Na has a 900-watt-hour (Wh) capacity, which is sufficient to power laptops, small appliances, or serve as a backup power source. The system has a standard output of 1,500 watts, with a “Power Lifting” mode for up to 2,250 W (for short bursts or heavy resistive devices like heaters).

The system can be recharged using solar power up to 1,900 W, and has a lifecycle of around 4,000 charge cycles. It is essential to note that the system is approximately 20–25% heavier than its LFP equivalents, with a total weight of 35 pounds (16 kg).

One of the system’s main selling points is its great performance in cold environments. According to reports, it can charge at 5 °F (–15 °C) and discharge at -13 °F (–25 °C).

At the lower end, still delivers 80% discharge capacity (most lithium batteries would shut down or degrade heavily). Even at around 14 °F (–10 °C), it can still recharge to 60% capacity.

To put that into perspective, most LFP batteries usually cannot charge at temperatures below the freezing point. To this end, Bluetti is marketing its new system for cold regions and expeditions (they’re even supplying one to an Antarctic explorer).

Due for release in October 2025

The choice of sodium-ion is a wise one, as sodium is far more abundant and cheaper than lithium. It also performs much better in sub-zero conditions. Battery systems based on it are also potentially safer (less prone to thermal runaway, may handle punctures better).

But, these benefits come with some drawbacks too, such as its relatively lower energy density. This is part of the reason why the system is so bulky and heavy. It is also important to note that it is still new tech, so long-term performance in real-world freeze-thaw cycles isn’t proven yet.

Bluetti’s Pioneer Na is the first real commercial test of sodium-ion batteries in portable power. It’s heavier and slightly less energy-dense than lithium, but it works in deep cold where lithium packs fail, has decent cycle life, and could be safer.

Bluetti is positioning it for households in cold climates and adventurers/explorers as a reliable off-grid backup.

🔗 Sumber: interestingengineering.com