MAROKO133 Eksklusif ai: Nvidia Stock Lurches Down as SoftBank Pulls Entire Investment Waji

📌 MAROKO133 Breaking ai: Nvidia Stock Lurches Down as SoftBank Pulls Entire Invest

Last month, we shared a story about Seaport Global Securities analyst Jay Goldberg, who defied the odds as the only voice out of 80 Bloomberg stock watchers who rated Nvidia as a “sell.” While Goldberg makes some very good points — mostly, that AI simply isn’t proficient enough, at least yet, to justify its incredible chokehold on the US economy — the consensus on Wall Street has been firmly against him.

Case in point, even after Goldberg made his assessment in the last week of October, investors continued to dump money into Nvidia, which is seen as the “shovel merchant” to the AI goldrush. So much money, in fact, that the company’s market cap briefly shot up above $5 trillion — the first company in history to do so.

All that is to say, conventional wisdom is that you’d be nuts to pull an investment in Nvidia right now. Yet that’s exactly what Japanese investment firm SoftBank did on Tuesday, sending numerous tech stocks into a slide.

According to Bloomberg, SoftBank disclosed it was pulling its $5.8 billion worth of holdings in Nvidia early Tuesday morning, ending its reign as one of the chip giant’s most prominent backers. The news immediately sent shockwaves throughout the market, dragging Nvidia’s stock down by around 2.6 percent as of Tuesday afternoon.

As if to show how codependent the AI industry is, other tech giants were dragged down too, with stock in companies like Tesla falling by 1.8 percent, Meta by 0.95 percent, and Intel around 0.9 percent. Overall, the tech-heavy Nasdaq Index was down by 0.2 percent, while the S&P 500 struggled to maintain a flatline.

Overall, it would be awful news for the AI industry if it wasn’t for one little wrinkle: SoftBank is planning on using its previous Nvidia holdings to back OpenAI, the private company behind ChatGPT. It’s already given OpenAI $7.5 billion, with plans for another $22.5 billion soon, as Bloomberg reports.

In other words, don’t be shocked if the sell-off evens out in the coming days, as Yahoo Finance suggests.

That said, there is at least one lesson to take away from the episode. While most other tech stocks tumbled, Apple soared by over 1.5 percent, its share price reaching an all-time high of $273.53. This is despite a disappointing announcement on Monday that the tech giant was delaying the release of the next iPhone Air in 2026. So what gives?

When you peel back the “big tech” trappings, Apple stands out as the largest tech corporation to maintain ambivalence about the AI boom. Back in June, Apple’s research lab dropped a bombshell paper calling out companies like OpenAI for selling “the illusion of thinking” in AI chatbots.

Apple has increasingly downplayed any efforts at building out a proprietary AI model, after early attempts failed rather spectacularly.

All in all, don’t count on Apple to weather the storm completely if the entire AI bubble begins to pop, but when it comes to the touch-and-go volatility of the AI news cycle, the company might just have the last laugh.

More on Nvidia: Nvidia CEO Says China Is “Going to Win” the AI Race

The post Nvidia Stock Lurches Down as SoftBank Pulls Entire Investment appeared first on Futurism.

🔗 Sumber: futurism.com

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

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