📌 TOPINDIATOURS Breaking ai: Evidence Grows That One of the Largest Known Stars Is

One of the largest known stars in the cosmos is poised for catastrophe.

After witnessing the massive object undergo a dramatic transformation, a team of astronomers say the star is on the verge of exploding in a powerful supernova, they report in a new study published in the journal Nature Astronomy. Or, they speculate, it could collapse directly into a black hole due its incredible mass.

Since it was first discovered some five decades ago, the star, WOH G64, has been classified as a red supergiant with a mass thirty times that of the Sun. But it’s the supergiant’s size that truly boggles the mind. With a radius over 1,500 times that of the Sun, it would stretch past the orbit of Jupiter if it were placed in the middle of our solar system.

Supergiant stars are short-lived. WOH G64 is only around five million years old, when our star is 4.6 billion. But they have a taste for the spectacular, ranking among the brightest stars in the cosmos, on top of their epic scale. This one is located some 165,000 light years away in a dense region of space called the Large Magellanic Cloud, a dwarf galaxy that orbits the Milky Way. It’s a fertile star-forming region packed with enough material to give birth to oversized behemoths like WOH G64. 

Being born so massive means that the stars grow quickly. As WOH G64 aged, it quickly burned through the hydrogen at its core and resorted to burning helium. This second wind of heating caused the star’s outer layers to quickly expand — which, as the core contracts, allows more heat to dissipate. This causes the star to cool, resulting in its red appearance. And thus, a red supergiant is formed.

But WOH G64 may be transforming yet again into something even more awe-inspiring. The astronomers noticed that, in 2014, WOH G64’s color and temperature dramatically but smoothly changed in under a year, suggesting that it may be evolving into a yellow hypergiant. The largest of these stars are so voluminous that they can fit several billion Suns inside them.

“Yellow hypergiants are extremely rare because they represent a short-lived transitional phase between the red supergiant stage and the eventual supernova explosion,” lead author Gonzalo Muñoz-Sanchez at the National Observatory of Athens told Space.com. “Consequently, only a small number of confirmed yellow hypergiants are currently known, amounting to just a few tens of objects.”

The transformation into a hypergiant occurred, the astronomers propose, after WOH G64 ejected a large portion of its outer layers into space. This was spurred by interactions with a companion star which stripped material from the WOH G64’s surface to form a vast shell of hydrogen — a common envelope — that swallowed both stars.

But the astronomers also can’t rule out the possibility that this transformation is taking place independently of the companion star’s interference.

“Even though the system is binary, the transition may have been driven by intrinsic stellar processes. In this case, the star may have undergone an extraordinary eruptive episode lasting more than 30 years and is now returning to a yellow, quiescent state,” Muñoz-Sanchez told Space.com. “Both possibilities are extremely rare, and witnessing either occur on human timescales is nearly unprecedented.”

Its unclear nature makes it difficult to predict how it will die, but it’s guaranteed to be a catastrophe of some kind, and one that will happen “soon” in cosmic terms, according to Muñoz-Sanchez, meaning anywhere from hundreds to thousands of years. The giant star could go supernova, exploding dramatically or instantly inverting into a black hole. Or it could collide with its companion star.

“The fate of stars with initial masses between 23 and 30 solar masses after evolving into red supergiants is still uncertain,” Muñoz-Sanchez told Space.com. WOH G64’s behaviour could suggest that red supergiants become yellow hypergiants before finally winking out.

More on stars: Scientists Intrigued as Prominent Star Suddenly Winks Out of Existence

The post Evidence Grows That One of the Largest Known Stars Is Poised to Explode in a Spectacular Blast appeared first on Futurism.

🔗 Sumber: futurism.com

📌 TOPINDIATOURS Update ai: MIT Researchers Unveil “SEAL”: A New Step Towards Self-

The concept of AI self-improvement has been a hot topic in recent research circles, with a flurry of papers emerging and prominent figures like OpenAI CEO Sam Altman weighing in on the future of self-evolving intelligent systems. Now, a new paper from MIT, titled “Self-Adapting Language Models,” introduces SEAL (Self-Adapting LLMs), a novel framework that allows large language models (LLMs) to update their own weights. This development is seen as another significant step towards the realization of truly self-evolving AI.

The research paper, published yesterday, has already ignited considerable discussion, including on Hacker News. SEAL proposes a method where an LLM can generate its own training data through “self-editing” and subsequently update its weights based on new inputs. Crucially, this self-editing process is learned via reinforcement learning, with the reward mechanism tied to the updated model’s downstream performance.

The timing of this paper is particularly notable given the recent surge in interest surrounding AI self-evolution. Earlier this month, several other research efforts garnered attention, including Sakana AI and the University of British Columbia’s “Darwin-Gödel Machine (DGM),” CMU’s “Self-Rewarding Training (SRT),” Shanghai Jiao Tong University’s “MM-UPT” framework for continuous self-improvement in multimodal large models, and the “UI-Genie” self-improvement framework from The Chinese University of Hong Kong in collaboration with vivo.

Adding to the buzz, OpenAI CEO Sam Altman recently shared his vision of a future with self-improving AI and robots in his blog post, “The Gentle Singularity.” He posited that while the initial millions of humanoid robots would need traditional manufacturing, they would then be able to “operate the entire supply chain to build more robots, which can in turn build more chip fabrication facilities, data centers, and so on.” This was quickly followed by a tweet from @VraserX, claiming an OpenAI insider revealed the company was already running recursively self-improving AI internally, a claim that sparked widespread debate about its veracity.

Regardless of the specifics of internal OpenAI developments, the MIT paper on SEAL provides concrete evidence of AI’s progression towards self-evolution.

Understanding SEAL: Self-Adapting Language Models

The core idea behind SEAL is to enable language models to improve themselves when encountering new data by generating their own synthetic data and optimizing their parameters through self-editing. The model’s training objective is to directly generate these self-edits (SEs) using data provided within the model’s context.

The generation of these self-edits is learned through reinforcement learning. The model is rewarded when the generated self-edits, once applied, lead to improved performance on the target task. Therefore, SEAL can be conceptualized as an algorithm with two nested loops: an outer reinforcement learning (RL) loop that optimizes the generation of self-edits, and an inner update loop that uses the generated self-edits to update the model via gradient descent.

This method can be viewed as an instance of meta-learning, where the focus is on how to generate effective self-edits in a meta-learning fashion.

A General Framework

SEAL operates on a single task instance (C,τ), where C is context information relevant to the task, and τ defines the downstream evaluation for assessing the model’s adaptation. For example, in a knowledge integration task, C might be a passage to be integrated into the model’s internal knowledge, and τ a set of questions about that passage.

Given C, the model generates a self-edit SE, which then updates its parameters through supervised fine-tuning: θ′←SFT(θ,SE). Reinforcement learning is used to optimize this self-edit generation: the model performs an action (generates SE), receives a reward r based on LMθ′’s performance on τ, and updates its policy to maximize the expected reward.

The researchers found that traditional online policy methods like GRPO and PPO led to unstable training. They ultimately opted for ReST^EM, a simpler, filtering-based behavioral cloning approach from a DeepMind paper. This method can be viewed as an Expectation-Maximization (EM) process, where the E-step samples candidate outputs from the current model policy, and the M-step reinforces only those samples that yield a positive reward through supervised fine-tuning.

The paper also notes that while the current implementation uses a single model to generate and learn from self-edits, these roles could be separated in a “teacher-student” setup.

Instantiating SEAL in Specific Domains

The MIT team instantiated SEAL in two specific domains: knowledge integration and few-shot learning.

• Knowledge Integration: The goal here is to effectively integrate information from articles into the model’s weights.
• Few-Shot Learning: This involves the model adapting to new tasks with very few examples.

Experimental Results

The experimental results for both few-shot learning and knowledge integration demonstrate the effectiveness of the SEAL framework.

In few-shot learning, using a Llama-3.2-1B-Instruct model, SEAL significantly improved adaptation success rates, achieving 72.5% compared to 20% for models using basic self-edits without RL training, and 0% without adaptation. While still below “Oracle TTT” (an idealized baseline), this indicates substantial progress.

For knowledge integration, using a larger Qwen2.5-7B model to integrate new facts from SQuAD articles, SEAL consistently outperformed baseline methods. Training with synthetically generated data from the base Qwen-2.5-7B model already showed notable improvements, and subsequent reinforcement learning further boosted performance. The accuracy also showed rapid improvement over external RL iterations, often surpassing setups using GPT-4.1 generated data within just two iterations.

Qualitative examples from the paper illustrate how reinforcement learning leads to the generation of more detailed self-edits, resulting in improved performance.

While promising, the researchers also acknowledge some limitations of the SEAL framework, including aspects related to catastrophic forgetting, computational overhead, and context-dependent evaluation. These are discussed in detail in the original paper.

Original Paper: https://arxiv.org/pdf/2506.10943

Project Site: https://jyopari.github.io/posts/seal

Github Repo: https://github.com/Continual-Intelligence/SEAL

The post MIT Researchers Unveil “SEAL”: A New Step Towards Self-Improving AI first appeared on Synced.

🔗 Sumber: syncedreview.com