Ebeyer published posts about famous people who had realized that they were aphantasic: Glen Keane, one of the leading Disney animators on "The Little Mermaid" and "Beauty and the Beast"; John Green, the author of "The Fault in Our Stars," whose books had sold more than fifty million copies; J. Craig Venter, the biologist who led the first team to sequence the human genome; Blake Ross, who co-created the Mozilla-Firefox web browser when he was nineteen. Ebeyer also wanted the Aphantasia Network to be a place where aphantasics could find recent scientific research. For instance, estimating the strength of a person's imagery had been thoroughly subjective until Joel Pearson, a cognitive neuroscientist at the University of New South Wales, in Australia, devised tests to measure it more precisely. In a paper from 2022, Pearson reported that when people with imagery visualized a bright object their pupils contracted, as though they were seeing a bright object in real life, but the pupils of aphantasics imagining a bright object stayed the same. Another study of his had shown that, although aphantasics had the same fear response (sweating) as typical imagers to a frightening image shown on a screen, when exposed to a frightening story they barely responded at all.

In Brownsville, Texas, some families found a buzzy new school's methods--surveillance of kids, software in lieu of teachers--to be an education in and of itself. At Alpha School's campus in Brownsville, Texas, a student works on exercises in a learning app. One day last fall, Kristine Barrios' 9-year-old daughter got stuck on a lesson in IXL, the personalized learning software that served as her math teacher. She had to multiply three three-digit numbers without using a calculator. Then she had to do it again, her mom says, more than 20 times, without making mistakes. At Alpha School, the private microschool the girl and her younger brother attended in Brownsville, Texas, she had been working a grade level ahead of her age in math, Barrios says. She could do three-digit multiplication correctly most of the time. But whenever she made an error in IXL, the software would determine she needed more practice and assign her more questions. She told her mom that she had asked her "guide," the adult who supervised her classroom in lieu of a teacher, to make an exception and let her move on. She said the guide's reply was that she needed to get it done, that it was expected of her. The adult guides in Alpha's classrooms "don't do any teaching," says the current head of the Brownsville school.

This paper introduces a novel dynamic knowledge distillation framework, Gompertz-CNN, which integrates the Gompertz growth model into the training process to address the limitations of traditional knowledge distillation. Conventional methods often fail to capture the evolving cognitive capacity of student models, leading to suboptimal knowledge transfer. To overcome this, we propose a stage-aware distillation strategy that dynamically adjusts the weight of distillation loss based on the Gompertz curve, reflecting the student's learning progression: slow initial growth, rapid mid-phase improvement, and late-stage saturation. Our framework incorporates Wasserstein distance to measure feature-level discrepancies and gradient matching to align backward propagation behaviors between teacher and student models. These components are unified under a multi-loss objective, where the Gompertz curve modulates the influence of distillation losses over time. Extensive experiments on CIFAR-10 and CIFAR-100 using various teacher-student architectures (e.g., ResNet50 and MobileNet_v2) demonstrate that Gompertz-CNN consistently outperforms traditional distillation methods, achieving up to 8% and 4% accuracy gains on CIFAR-10 and CIFAR-100, respectively.

Large Language Models (LLMs) have demonstrated impressive capabilities as intelligent agents capable of solving complex problems. However, effective planning in scenarios involving dependencies between API or tool calls-particularly in multi-turn conversations-remains a significant challenge. To address this, we introduce T1, a tool-augmented, multi-domain, multi-turn conversational dataset specifically designed to capture and manage inter-tool dependencies across diverse domains. T1 enables rigorous evaluation of agents' ability to coordinate tool use across nine distinct domains (4 single domain and 5 multi-domain) with the help of an integrated caching mechanism for both short- and long-term memory, while supporting dynamic replanning-such as deciding whether to recompute or reuse cached results. Beyond facilitating research on tool use and planning, T1 also serves as a benchmark for evaluating the performance of open-weight and proprietary large language models. We present results powered by T1-Agent, highlighting their ability to plan and reason in complex, tool-dependent scenarios.

Preference datasets are essential for training general-domain, instruction-following language models with Reinforcement Learning from Human Feedback (RLHF). Each subsequent data release raises expectations for future data collection, meaning there is a constant need to advance the quality and diversity of openly available preference data. To address this need, we introduce HelpSteer3-Preference, a permissively licensed (CC-BY-4.0), high-quality, human-annotated preference dataset comprising of over 40,000 samples. These samples span diverse real-world applications of large language models (LLMs), including tasks relating to STEM, coding and multilingual scenarios. Using HelpSteer3-Preference, we train Reward Models (RMs) that achieve top performance on RM-Bench (82.4%) and JudgeBench (73.7%). This represents a substantial improvement (~10% absolute) over the previously best-reported results from existing RMs. We demonstrate HelpSteer3-Preference can also be applied to train Generative RMs and how policy models can be aligned with RLHF using our RMs. Dataset (CC-BY-4.0): https://huggingface.co/datasets/nvidia/HelpSteer3#preference Models (NVIDIA Open Model): https://huggingface.co/collections/nvidia/reward-models-68377c5955575f71fcc7a2a3

Clustering tabular data remains a significant open challenge in data analysis and machine learning. Unlike for image data, similarity between tabular records often varies across datasets, making the definition of clusters highly dataset-dependent. Furthermore, the absence of supervised signals complicates hyperparameter tuning in deep learning clustering methods, frequently resulting in unstable performance. To address these issues and reduce the need for per-dataset tuning, we adopt an emerging approach in deep learning: zero-shot learning. We propose ZEUS, a self-contained model capable of clustering new datasets without any additional training or fine-tuning. It operates by decomposing complex datasets into meaningful components that can then be clustered effectively. Thanks to pre-training on synthetic datasets generated from a latent-variable prior, it generalizes across various datasets without requiring user intervention. To the best of our knowledge, ZEUS is the first zero-shot method capable of generating embeddings for tabular data in a fully unsupervised manner. Experimental results demonstrate that it performs on par with or better than traditional clustering algorithms and recent deep learning-based methods, while being significantly faster and more user-friendly.

We show that reinforcement learning with verifiable reward using one training example (1-shot RLVR) is effective in incentivizing the math reasoning capabilities of large language models (LLMs). Applying RLVR to the base model Qwen2.5-Math-1.5B, we identify a single example that elevates model performance on MATH500 from 36.0% to 73.6% (8.6% improvement beyond format correction), and improves the average performance across six common mathematical reasoning benchmarks from 17.6% to 35.7% (7.0% non-format gain). This result matches the performance obtained using the 1.2k DeepScaleR subset (MATH500: 73.6%, average: 35.9%), which contains the aforementioned example. Furthermore, RLVR with only two examples even slightly exceeds these results (MATH500: 74.8%, average: 36.6%). Similar substantial improvements are observed across various models (Qwen2.5-Math-7B, Llama3.2-3B-Instruct, DeepSeek-R1-Distill-Qwen-1.5B), RL algorithms (GRPO and PPO), and different math examples. In addition, we identify some interesting phenomena during 1-shot RLVR, including cross-category generalization, increased frequency of self-reflection, and sustained test performance improvement even after the training accuracy has saturated, a phenomenon we term post-saturation generalization. Moreover, we verify that the effectiveness of 1-shot RLVR primarily arises from the policy gradient loss, distinguishing it from the "grokking" phenomenon. We also show the critical role of promoting exploration (e.g., by incorporating entropy loss with an appropriate coefficient) in 1-shot RLVR training. We also further discuss related observations about format correction, label robustness and prompt modification. These findings can inspire future work on RLVR efficiency and encourage a re-examination of recent progress and the underlying mechanisms in RLVR. All resources are open source at https://github.com/ypwang61/One-Shot-RLVR.

AI agents hold the potential to revolutionize scientific productivity by automating literature reviews, replicating experiments, analyzing data, and even proposing new directions of inquiry; indeed, there are now many such agents, ranging from general-purpose "deep research" systems to specialized science-specific agents, such as AI Scientist and AIGS. Rigorous evaluation of these agents is critical for progress. Yet existing benchmarks fall short on several fronts: they (1) fail to provide holistic, product-informed measures of real-world use cases such as science research; (2) lack reproducible agent tools necessary for a controlled comparison of core agentic capabilities; (3) do not account for confounding variables such as model cost and tool access; (4) do not provide standardized interfaces for quick agent prototyping and evaluation; and (5) lack comprehensive baseline agents necessary to identify true advances. In response, we define principles and tooling for more rigorously benchmarking agents. Using these, we present AstaBench, a suite that provides the first holistic measure of agentic ability to perform scientific research, comprising 2400+ problems spanning the entire scientific discovery process and multiple scientific domains, and including many problems inspired by actual user requests to deployed Asta agents. Our suite comes with the first scientific research environment with production-grade search tools that enable controlled, reproducible evaluation, better accounting for confounders. Alongside, we provide a comprehensive suite of nine science-optimized classes of Asta agents and numerous baselines. Our extensive evaluation of 57 agents across 22 agent classes reveals several interesting findings, most importantly that despite meaningful progress on certain individual aspects, AI remains far from solving the challenge of science research assistance.

Understanding how information is dynamically accumulated and transformed in human reasoning has long challenged cognitive psychology, philosophy, and artificial intelligence. Existing accounts, from classical logic to probabilistic models, illuminate aspects of output or individual modelling, but do not offer a unified, quantitative description of general human reasoning dynamics. To solve this, we introduce Information Flow Tracking (IF-Track), that uses large language models (LLMs) as probabilistic encoder to quantify information entropy and gain at each reasoning step. Through fine-grained analyses across diverse tasks, our method is the first successfully models the universal landscape of human reasoning behaviors within a single metric space. We show that IF-Track captures essential reasoning features, identifies systematic error patterns, and characterizes individual differences. Applied to discussion of advanced psychological theory, we first reconcile single- versus dual-process theories in IF-Track and discover the alignment of artificial and human cognition and how LLMs reshaping human reasoning process. This approach establishes a quantitative bridge between theory and measurement, offering mechanistic insights into the architecture of reasoning.

Biological and artificial learners are inherently exposed to a stream of data and experience throughout their lifetimes and must constantly adapt to, learn from, or selectively ignore the ongoing input. Recent findings reveal that, even when the performance remains stable, the underlying neural representations can change gradually over time, a phenomenon known as representational drift. Studying the different sources of data and noise that may contribute to drift is essential for understanding lifelong learning in neural systems. However, a systematic study of drift across architectures and learning rules, and the connection to task, are missing. Here, in an online learning setup, we characterize drift as a function of data distribution, and specifically show that the learning noise induced by task-irrelevant stimuli, which the agent learns to ignore in a given context, can create long-term drift in the representation of task-relevant stimuli. Using theory and simulations, we demonstrate this phenomenon both in Hebbian-based learning -- Oja's rule and Similarity Matching -- and in stochastic gradient descent applied to autoencoders and a supervised two-layer network. We consistently observe that the drift rate increases with the variance and the dimension of the data in the task-irrelevant subspace. We further show that this yields different qualitative predictions for the geometry and dimension-dependency of drift than those arising from Gaussian synaptic noise. Overall, our study links the structure of stimuli, task, and learning rule to representational drift and could pave the way for using drift as a signal for uncovering underlying computation in the brain.