Named after the hundred-eyed watchman of Greek myth, Argus watches the education landscape: spotting new opportunities, pressure-testing the ventures we're building, and tracing every read back to the real-world signals behind it.
The evidence library: the raw signals the pipeline is watching across the education ecosystem. Every idea is built from these.
Some students with disabilities rely on assistive technology to learn, and they worry it could be swept up in the movement to get screens out of schools.
Hi HN, I'm Rosa, a massage therapist for over 30 years. I noticed my clients felt relaxed after a massage, but their stress and muscle tension always came back. A one-hour massage isn't always enough to combat a long workweek. I saw that people needed info on how to take care of their bodies between appointments, not just treatment. That's why we built MASSAGE BY ROSA – a wellness platform to meet this need. Here’s what we offer: It’s a two-part deal: -Massage Therapy: Hands-on therapy for pain relief – based on methods from South Florida. -Online Body Therapy Courses: This is what I want to share. I turned my knowledge into video courses teaching self-massage, workstation adjustments, and ways to release tension. It’s like having a therapist help you stay well. The Tech: We’re keeping it basic with a static site for course content and subscriptions, which lets us focus on making great video lessons. We're launching this to solve the problem of upkeep in physical wellness. It's for peo
As someone who’s dedicated my career to advancing the Science of Reading movement, I’ve seen firsthand what it takes to help every child become a strong, fluent reader.
The House VA Committee voted 19-0 to subpoena Oracle Health executives Larry Ellison and Mike Sicilia after learning the VA’s EHR contract ballooned from $10 billion to $27 billion — despite Oracle’s promise to Congress that it would absorb any costs beyond the original cap. The post Oracle Health Execs Subpoenaed After VA Contract Costs Nearly Triple appeared first on MedCity News .
There are small and large ways teachers, parents and administrators can cut back on the amount of screen time in school that detracts from learning.
Food stamps have a direct link to other critical resources, such as free breakfast and lunch at school. Now, experts say they're starting to see that link break.
A new model in Vermont emphasizes hands-on learning while stripping away frills like gyms and meal plans that increase costs. Other newly created institutions are also reimagining college.
Gov. Greg Abbott directed the Texas Higher Education Coordinating Board on Wednesday to work on pathways for students to earn bachelor’s degrees in three years. Such a program could have college students take about 90 hours of classes instead of 120 to earn a degree, thus reducing the cost of higher education, shortening the time […]
Ionis Pharmaceuticals’ zilganersen, brand name Zanvastro, is the first disease-modifying therapy approved for the neurological disorder Alexander disease. While Ionis has experience developing medicines for rare neuroscience indications, Zanvastro will be the first neurology product the company brings to the market without a commercialization partner. The post FDA Approves Ionis Pharma Drug, the First for Ultra-Rare Alexander Disease appeared first on MedCity News .
Before the beginning of each school year, teachers spend time getting their classrooms ready for students. Everything has its place, from whiteboards to pens, pencils and pushpins. With the adoption of cloud computing by school districts, the IT departments of K–12 schools also must figure out what goes where, but on a much larger level. Many districts operate in hybrid cloud environments, where some data and applications reside on servers that are on-premises and others reside with public cloud providers. It’s not a decision that should be taken lightly. Click the banner below to see how a…
The bill seeks to keep all special education, postsecondary, Native American, and elementary and secondary education activities within the agency.
The reality of K–12 IT teams is that most are very small — and, as a result, stretched thin. More than half of K–12 districts say they’re understaffed for everyday classroom technical support needs. On a given day, technicians may handle dozens of support requests for password resets, basic device troubleshooting, network connectivity issues, broken devices and more. When a district’s technical needs outweigh its team’s capacity to support it, something has to give. The solution? Agentic artificial intelligence. Agentic AI isn’t here to replace IT staff. Instead, it gives small departments a…
As teenagers return to school this fall, a new report suggests that many of them are moving through adolescence without the experiences that help prepare them for adulthood, activities like a part-time job, a club, a team and a regular family meal. The clearest evidence is teen employment, which alone has been declining for nearly […]
Three things ski patrol can teach about building a business The post Bringing Ski Patrol Lessons to Healthtech Entrepreneurship appeared first on MedCity News .
By approving the policy Thursday, the State University System of Florida’s board effectively shut undocumented students out of its 12 institutions.
The school board voted to close nine schools last year, and 27 were recently being evaluated to determine whether they should "remain operational." The post New Miami-Dade superintendent pauses look at closing 27 schools appeared first on District Administration .
The data, from nearly every public school in the country, reveals disparities in the way students are treated based on race, ethnicity, disability, sex and other factors. The post Federal civil rights data about students is finally out. It’s different under Trump appeared first on District Administration .
For students starting at a new school, the first day holds many questions: Will they see any familiar faces? Who will they sit with at lunch? Who will they play with at recess? “I’m this much nervous,” said Iman Fair-Seldon, 5, holding her hands about 4 inches (10 centimeters) apart. Iman will start first grade next […]
Schools are racing to teach students how to use AI. That is necessary, but it is not enough. The workplace advantage will increasingly belong to people who can decide when an AI output is useful, when it is incomplete and when it is wrong.
What if the next breakthrough in school design did not require a new building, a charter, or a pilot program imported from somewhere else? At Hidden Valley Middle School in Escondido, a public microschool operating inside a traditional campus has cut suspensions by 60 percent and reduced chronic absenteeism by 12 percentage points, schoolwide. This piece by Dr. Katie Martin and Dr. Devin Vodicka explores how California districts are building the internal capacity to generate and spread learner-centered models, and why that shift from showcase schools to generative systems may be the most important move in education reform right now. The post Districts Don’t Need Another New Model. They Need Systems That Build Them. appeared first on Getting Smart .
With the rapid spread of Gen AI, we’re entering a time when human connection is becoming more valuable and more vulnerable. The post Can high-tech scale high-touch? appeared first on eCampus News .
(Some) Starter Questions for Measuring Learning johnw@mcsweeneys.net Fri, 09/04/2026 - 03:00 AM Let’s do it, if we’re going to do it. Byline(s) John Warner (Some) Starter Questions for Measuring Learning johnw@mcsweeneys.net Fri, 09/04/2026 - 03:00 AM Let’s do it, if we’re going to do it. Byline(s) John Warner
More U.S. Researchers Secure European Research Council Grants Susan H. Greenberg Fri, 09/04/2026 - 03:00 AM The program for early-career researchers sees record demand from outside Europe, even as the E.U. has less money to distribute than last year. Byline(s) Seher Asaf for Times Higher Education More U.S. Researchers Secure European Research Council Grants Susan H. Greenberg Fri, 09/04/2026 - 03:00 AM The program for early-career researchers sees record demand from outside Europe, even as the E.U. has less money to distribute than last year. Byline(s) Seher Asaf for Times Higher Education
4 Ways Colleges Are Making Support Easier to Access Joshua.Bay Fri, 09/04/2026 - 03:00 AM From offering mental health services after hours to putting all essential needs under one roof, colleges are designing support systems around students’ lives. Byline(s) Joshua Bay 4 Ways Colleges Are Making Support Easier to Access Joshua.Bay Fri, 09/04/2026 - 03:00 AM From offering mental health services after hours to putting all essential needs under one roof, colleges are designing support systems around students’ lives. Byline(s) Joshua Bay
U of Washington Pays Professor $600K, Allows ‘Parody Land Acknowledgment’ Ryan Quinn Fri, 09/04/2026 - 03:00 AM Byline(s) Ryan Quinn U of Washington Pays Professor $600K, Allows ‘Parody Land Acknowledgment’ Ryan Quinn Fri, 09/04/2026 - 03:00 AM Byline(s) Ryan Quinn
Reading the Rules for Rule Makers, Part 1 Sara Brady Fri, 09/04/2026 - 03:00 AM Conflating academic freedom with perceived viewpoint diversity: such a fundamental category error as to imply bad faith. Byline(s) Matt Reed Reading the Rules for Rule Makers, Part 1 Sara Brady Fri, 09/04/2026 - 03:00 AM Conflating academic freedom with perceived viewpoint diversity: such a fundamental category error as to imply bad faith. Byline(s) Matt Reed
The House Republican Trying to Redefine Graduate Student Loan Caps jessica.blake@… Fri, 09/04/2026 - 03:00 AM Inside Higher Ed called up Rep. Michael Lawler to talk about his push to expand the list of programs whose students get access to higher loan caps. Byline(s) Jessica Blake The House Republican Trying to Redefine Graduate Student Loan Caps jessica.blake@… Fri, 09/04/2026 - 03:00 AM Inside Higher Ed called up Rep. Michael Lawler to talk about his push to expand the list of programs whose students get access to higher loan caps. Byline(s) Jessica Blake
OMB Grants Rule Delayed Until Mid-December Katherine Knott Fri, 09/04/2026 - 03:00 AM Byline(s) Katherine Knott OMB Grants Rule Delayed Until Mid-December Katherine Knott Fri, 09/04/2026 - 03:00 AM Byline(s) Katherine Knott
Trump Administration Threatens Colleges’ Tax-Exempt Status Emma Whitford Fri, 09/04/2026 - 03:00 AM Under the proposed rule, any college that adopts or maintains admissions policies, scholarships, loans or other programs for students of color could lose its tax-exempt status. Byline(s) Emma Whitford Trump Administration Threatens Colleges’ Tax-Exempt Status Emma Whitford Fri, 09/04/2026 - 03:00 AM Under the proposed rule, any college that adopts or maintains admissions policies, scholarships, loans or other programs for students of color could lose its tax-exempt status. Byline(s) Emma Whitford
House Passes Bill Blocking Aid to Colleges That Boycott Israel Susan H. Greenberg Fri, 09/04/2026 - 03:00 AM Byline(s) Susan H. Greenberg House Passes Bill Blocking Aid to Colleges That Boycott Israel Susan H. Greenberg Fri, 09/04/2026 - 03:00 AM Byline(s) Susan H. Greenberg
Trump’s Latest Move Threatens Any College That Seeks to Promote Equality Susan H. Greenberg Fri, 09/04/2026 - 03:00 AM The administration’s plan to revoke tax-exempt status from schools that offer Black or other minority students targeted support willfully distorts the law. Byline(s) Austin Sarat Trump’s Latest Move Threatens Any College That Seeks to Promote Equality Susan H. Greenberg Fri, 09/04/2026 - 03:00 AM The administration’s plan to revoke tax-exempt status from schools that offer Black or other minority students targeted support willfully distorts the law. Byline(s) Austin Sarat
Among survyed middle-income high school students, 66% said educators heavily emphasized four-year college when talking about future pathways.
From an increase in after-school activity participation to a major school system's AI pause, what did you learn from our recent stories?
Mikiah Roberson expected to receive her federal $6,700 student loan disbursement sometime between May 27 and June 3, just in time to make her rent and car payments. Instead, June 3 came and went. Days, weeks and finally over a month passed. Roberson, a graduate student in University of Maryland Global Campus’s digital forensics and […] The post Colleges are scrambling to figure out new student loan rules. Students are paying the price appeared first on The Hechinger Report .
Cathy Vatterott, author of “The Teens Are Not Alright,” believes schools are treating the symptoms of student stress instead of looking at the real problem.
There are hundreds of great nurse discounts available in 2026! We even have Nurse.org exclusive discounts that we've partnered with your favorite nurse brands to offer our readers. From Verizon…
Anesthesia technicians are healthcare professionals who support anesthesia providers by preparing, maintaining, cleaning, and troubleshooting the equipment used during anesthesia care. If you're interested in surgery and anesthesia but don't…
Anesthesia technicians are healthcare professionals who support anesthesia providers by preparing, maintaining, cleaning, and troubleshooting the equipment used during anesthesia care. If you're interested in surgery and anesthesia but don't…
arXiv:2608.19491v2 Announce Type: replace-cross Abstract: Most modern optimizers form their momentum as an exponential moving average (EMA) of past gradients, forgetting every direction at one fixed rate. However, the inputs a deep network sees during training can be highly anisotropic, with a few directions queried frequently while most are seen rarely. Preconditioning methods address this anisotropy by wrapping extra processing around this buffer and leave the momentum update itself unchanged. We propose Activation-Keyed Momentum (AK-Momentum), which builds direction-awareness into the momentum update rule. The gradient of a linear layer splits into an input activation that acts as a key and an output-side error that acts as a value. Keying on that activation, AK-Momentum updates the momentum buffer by the canonical delta rule, so each direction is forgotten at a rate set by how often it appears. We prove that it is a valid momentum, that it applies the input-side curvature correctio
arXiv:2606.21657v2 Announce Type: replace-cross Abstract: Do people perceive the same facial expression in the same way? Should we expect vision models to be flexible in how they perceive facial expressions? Facial expressions are nonverbal social signals used in human interaction, but facial expression recognition datasets often focus on a single deterministic annotation per sample. We introduce Chehre, an emoji-prompted video dataset with a wide range of dynamic facial expressions for exploring perceptual variation. In Chehre, 203 participants were prompted to express and record 40 facial emojis. Later, their facial motions were transferred onto synthetic faces to preserve privacy. A separate group annotated the videos, resulting in 2,111 videos annotated by 1,242 perceivers, with ~30 annotators per video. Chehre enables us to define a new task: "distributional expression recognition", which tests whether a model can reproduce the variation observed across annotator responses. We tes
arXiv:2606.19264v2 Announce Type: replace-cross Abstract: The knowledge encoded in large language models (LLMs) can serve as a substrate for structured reasoning over variables describing a complex world, but accessing this knowledge in a probabilistically coherent manner poses a difficult inference problem. We propose Large Language Gibbs, a scheme for structured probabilistic inference that uses conditional distributions of an LLM as transition operators. Rather than sampling structured objects through single-pass autoregressive generation, we iteratively resample individual variables conditioned on others using an LLM's next-token conditionals. This approach avoids order-dependent biases and produces a stationary distribution that reflects a compromise between all local conditionals. We apply this approach to sampling from synthetic distributions, consistent reasoning tasks, and Bayesian structure learning. The results suggest that the use of LLM conditionals in MCMC is a practical
arXiv:2606.18388v2 Announce Type: replace-cross Abstract: RL post-training strategies are dataset-dependent and reveal a recurring empirical pattern: capacity parameters accumulate monotonically across stages, while regularization parameters predominantly oscillate in response to shifting training dynamics. This distinction highlights a potential flaw in fixed training schedules: by forcing all parameters along rigid paths, they fail to capture the dynamic exploration-exploitation tradeoffs that regularization must track. We uncover this through LLMZero, an agentic system that optimizes training trajectories via tree search by diagnosing pathologies at each checkpoint and proposing coordinated multi-parameter transitions. Across four diverse GRPO tasks, LLMZero discovers strategies that improve over the base model by 9% to 140% and over grid search by 6% to 15% (relative), consistently outperforming random search and a skill-based agent under a matched compute budget. The capacity--reg
arXiv:2606.07451v2 Announce Type: replace-cross Abstract: Vision-language models such as CLIP are highly useful for diverse tasks due to their shared image-text embedding space. Despite this, the image and text embeddings are often poorly aligned, affecting downstream performance. Recent work has hypothesized that this can be attributed to an information imbalance: images contain more information than their captions describe. In this work, we propose TEVI, a framework that uses captions as a signal for what to retain from image embeddings. Specifically, we use sparse autoencoders to disentangle image embeddings and train a masking module to selectively reconstruct the embedding based on a given caption. In a controlled setup with synthetic captions, we show that TEVI is effective at preserving caption-described attributes while discarding others. We find that this extends to CLIP models trained on natural images, where TEVI learns to mask meaningfully and allows retrieval based on cond
arXiv:2606.02914v3 Announce Type: replace-cross Abstract: Background: Oral diseases affect nearly 3.5 billion people worldwide, yet the comparative clinical potential of large-scale AI models in dentistry remains poorly understood. Three distinct model categories have emerged: language-generative models, discriminative vision foundation models, and dental-specific foundation models, with no unified review examining their relationships and collective limitations. Methods: Following PRISMA-ScR guidelines, we systematically searched four databases (PubMed, Google Scholar, Scopus, arXiv), screened independently by two reviewers. After applying inclusion/exclusion criteria, 97 studies (2020-2026) were included. We propose a two-dimensional classification framework organizing models by architectural paradigm and dental specialization degree. Results: Language-generative models excel at text-based tasks (clinical reasoning, licensing exams, patient communication) but show inconsistent perform
arXiv:2606.02372v2 Announce Type: replace-cross Abstract: Equipping language agents with world models enables them to anticipate environment dynamics and evaluate candidate actions before execution. However, existing textual world models are typically fixed after training, preventing them from adapting to the on-policy state-action distributions induced by an evolving agent. Meanwhile, agent-improvement methods often rely on external rewards or verifiers, limiting their applicability in realistic interactive environments. In this paper, we propose COMAP, a novel framework that co-evolves textual world models and agent policies through closed-loop interaction. At each decision step, the world model predicts future state feedback for candidate actions, and the agent performs future-aware reflection by estimating the reliability of this feedback and refining its action accordingly. The resulting on-policy trajectories are then used to update the world model via self-distillation, allowing
arXiv:2605.30912v2 Announce Type: replace-cross Abstract: Reinforcement learning with verifiable rewards (RLVR) improves vision-language models (VLMs) by optimizing outcome rewards derived from final answers. However, such outcome-only rewards do not tell the model which image regions justify an answer. For questions that require visual grounding, these rewards cannot distinguish responses supported by relevant visual evidence from those produced by language-prior shortcuts or lucky guesses. We introduce EASE (Evidence-Anchored Spatial Attention), which augments multimodal RLVR with visual-evidence process supervision. EASE converts annotated evidence regions into a smoothed visual-token target and uses it to guide response-to-image attention during RL training, but only on high-reward trajectories. The annotations are used solely as privileged training labels, while inference requires only the original image and question. Across Qwen2.5-VL-7B, Qwen3-VL-4B, and Qwen3-VL-8B, EASE raises
arXiv:2602.07106v3 Announce Type: replace-cross Abstract: Omni-modal large language models (OLLMs) aim to unify multimodal understanding and generation, yet extending them to jointly produce speech and 3D facial animation remains largely underexplored. A key challenge is the mismatch between the discrete semantic reasoning of LLMs and the dense temporal dynamics required for 3D facial motion. We propose Expressive Omni (Ex-Omni), a framework that augments OLLMs with speech-accompanied 3D facial animation. Ex-Omni decouples semantic reasoning from temporal generation through a speech-unit generator with blendshape co-supervision and a non-autoregressive blendshape decoder, where speech units provide temporal scaffolding and hidden speech representations carry facially relevant cues. We further introduce a token-as-query gated fusion (TQGF) interface for controlled semantic injection, as well as InstructS2SF-1200K, a 1.2M-sample weakly supervised dataset for speech-accompanied facial ani
arXiv:2602.06065v4 Announce Type: replace-cross Abstract: Understanding how the structure of language can be learned from sentences alone is a central question in both cognitive science and machine learning. Studies of the internal representations of Large Language Models (LLMs) support their ability to parse text when predicting the next word, while representing semantic notions independently of surface form. Yet, which data statistics make these feats possible, and how much data is required, remain largely unknown. Probabilistic context-free grammars (PCFGs) provide a tractable testbed for studying these questions. However, prior work has focused either on the post-hoc characterization of the parsing-like algorithms used by trained networks; or on the learnability of PCFGs with fixed syntax, where parsing is unnecessary. Here, we (i) introduce a tunable class of PCFGs in which both the degree of ambiguity and the correlation structure across scales can be controlled; (ii) provide a l
arXiv:2503.03313v4 Announce Type: replace-cross Abstract: Text-Attributed Graphs (TAGs), where each node is associated with text descriptions, are ubiquitous in real-world scenarios. They typically exhibit distinctive structure and domain-specific knowledge, motivating the development of a Graph Foundation Model (GFM) that generalizes across diverse graphs and tasks. Despite large efforts to integrate Large Language Models (LLMs) and Graph Neural Networks (GNNs) for TAGs, existing approaches suffer from decoupled architectures with two-stage alignment, limiting their synergistic potential. Even worse, existing methods assign out-of-vocabulary (OOV) tokens to graph nodes, leading to graph-specific semantics, token explosion, and incompatibility with task-oriented prompt templates, which hinders cross-graph and cross-task transferability. To address these challenges, we propose PromptGFM, a versatile GFM for TAGs grounded in graph vocabulary learning. PromptGFM comprises two key componen
arXiv:2405.21047v4 Announce Type: replace-cross Abstract: Large Language Models (LLMs) struggle with reliably generating highly structured outputs, such as program code, mathematical formulas, or well-formed markup. Constrained decoding approaches mitigate this problem by greedily restricting what tokens an LLM can output at each step to guarantee that the output matches a given constraint. Specifically, in grammar-constrained decoding (GCD), the LLM's output must follow a given grammar. In this paper, we demonstrate that GCD techniques (and in general constrained decoding techniques) can distort the LLM's distribution, leading to outputs that are grammatical but appear with likelihoods that are not proportional to the ones given by the LLM, and so ultimately are low-quality. We call the problem of aligning sampling with a grammar constraint, grammar-aligned decoding (GAD), and propose adaptive sampling with approximate expected futures (ASAp), a decoding algorithm that guarantees the