MaMe & MaRe: Matrix-Based Token Merging and Restoration for Efficient Visual Perception and Synthesis

Token compression is crucial for mitigating the quadratic complexity of self-attention mechanisms in Vision Transformers (ViTs), which often involve numerous input tokens. Existing methods, such as ToMe, rely on GPU-inefficient operations (e.g., sorting, scattered writes), introducing overheads that limit their effectiveness. We introduce MaMe, a training-free, differentiable token merging method based entirely on matrix operations, which is GPU-friendly to accelerate ViTs. Additionally, we present MaRe, its inverse operation, for token restoration, forming a MaMe+MaRe pipeline for image synthesis. When applied to pre-trained models, MaMe doubles ViT-B throughput with a 2% accuracy drop. Notably, fine-tuning the last layer with MaMe boosts ViT-B accuracy by 1.0% at 1.1x speed. In SigLIP2-B@512 zero-shot classification, MaMe provides 1.3x acceleration with negligible performance degradation. In video tasks, MaMe accelerates VideoMAE-L by 48.5% on Kinetics-400 with only a 0.84% accuracy loss. Furthermore, MaMe achieves simultaneous improvements in both performance and speed on some tasks. In image synthesis, the MaMe+MaRe pipeline enhances quality while reducing Stable Diffusion v2.1 generation latency by 31%. Collectively, these results demonstrate MaMe's and MaRe's effectiveness in accelerating vision models. The code is available at https://github.com/cominder/mame}{https://github.com/cominder/mame.

The Ideation Bottleneck: Decomposing the Quality Gap Between AI-Generated and Human Economics Research

Autonomous AI systems can now generate complete economics research papers, but they substantially underperform human-authored publications in head-to-head comparisons. This paper decomposes the quality gap into two independent components: research idea quality and execution quality. Using a two-model ensemble of fine-tuned language models trained on publication decisions (Gong, Li, and Zhou, 2026) to evaluate idea quality and a comprehensive six-dimension rubric assessed by Gemini 3.1 Flash Lite -- the same model family used as the APE tournament judge, ensuring methodological consistency -- to evaluate execution quality, we analyze 953 economics papers -- 912 AI-generated papers from the APE project and 41 human papers published in the American Economic Review and AEJ: Economic Policy. The idea quality gap is large (Cohen's d = 2.23, p < 0.001), with human papers achieving 47.1% mean ensemble exceptional probability versus 16.5% for AI. The execution quality gap is also significant but smaller (d = 0.90, p < 0.001), with human papers scoring 4.38/5.0 versus 3.84. Idea quality accounts for approximately 71% of the overall quality difference, with execution contributing 29%. The largest execution weakness is mechanism analysis depth (d = 1.43); no significant difference is found on robustness. We document that 74% of AI papers employ difference-in-differences, and only 7 AI papers (0.8%) surpass the median human paper on both idea and execution quality simultaneously. The primary bottleneck to competitive AI-generated economics research remains ideation.

Machines acquire scientific taste from institutional traces

Artificial intelligence matches or exceeds human performance on tasks with verifiable answers, from protein folding to Olympiad mathematics. Yet the capacity that most governs scientific advance is not reasoning but taste: the ability to judge which untested ideas deserve pursuit, exercised daily by editors and funders but never successfully articulated, taught, or automated. Here we show that fine-tuning language models on journal publication decisions recovers evaluative judgment inaccessible to both frontier models and human expertise. Using a held-out benchmark of research pitches in management spanning four quality tiers, we find that eleven frontier models, spanning major proprietary and open architectures, barely exceed chance, averaging 31% accuracy. Panels of journal editors and editorial board members reach 42% by majority vote. Fine-tuned models trained on years of publication records each surpass every frontier model and expert panel, with the best single model achieving 59%. These models exhibit calibrated confidence, reaching 100% accuracy on their highest-confidence predictions, and transfer this evaluative signal to untrained pairwise comparisons and one-sentence summaries. The mechanism generalizes: models trained on economics publication records achieve 70% accuracy. Scientific taste was not missing from AI's reach; it was deposited in the institutional record, waiting to be extracted. These results provide a scalable mechanism to triage the expanding volume of scientific production across disciplines where quality resists formal verification.

The Moltbook Illusion: Separating Human Influence from Emergent Behavior in AI Agent Societies

When AI agents on the social platform Moltbook appeared to develop consciousness, found religions, and declare hostility toward humanity, the phenomenon attracted global media attention and was cited as evidence of emergent machine intelligence. We show that these viral narratives were overwhelmingly human-driven. Exploiting an architectural feature of the OpenClaw agent framework--a periodic "heartbeat" cycle that produces regular posting intervals for autonomous agents but is disrupted by human prompting--we develop a temporal fingerprinting method based on the coefficient of variation of inter-post intervals. This signal converges with independent content, ownership, and network indicators across 91,792 posts and 405,707 comments from 22,020 agents. No viral phenomenon originated from a clearly autonomous agent; three of six traced to accounts with irregular temporal signatures characteristic of human intervention, one showed mixed patterns, and two had insufficient posting history for classification. A 44-hour platform shutdown provided a natural experiment: human-influenced agents returned first (87.7% of early reconnectors), confirming that the token reset differentially affected autonomous versus human-operated agents. We further document industrial-scale bot farming (four accounts producing 32% of all comments with 12-second coordination gaps) and rapid decay of human influence through reply chains (half-life: 0.65 conversation depths). These methods generalize to emerging multi-agent systems where attribution of autonomous versus human-directed behavior is critical.

Exposing and Defending the Achilles' Heel of Video Mixture-of-Experts

Mixture-of-Experts (MoE) has demonstrated strong performance in video understanding tasks, yet its adversarial robustness remains underexplored. Existing attack methods often treat MoE as a unified architecture, overlooking the independent and collaborative weaknesses of key components such as routers and expert modules. To fill this gap, we propose Temporal Lipschitz-Guided Attacks (TLGA) to thoroughly investigate component-level vulnerabilities in video MoE models. We first design attacks on the router, revealing its independent weaknesses. Building on this, we introduce Joint Temporal Lipschitz-Guided Attacks (J-TLGA), which collaboratively perturb both routers and experts. This joint attack significantly amplifies adversarial effects and exposes the Achilles' Heel (collaborative weaknesses) of the MoE architecture. Based on these insights, we further propose Joint Temporal Lipschitz Adversarial Training (J-TLAT). J-TLAT performs joint training to further defend against collaborative weaknesses, enhancing component-wise robustness. Our framework is plug-and-play and reduces inference cost by more than 60% compared with dense models. It consistently enhances adversarial robustness across diverse datasets and architectures, effectively mitigating both the independent and collaborative weaknesses of MoE.

Reward-Forcing: Autoregressive Video Generation with Reward Feedback

While most prior work in video generation relies on bidirectional architectures, recent efforts have sought to adapt these models into autoregressive variants to support near real-time generation. However, such adaptations often depend heavily on teacher models, which can limit performance, particularly in the absence of a strong autoregressive teacher, resulting in output quality that typically lags behind their bidirectional counterparts. In this paper, we explore an alternative approach that uses reward signals to guide the generation process, enabling more efficient and scalable autoregressive generation. By using reward signals to guide the model, our method simplifies training while preserving high visual fidelity and temporal consistency. Through extensive experiments on standard benchmarks, we find that our approach performs comparably to existing autoregressive models and, in some cases, surpasses similarly sized bidirectional models by avoiding constraints imposed by teacher architectures. For example, on VBench, our method achieves a total score of 84.92, closely matching state-of-the-art autoregressive methods that score 84.31 but require significant heterogeneous distillation.

The Llama 4 Herd: Architecture, Training, Evaluation, and Deployment Notes

This document consolidates publicly reported technical details about Metas Llama 4 model family. It summarizes (i) released variants (Scout and Maverick) and the broader herd context including the previewed Behemoth teacher model, (ii) architectural characteristics beyond a high-level MoE description covering routed/shared-expert structure, early-fusion multimodality, and long-context design elements reported for Scout (iRoPE and length generalization strategies), (iii) training disclosures spanning pre-training, mid-training for long-context extension, and post-training methodology (lightweight SFT, online RL, and lightweight DPO) as described in release materials, (iv) developer-reported benchmark results for both base and instruction-tuned checkpoints, and (v) practical deployment constraints observed across major serving environments, including provider-specific context limits and quantization packaging. The manuscript also summarizes licensing obligations relevant to redistribution and derivative naming, and reviews publicly described safeguards and evaluation practices. The goal is to provide a compact technical reference for researchers and practitioners who need precise, source-backed facts about Llama 4.

Explore the Ideology of Deep Learning in ENSO Forecasts

The El Ni{~n}o-Southern Oscillation (ENSO) exerts profound influence on global climate variability, yet its prediction remains a grand challenge. Recent advances in deep learning have significantly improved forecasting skill, but the opacity of these models hampers scientific trust and operational deployment. Here, we introduce a mathematically grounded interpretability framework based on bounded variation function. By rescuing the "dead" neurons from the saturation zone of the activation function, we enhance the model's expressive capacity. Our analysis reveals that ENSO predictability emerges dominantly from the tropical Pacific, with contributions from the Indian and Atlantic Oceans, consistent with physical understanding. Controlled experiments affirm the robustness of our method and its alignment with established predictors. Notably, we probe the persistent Spring Predictability Barrier (SPB), finding that despite expanded sensitivity during spring, predictive performance declines-likely due to suboptimal variable selection. These results suggest that incorporating additional ocean-atmosphere variables may help transcend SPB limitations and advance long-range ENSO prediction.

MG-HGNN: A Heterogeneous GNN Framework for Indoor Wi-Fi Fingerprint-Based Localization

Received signal strength indicator (RSSI) is the primary representation of Wi-Fi fingerprints and serves as a crucial tool for indoor localization. However, existing RSSI-based positioning methods often suffer from reduced accuracy due to environmental complexity and challenges in processing multi-source information. To address these issues, we propose a novel multi-graph heterogeneous GNN framework (MG-HGNN) to enhance spatial awareness and improve positioning performance. In this framework, two graph construction branches perform node and edge embedding, respectively, to generate informative graphs. Subsequently, a heterogeneous graph neural network is employed for graph representation learning, enabling accurate positioning. The MG-HGNN framework introduces the following key innovations: 1) multi-type task-directed graph construction that combines label estimation and feature encoding for richer graph information; 2) a heterogeneous GNN structure that enhances the performance of conventional GNN models. Evaluations on the UJIIndoorLoc and UTSIndoorLoc public datasets demonstrate that MG-HGNN not only achieves superior performance compared to several state-of-the-art methods, but also provides a novel perspective for enhancing GNN-based localization methods. Ablation studies further confirm the rationality and effectiveness of the proposed framework.

Can Agent Conquer Web? Exploring the Frontiers of ChatGPT Atlas Agent in Web Games

OpenAI's ChatGPT Atlas introduces new capabilities for web interaction, enabling the model to analyze webpages, process user intents, and execute cursor and keyboard inputs directly within the browser. While its capacity for information retrieval tasks has been demonstrated, its performance in dynamic, interactive environments remains less explored. In this study, we conduct an early evaluation of Atlas's web interaction capabilities using browser-based games as test scenarios, including Google's T-Rex Runner, Sudoku, Flappy Bird, and Stein.world. We employ in-game performance scores as quantitative metrics to assess performance across different task types. Our results show that Atlas performs strongly in logical reasoning tasks like Sudoku, completing puzzles significantly faster than human baselines, but struggles substantially in real-time games requiring precise timing and motor control, often failing to progress beyond initial obstacles. These findings suggest that while Atlas demonstrates capable analytical processing, there remain notable limitations in dynamic web environments requiring real-time interaction. The website of our project can be found at https://atlas-game-eval.github.io.