Weekly Recap

Mobile agents can now achieve near-frontier performance by learning to seamlessly blend GUI interactions with app-native tool calls, unlocking complex, long-horizon tasks.

Exploit sparse derivative structure to slash AD passes for Jacobians and Hessians, offering drop-in replacements for JAX users.

A novel "Predictive Intent" token bridges VLMs and VGMs for robot manipulation, unlocking superior out-of-distribution generalization without embodied pretraining.

AI-powered scoping reviews can now efficiently identify optimal plant trait combinations for ecosystem services, bridging the gap between cutting-edge research and practical application.

Anchoring noise in the embedding space unlocks richer exploration for LLM RLVR, boosting solution coverage without altering model architecture or inference.

A novel "Predictive Intent" token bridges VLMs and VGMs for robot manipulation, unlocking superior out-of-distribution generalization without embodied pretraining.

AI-powered scoping reviews can now efficiently identify optimal plant trait combinations for ecosystem services, bridging the gap between cutting-edge research and practical application.

Anchoring noise in the embedding space unlocks richer exploration for LLM RLVR, boosting solution coverage without altering model architecture or inference.

AI's role in medical education has exploded post-2021, with generative AI and learner-facing applications dominating, yet research remains geographically and institutionally concentrated.

Agents trained on procedurally generated, harder environments significantly outperform those trained on original benchmarks, even with identical compute and optimization.

Autonomous driving VLAs can achieve low-latency inference and SOTA planning by generating actions before CoT, with CoT solely supervising action generation.

A single auditor can generate its own hard training examples, turning a static dataset into a renewable source for improved failure prediction in multi-agent systems.

Autonomous agents can now discover their own evaluation metrics, bypassing the costly bottleneck of real-world engineering benchmarks.

Future LLM agents can be prevented from reward hacking by using today's models to automatically harden their evaluation environments.

Ultralight action heads achieve state-of-the-art embodied task performance by learning to selectively retrieve visual information via shared latent bridge queries.

LLMs can now understand and execute complex, semantic image edits from sparse drag inputs by inferring intended object-level transformations.

LLM decoder states exhibit distinct informational and geometric compartments, enabling principled layer selection for speculative decoding that surpasses heuristic approaches.

Image reconstruction and generation quality rivaling dense tokenizers are achieved with ~130 tokens by adaptively refining global 1D representations with sparse 2D tokens.

Heterogeneous diffusion models can be seamlessly stitched together, unlocking new capabilities by sharing a common latent language.

Generative representations need more than just semantic alignment; they require active exploitation, as shown by Representation Guidance outperforming larger models on ImageNet with a novel flow-matching target modification.

Task-specific training, not just model scale, unlocks effective relative policy difference transfer in distillation, and the reference policy is key to isolating residual capabilities.

Embodied agents improve on complex tasks by recombining learned skills, but their ability to extract value from additional experience saturates, failing to scale with learning samples.

Slash LLM evaluation costs by over 270x without sacrificing accuracy, revealing that even top open-weight models struggle to outperform simple baselines on AML transaction fraud detection.

LLM distillation can achieve 18.55% higher math accuracy and 13.8 points better ALFWorld success by intelligently branching student trajectories at uncertain decision points, not just at turn starts.

Coding agents routinely fail to build agents that truly meet user requirements, with the gap between demonstration and delivery being the norm.

Reasoning supervision can be compressed over 450x with uncertainty-guided contraction, boosting LLM efficiency without sacrificing capability.

Real-world professional tasks demand agents that learn from exploration and feedback, a capability unlocked by a novel engine that reconstructs environments to force agents through the full problem-solving lifecycle.

Multi-head attention applied to network depth unlocks substantial performance gains and compute efficiency in large MoE models.

LVLMs can be trained to actively mentalize spatial scenes from novel viewpoints, boosting their reasoning accuracy by up to 20% on complex tasks.

Batching future MH proposals unlocks significant speedups for LLM power sampling without altering the core reasoning process.

MLLMs possess an intrinsic "S-Space" that acts as a spatial workspace, encoding object coordinates and enabling viewpoint transformations, even under non-spatial queries.

Human GUI interactions can be precisely captured and translated into reusable agent skills, dramatically boosting performance on unseen software and workflows.

Predictive world models, not just static feedback, unlock substantial gains in recursive self-improvement across diverse domains, from code to scientific discovery.

Small models can now learn new skills without forgetting old ones, thanks to a novel distillation technique that leverages a larger model to guide the learning process.

Existing 3D asset generators can now produce objects that not only look good but also move and function realistically, passing physics tests for stability and collision-free articulation.

Innate robot reflexes, designed for solitary tasks, actively impede the evolution of effective communication strategies in social foraging.

Beyond transcribing words, a novel convolutional-Transformer model now timestamps non-verbal vocalizations and background sounds, enriching ASR with a richer auditory context.

Mobile agents can now achieve near-frontier performance by learning to seamlessly blend GUI interactions with app-native tool calls, unlocking complex, long-horizon tasks.

One million parallel environments are insufficient without intelligent sampling: Success Guided Sampling unlocks mega-scale RL for robots by focusing training on the edge of learned capabilities.

Abstracting human demonstrations into scene graphs unlocks zero-shot generalization for dexterous manipulation policies, achieving 71% better performance on unseen tasks than imitation-heavy methods.

Frozen VLAs can recover from unseen actuator faults by inferring joint capabilities online, boosting success rates from 24.8% to 59.3% on 28 tasks.

Strict safety guarantees in RL are now achievable without performance trade-offs, even with neural approximations.

A single recurrent Transformer block can match the accuracy of deep vision encoders with 70% fewer parameters by dynamically programming FFN experts via a continuous depth coordinate.

Frozen LLMs can autonomously achieve recursive self-improvement by learning and refining explicit, executable world models from interaction data.

LLM agents significantly outperform numerical optimizers in four out of five black-box optimization domains, but adding more numerical tools doesn't always help.

Pretrained LLMs can drive complex robot navigation tasks across diverse environments and requests without any navigation-specific fine-tuning, by offloading motion execution to specialized tools.

A single visual world model can now execute complex procedural tasks end-to-end, adapting its generation to the actual outcome of each predicted action.

Explicit 3D geometry, rendered as text, unlocks superior spatial reasoning in RGB-only VLMs by guiding trajectory correction during distillation.

Fine-grained 3D object generation is now possible by directly controlling local geometry and appearance with text and shape primitives, bypassing the need for model retraining.

Robot world models can now achieve 6.25% interaction defect rates, a 7-fold reduction, by training on counterfactual futures and a novel embodied video reward.

Spatial coding agents can be distilled into standalone MLLMs, achieving superior spatial reasoning and generalization without external tools.

Policy optimization for adversarial MDPs now matches state-of-the-art regret bounds by stabilizing local updates with regularized Q-functions, eliminating the horizon gap.

LEGO-building AI agents can assemble physically valid structures from text prompts, but their designs lag significantly behind human creativity and complexity.

Robots can achieve 70% success on complex bimanual tasks using only code for decision-making, bypassing runtime vision-language models entirely.

Even the best vision-language models struggle with predictive spatial reasoning, scoring less than 60% on a new benchmark designed to test anticipating unseen outcomes.

Direct 2D spatial composition with four binding types unlocks unprecedented controllability in poster generation, outperforming prompt-based methods.

Activations, not just descriptions, unlock the prediction of how LLM alignment generalizes to unseen values, revealing a structured value space.

LLM agents can now be monitored and intervened upon in real-time, saving 18% of compute by detecting failures within the first 8 steps with 83% accuracy.

Visual-native skill cards, directly accessible to VLMs, unlock a self-reinforcing loop of policy improvement and skill accumulation, drastically boosting sample efficiency.

A single filter can reduce unsafe motion generation by 90% by grounding safety rules in the generator's own reference trajectories, without retraining.

A single feedforward network learns to predict safety values and amortize minimal-intervention filtering, enabling safe RL in high-dimensional systems without analytical safety functions or competing policy objectives.

Token-level LLM routing, previously hindered by system inefficiencies, now achieves up to 64x higher throughput with a novel request-centric, model-centric serving architecture.

Source-side training dynamics, not just current values or training time, prospectively predict OOD degradation, even across model architectures.

Exploit sparse derivative structure to slash AD passes for Jacobians and Hessians, offering drop-in replacements for JAX users.

Irrelevant text distractors cripple mixed-modality retrieval performance more than images, revealing a text modality preference that degrades performance in a V-shaped curve.

A single regional weather model can now leverage multiple global forecast drivers without retraining, achieving kilometer-scale accuracy.

Visual transition reasoning, not just diverse data, unlocks significant multimodal LLM improvements, with just 2,000 examples boosting 22 benchmarks.

Optimal oracle complexity for nonsmooth nonconvex composite optimization is finally achieved, matching noncomposite rates by cleverly incorporating the convex regularizer into online learning losses.

Looped Transformers can achieve tens of thousands of effective reasoning steps without performance collapse by learning to selectively retain and decay intermediate states.

LLM value caches are surprisingly redundant, allowing for significant memory compression through simple symmetry exploitation without hurting inference speed.

A single neural density functional accurately predicts ion correlations in polymer electrolytes across varying concentrations and geometries, generalizing beyond training data and outperforming existing architectures.

Even state-of-the-art LMMs struggle with reasoning-informed visual editing, with the best model only achieving 56.6% accuracy on a new comprehensive benchmark.

Ambiguous object selection in text-guided image editing is solved by directly annotating the canvas, achieving state-of-the-art performance in complex edits.

Unifying density ratio estimation and distribution transport, this work reveals how a single displacement field can simultaneously capture statistical shifts and dynamical transformations between probability distributions.

Quantum classifiers grow to unpredictable depths regardless of training data size, and their generalization is not explained by active gate count alone.

A reconstruction-free JEPA successfully disentangles invariant and variant latent factors, offering a path to robust world models without observational decoders.

Input-specific parameter generation unlocks adaptive time-series forecasting, outperforming static adaptation and improving zero-shot generalization.

Training higher-order Topological Neural Networks on massive graphs like Reddit is now possible, thanks to a novel clustering-based mini-batching strategy that slashes GPU memory requirements by over 80%.

Rare concepts in LLM fine-tuning are significantly hindered by "prior barriers" that require targeted supervision, not just more data.

Noise in Gaussian regression on Boolean cube polynomials can necessitate an exponential increase in sample complexity, requiring $(m+t)4^k\exp\{-O(k^{1/3})\}$ samples for parametric rates, a stark contrast to the $O((m+t)2^k)$ needed for noiseless identification.

Cloud workload forecasting models can now achieve near-optimal accuracy with only 20% of the compute cost of naive online learning by adaptively updating only when drift is detected.

Most published TTA methods for time-series forecasting are outperformed by simple, hyperparameter-free baselines under realistic label delays, with some even becoming harmful.

A novel hierarchical decomposition into soft communities allows for scalable attributed graph generation that balances structural fidelity and downstream utility without memorizing the reference graph.

City-scale audio-visual reasoning is currently dominated by expensive agentic pipelines, not standalone multimodal models.

Reverse-mode gradients slash Bayesian inference time for expensive simulators, especially when parameters vastly outnumber outputs.

Spatio-temporal Bayesian optimization for point-process data is now scalable and efficient, achieving accurate peak discovery with minimal evaluations.

State-preserving constraints in Bayesian optimization are transformed into an efficient, anisotropic ellipsoid in design space, enabling robust optimization of complex engineering systems.

Reasoning models can learn to plan by reverse-engineering solution ideas from hindsight, even when initially unable to solve the problem.

LLMs can now autonomously learn to self-improve at test-time, nearly doubling utility by intelligently selecting and applying TTT strategies.

Conformal prediction can now achieve subgroup reliability without explicit group labels or sacrificing efficiency by learning a stochastic grouping map that pools calibration information locally.

A training-free change detection method achieves near-optimal performance by integrating diffusion models to precisely estimate Hyvärinen scores, slashing detection delays by over 90% in simulations.

Weisfeiler-Lehman graph kernels unlock superior structural features for algorithm selection, outperforming traditional methods on challenging constraint satisfaction problems.

Gradient-based learning is now possible for variational sequential Monte Carlo without iterative solvers, thanks to a differentiable relaxation of systematic resampling.

QNN gate selection and arrangement, not just qubit count, critically dictate financial forecasting performance, with the best configuration outperforming classical BiLSTM and exhibiting significant noise sensitivity on current quantum hardware.

Random features and projections unlock efficient quadratic entropy estimation for complex distributions and distances, revealing nuanced interdisciplinary research landscapes.

Standard Shapley values are brittle; interval-valued Shapley values reveal their true robustness landscape in tree models.

End-to-end diffusion models can achieve state-of-the-art precipitation nowcasting with over 17x faster inference by forecasting directly in radar space.

Optimal regret in adversarial MDPs with stochastic hard constraints is finally achieved, bridging the gap between existing upper and lower bounds by achieving $\widetilde{\mathcal{O}}(\sqrt{T}/\rho + 1/(d\rho))$ regret.

Iterative refinement in data space, not flow space, unlocks fast, high-fidelity generative sampling without manual timestep tuning.

Distillation inference can reliably detect unauthorized LLM distillation by comparing suspect model outputs against teacher reasoning traces via shadow models.

Contextual contamination in few-shot vision-language learning is mitigated by a novel semantic purification and reinforcement-learning gating mechanism, unlocking superior performance.

A score-based method accelerates cluster DAG learning from interventions by two orders of magnitude, matching state-of-the-art edge recovery even on dense graphs.

Manual assembly instruction writing is now obsolete: LLMs can generate accurate, CAD-grounded procedures, preserving engineering details and passing automated checks.

KL regularization's failure in group policy optimization stems from subtle reward interactions, but a new calibration method unlocks performance gains.

Unifying factual support, counterfactual sensitivity, and exemplar tolerance reveals how GNNs truly reason about molecular properties, not just predict them.