d-OPSD is the first on-policy self-distillation framework designed for diffusion LLMs. It uses self-generated answers as suffix conditioning and step-level supervision, enabling efficient post-training with only about 10% of RLVR's optimization steps while outperforming RLVR and SFT baselines on four reasoning benchmarks.
RubricsTree introduces a hierarchical taxonomy of over 100 clinically-verifiable Boolean rubrics, evolved from 4,000 real user queries via human-in-the-loop curation. It enables scalable, expert-aligned evaluation of personal health agents by dynamically routing queries to relevant rubrics and outperforms baseline methods in alignment, context sensitivity, and model performance gains of up to 66% on HealthBench.
Looped World Models (LoopWM) introduce a looped architecture that iteratively refines latent environment states using a parameter-shared transformer. This approach achieves up to 100x parameter efficiency over conventional world models by adapting computation depth to each prediction step, offering a new scaling dimension for world simulation.
Zone of Proximal Policy Optimization (ZPPO) integrates teacher knowledge directly into prompts rather than policy gradients. It uses Binary and Negative Candidate-included Questions to surface student failure modes and amplifies learning through a prompt replay buffer, achieving superior performance on hard questions across student scales, especially at smaller model sizes.
Darshana Graph presents a corpus of over 125,000 text records from Hindu, Buddhist, and Jain philosophical sources. It includes a unique subset of 8,500 aligned records from 18 commentators across five schools, enabling cross-commentator comparison. The corpus supports stylometric analysis and a large language model pipeline that extracts philosophical concept relationships, revealing disagreement patterns and extraction limitations.
ReLAR introduces a reinforcement-guided framework that iteratively refines hidden states to improve LLM reasoning stability. It uses learned depth and action controllers trained via policy gradients to adaptively determine refinement steps, achieving better accuracy and generation quality with lower inference overhead than explicit reasoning methods.
A domain-validity rubric screens candidate metamorphic relations by ensuring tolerance exceeds numerical floor and preconditions are met. The method transforms valid relations into executable, oracle-free test assets, validated across multiple CFD tasks and PDE families, distinguishing model violations from out-of-domain applications.
A new method integrates persistent homology into non-negative matrix factorisation to regularise the topology of basis functions. This approach enables spatially coherent image components, periodic time-series, and clique-like graph signals by using threshold-free topological scores as regularisers in the NMF objective.
Offline evaluation of agentic systems often produces tied comparisons in 75% of cases using standard success-based metrics. Preference-based trajectory evaluation reduces ties to 35% by comparing progress and time-to-return profiles, enhancing discriminative power and data efficiency. These results suggest benchmark saturation may stem from evaluation method choice, not just data or problem difficulty.
CARLOS uses an aggregate deep neural network to learn a joint space-time exercise boundary for optimal stopping problems. It progressively refines stopping decisions at finer time resolutions and employs adaptive sampling to focus training near the stopping boundary. Benchmarked results show CARLOS outperforms existing Bermudan solvers, approaching the American upper bound with high efficiency.
Reversal Q-Learning (RQL) is a new off-policy reinforcement learning algorithm that trains a flow policy using prior data. By modeling flow refinement steps as actions in an expanded Markov decision process and applying virtual on-policy trajectories via reversal, RQL enables effective offline learning without backpropagation through time. Experiments on 50 robotic tasks show RQL achieves the best average performance among state-of-the-art flow-based offline RL methods.
SpatioTemporal Causal Network Diagnostics (ST-CND) introduces a data-driven framework to detect geographic tipping points by modeling spatial fields as time-evolving causal networks. It outperforms existing methods on sea-surface temperature benchmarks, achieving an AUROC of 0.783 and a critical-subnetwork IoU of 0.378 for the North Atlantic AMOC.
A new method called Credit-in-Event identifies and addresses temporal credit dilution in learned dynamics models. CREST, a label-free and training-free readout, re-anchors pooled representations by estimating transient event cores and applying event-versus-rest contrast, reducing out-of-distribution error across multiple systems and data types. Ablations confirm the improvement stems from event-core credit re-anchoring, not generic locality or stability priors.
Concatenating LLM-generated features to graph neural networks systematically reduces accuracy on homophilous benchmarks, with PubMed accuracy dropping by -17.0 +/- 0.3 pp. A measure of LLM-alone discriminability, Delta_sig, correlates strongly with concatenation performance (r^2 = 0.38), and a rule based on Delta_sig <= 13.8 pp correctly predicts non-positive impact in 7 out of 9 datasets.
SelFix improves fixed-point inversion by selecting solutions that produce straighter inverse trajectories, enhancing real-image reconstruction and source-preserving editing. Experiments on FLUX.1-dev and PIE-Bench show it outperforms prior baselines in both reconstruction quality and editing fidelity.
SkillMigrator learns reusable web skills by matching layout structures instead of element references. It stores each skill as a transferable interaction pattern with a structural sketch, enabling efficient skill transfer across sites. Compared to state-of-the-art methods, it reduces average LLM-action counts by 8-10% on WebArena and Mind2Web at matched success rates.
A new framework decomposes pre-hoc fine-tuning prediction risk into intrinsic limits and optimization variance. It proves a necessary lower bound on variance decay and introduces a budget-optimal probing strategy, validated across synthetic and real-world benchmarks through three distinct prediction regimes.
A study enhances physics-constrained neural networks by introducing an upgraded numerical solver, a unified autoregressive block, and two neural backbones. These improvements reduce root mean squared error by 8-22% in short-term forecasts over the South Pacific and better preserve physical consistency.
TUNEAHEAD is a lightweight framework that predicts fine-tuning performance using meta-feature vectors from dataset descriptors and short probe runs. It outperforms baselines like Early-Stop Extrapolation and ProxyLM, achieving an RMSE of 1.47 percentage points and 95.1% of predictions within ±3 percentage points of true scores on 370 held-out runs.
We propose learnable graph patches as the smallest semantic units in graph data to address feature heterogeneity without textual information. Our framework uses patch encoders and aggregators to extract and combine knowledge across domains, enabling universal pre-training and improved downstream performance with more pre-training data.