Search signals, briefings, benchmarks and glossary terms.
Search signals, briefings, benchmarks and glossary terms.
Use this view to inspect the underlying evidence corpus. For ranked developments, decision posture and interpretation, use Signals.
Raw feed or Signals?
Raw feed is chronological evidence. Signals ranks and interprets material change.
DC-Leap proposes a training-free framework to accelerate Diffusion Large Language Models (dLLMs) by addressing conservative confidence thresholds.
New research introduces DFAH-Bench, a replay benchmark designed to measure observable behavioral instability in financial agent decision-making processes.
GaugeQuant proposes a method to learn quantization-optimal bases for transformers during training by breaking internal symmetries, reducing activation outliers.
SonicSampler presents unified tile-aware kernels for LLM sampling and speculative verification, improving efficiency for dynamic serving workloads.
Research introduces an attention-based experience replay framework for continual learning in time series forecasting models, addressing data distribution shifts.
Research finds LLMs struggle with dynamic user intent in iterative tasks, as current evaluation focuses on single-turn, fully-specified settings.
A research paper proposes a knowledge-injection framework to improve zero-shot delirium prediction using LLMs and electronic health records.
Research details higher-order optimizers like Muon and SOAP for faster LLM pretraining convergence, addressing stability issues at scale.
Researchers developed a neural-symbolic pipeline combining compact transformers with a linguistically-informed mediator for gaming toxicity detection.
Research explores the scaling behavior and generalization of deep parameterized quantum circuits (PQCs), a key challenge in quantum machine learning.
Research explores how architectural conditions in active inference agents lead to 'causal emergence,' linking predictive organization to information theory.
Domyn-Small, a 10B parameter multilingual open-weight reasoning LLM developed in Europe, achieves 32K context window via continued pre-training.
Research finds combining regex filters with LLM alignment provides no measurable safety coverage gain against adversarial input, even when designed to bypass regex.
Research proposes a Fisher-market formulation to allocate fractional credit between planned tasks and performed actions, moving beyond all-or-nothing links.
Researchers propose Graph Wavelet Compressed Sensing (GWCS) for efficient, offline compression of graph signals to reduce data and training costs.
Researchers propose "JITAI-Twins," digital twins of subpopulations using diffusion models for mobile health intervention algorithm testing.
Research proposes Adaptive Multi-Horizon Reinforcement Learning, allowing AI agents to flexibly adapt temporal discounting for decision-making.
Research presents a method for learning safe, high-dimensional feedback control using control barrier functions (CBFs) to enforce safety constraints.
Research introduces 'Parity Bottleneck Layers' to enable interpretable transformer models by addressing memory and compute costs of per-layer bottlenecks.
Research explores offline goal-conditioned reinforcement learning with hierarchical action chunking to address long-horizon task challenges.
Research paper proposes Context-weighted Discrete Flow Matching to improve generative modeling on discrete structures by addressing varying token difficulty.
Research explores explanation-based runtime verification for trustworthy ML models in optical network automation, mitigating risks from incorrect decisions.
Research shows Bayesian uncertainty estimation (Monte Carlo dropout) improves confidence calibration and error detection in medical image AI models.
Research identifies 'Spectral Drift'—frequency-domain instability in internal activations—as a signature of neural network misclassifications.
A research paper introduces STeMP, a spatio-temporal modeling protocol to address data sensitivity and methodological choices in ML model quality estimation.
Research investigates when weight-tied looped transformers implement algorithms, finding a linear computation frontier dictated by training budget.
Research finds sparse autoencoder (SAE) features' causal roles vary across SAE families and layer depths, challenging their stability for LLM interpretation.
Research shows LLMs hallucinate more when forced to fill structured forms like JSON or function arguments, even when the input lacks an answer.
Research paper introduces Fisher widths to measure local parameter fluctuations and provide anisotropic recovery in statistical manifolds.
Research introduces heat-kernel entropy profiles for weighted measures on manifolds to account for geometric support in effective sample size.
Research introduces Pairwise Quantile Regression, extending conditional distribution modeling for high-dispersion variables beyond standard least squares.
Research proposes using neural predicates to formalize and scale view generation in the Black-Litterman model for portfolio construction, reducing subjectivity.
New benchmark, DataPrep-Bench, introduced to evaluate LLMs as end-to-end training data preparators, focusing on data construction and quality evaluation.
Researchers released GLAN-QnA-KR, a 303K-row Korean instruction-QA corpus generated by Microsoft's Phi-3.5-MoE-instruct model.
Research proposes analytic federated learning for medical image classification, enabling collaboration across institutions with task-heterogeneous data in one round.
Researchers released Naver-News-KO, a 27,400-pair Korean news summarization dataset from Naver News, for open-source model fine-tuning.
Researchers propose CEDAR, a constraint-based method for lagged causal edge discovery in sparse autoregressive time series data.
Research proposes Minimum Bayes Risk decoding for Error Span Detection in automatic machine translation evaluation, improving error localization.
Research explores compiling discrete programming structures, like conditionals and iteration, into differentiable recurrent neural networks.
Research proposes sparse autoencoders (SAEs) to create interpretable embeddings for analyzing large text corpora, addressing limitations of LLM-based and dense embedding methods.
Research presents enhanced Membership Inference Attacks (MIAs) on diffusion models using a frequency-domain approach, improving data privacy breach detection.
Researchers introduced ER-JEPA, a hierarchical self-supervised learning framework for multivariate time series, demonstrated in ECG analysis.
Research explores parameter-free RDB encoders for foundation models to predict missing values across diverse enterprise tasks without retraining.
Research suggests long-horizon time-series forecasting benchmarks may not require complex transformer architectures, finding simpler linear models competitive.
Research explores using Quantum Kitchen Sinks (QKS) for RF spectrogram anomaly detection to secure wireless spectrum management.
Research introduces Exo-MDPs, a structured class of Markov Decision Processes for sample-efficient reinforcement learning in operations research.
Gumbel Distillation is a new technique for parallel text generation, improving non-autoregressive model quality to rival sequential models.
Research characterizes the VC dimension and sample complexity of Transformers, providing tight bounds for depth-L models with W parameters and sequence length T.
Research proposes a category theory approach using Kan extensions to define and compute structural invariants for transfer learning between tasks.
Research introduces 'swarm-attack,' an open-source adversarial framework using coordinating LLM agents to bypass safety and discover software vulnerabilities.
© 2026 OneBench: AI Insights. All rights reserved.
Evidence before opinion