Magis-Bench is a new benchmark of 74 magistrate-level legal writing tasks from Brazilian exams where the strongest LLMs reach only 6.97/10, showing judicial reasoning remains difficult for current models.
hub Mixed citations
A New Measure of Rank Correlation.Biometrika1938; 30(1–2): 81–93
Mixed citation behavior. Most common role is background (67%).
16
Pith papers citing it
3,786
external citations · Crossref
Background
67% of classified citations
hub tools
citation-role summary
background 4
method 2
citation-polarity summary
representative citing papers
EstGraph benchmark evaluates LLMs on estimating properties of very large graphs from random-walk samples that fit in context limits.
A large benchmark finds traditional imputation methods for scRNA-seq data generally outperform deep learning ones, but numerical recovery does not reliably improve biological downstream analyses and no method wins across all settings.
Auditing three performance-optimization benchmarks shows reference-patch validity collapses under cross-machine replay, rankings depend on scoring rules, and 384 of 450 replay-valid tasks are already solved by at least one public submission.
Human readers prefer human literary translations over AI-generated ones for immersion and clarity despite finding MT adequate and struggling to identify the source.
FORSS is a new formula-based super-sample framework for power and sample size calculations with win statistics on hierarchical endpoints that incorporates marginal effects and a flexible joint distribution.
Attention sinks arise from variance discrepancy in self-attention value aggregation, amplified by super neurons and first-token dimension disparity, and can be mitigated by head-wise RMSNorm to accelerate pre-training convergence.
EgoDyn-Bench finds a Perception Bottleneck: foundation models hold ego-motion logic in language but misalign it with vision, underperforming geometric baselines until given explicit trajectories.
A Dirichlet-prior Bayesian estimator for model success probability replaces Pass@k, delivering faster-converging and more stable rankings with credible intervals on math benchmarks.
Introduces consensus objective aggregation for meta-optimization of scientific discovery and reports improved scaling and speedup for 3-SAT algorithm discovery using digital MemComputing machines.
APEX is an assumption-free image quality metric using Sliced Wasserstein Distance on CLIP and DINOv2 embeddings that claims superior robustness to degradations and cross-dataset stability.
Statistical multipoint observations indicate ICME shock acceleration efficiency increases with heliocentric distance up to 0.7 AU before decreasing.
The SFR-M_* relation develops a high-mass decline at low redshifts, driven mainly by morphological quenching from internal structure rather than environmental effects on star-forming galaxies.
Kernel ridge regression combined with mRMR feature selection improves prediction of full benchmark scores from question subsets over existing efficient benchmarking techniques.
The thesis identifies theoretical, empirical, and conceptual flaws in offline fairness measures for recommender systems and contributes new evaluation methods and practical guidelines.
Off-the-shelf German NER tools produce divergent toponym sets that lead to distinct country assignments for climate event news, affecting assessments of national prominence in media coverage.
citing papers explorer
-
Magis-Bench: Evaluating LLMs on Magistrate-Level Legal Tasks
Magis-Bench is a new benchmark of 74 magistrate-level legal writing tasks from Brazilian exams where the strongest LLMs reach only 6.97/10, showing judicial reasoning remains difficult for current models.
-
Evaluating LLMs on Large-Scale Graph Property Estimation via Random Walks
EstGraph benchmark evaluates LLMs on estimating properties of very large graphs from random-walk samples that fit in context limits.
-
A Large-Scale Comparative Analysis of Imputation Methods for Single-Cell RNA Sequencing Data
A large benchmark finds traditional imputation methods for scRNA-seq data generally outperform deep learning ones, but numerical recovery does not reliably improve biological downstream analyses and no method wins across all settings.
-
Are Performance-Optimization Benchmarks Reliably Measuring Coding Agents?
Auditing three performance-optimization benchmarks shows reference-patch validity collapses under cross-machine replay, rankings depend on scoring rules, and 384 of 450 replay-valid tasks are already solved by at least one public submission.
-
AI translation of literary texts is "fine", but readers still prefer human translations
Human readers prefer human literary translations over AI-generated ones for immersion and clarity despite finding MT adequate and struggling to identify the source.
-
The FORSS Framework for Sample Size and Power Calculations With Win Statistics for Hierarchical Endpoints
FORSS is a new formula-based super-sample framework for power and sample size calculations with win statistics on hierarchical endpoints that incorporates marginal effects and a flexible joint distribution.
-
The Structural Origin of Attention Sink: Variance Discrepancy, Super Neurons, and Dimension Disparity
Attention sinks arise from variance discrepancy in self-attention value aggregation, amplified by super neurons and first-token dimension disparity, and can be mitigated by head-wise RMSNorm to accelerate pre-training convergence.
-
EgoDyn-Bench: Evaluating Ego-Motion Understanding in Vision-Centric Foundation Models for Autonomous Driving
EgoDyn-Bench finds a Perception Bottleneck: foundation models hold ego-motion logic in language but misalign it with vision, underperforming geometric baselines until given explicit trajectories.
-
Don't Pass@k: A Bayesian Framework for Large Language Model Evaluation
A Dirichlet-prior Bayesian estimator for model success probability replaces Pass@k, delivering faster-converging and more stable rankings with credible intervals on math benchmarks.
-
Scientific discovery as meta-optimization: a combinatorial optimization case study
Introduces consensus objective aggregation for meta-optimization of scientific discovery and reports improved scaling and speedup for 3-SAT algorithm discovery using digital MemComputing machines.
-
APEX: Assumption-free Projection-based Embedding eXamination Metric for Image Quality Assessment
APEX is an assumption-free image quality metric using Sliced Wasserstein Distance on CLIP and DINOv2 embeddings that claims superior robustness to degradations and cross-dataset stability.
-
Radial Dependency of ICME-associated Particle Acceleration Processes: Statistical Multipoint Observations from 2016-2023
Statistical multipoint observations indicate ICME shock acceleration efficiency increases with heliocentric distance up to 0.7 AU before decreasing.
-
The Evolution of the SFR-M_* relation at 0.1<z<4: Environmental and Morphological Dependencies
The SFR-M_* relation develops a high-mass decline at low redshifts, driven mainly by morphological quenching from internal structure rather than environmental effects on star-forming galaxies.
-
Efficient Benchmarking Is Just Feature Selection and Multiple Regression
Kernel ridge regression combined with mRMR feature selection improves prediction of full benchmark scores from question subsets over existing efficient benchmarking techniques.
-
Offline Evaluation Measures of Fairness in Recommender Systems
The thesis identifies theoretical, empirical, and conceptual flaws in offline fairness measures for recommender systems and contributes new evaluation methods and practical guidelines.
-
Geolocating News about Extreme Climate Events: A Comparative Analysis of Off-the-Shelf Tools for Toponym Identification in German
Off-the-shelf German NER tools produce divergent toponym sets that lead to distinct country assignments for climate event news, affecting assessments of national prominence in media coverage.