The Flow has not summarised this story yet — read it at arXiv Computation and Language.
arXiv:2511. 21692v3 Announce Type: replace-cross Abstract: We investigate how well large language models (LLMs) generalize across different task difficulties, a key question for effective data curation and evaluation.
The paper investigates using open‑weight large language models (LLMs) as both judges and simulated examinees to predict medical exam item difficulty from text alone. LLMs provided absolute difficulty ratings, pairwise comparisons, and response‑based signals such as correctness and option probabilities, which were combined with baseline features to improve predictions. The best models achieved Spearman correlations of .447 on USMLE and .494 on PANCE, surpassing prior shared‑task systems and matching proprietary models, demonstrating that LLM signals capture difficulty information beyond traditional text representations.
arXiv:2509. 14704v3 Announce Type: replace Abstract: Benchmark saturation and training-data contamination increasingly obscure whether reported gains in large language models (LLMs) reflect genuine advances in reasoning or familiarity with recurring patterns in benchmark problems.
arXiv:2607. 28634v1 Announce Type: cross Abstract: The estimation of item difficulty plays a key role in both formative assessment and large-scale high-stakes summative assessments.
The paper introduces the Grounded Integration Measure (GIM), a benchmark of 820 expert‑authored problems designed to test models on tasks that integrate multiple cognitive operations such as constraint satisfaction, state tracking, epistemic vigilance, and audience calibration. GIM emphasizes realistic, broadly accessible knowledge rather than specialized expertise, and uses a judge‑aware 2‑parameter logistic IRT model to produce robust ability estimates across 53 model‑thinking‑level configurations. The authors provide a comprehensive leaderboard of 22 models and 47 test configurations, and conduct an extensive study on how test‑time compute affects model capability, finding that configuration choices like thinking budget and quantization can be as influential as model selection itself. whyItMatters":"By focusing on integration of multiple cognitive domains, GIM offers a more realistic assessment of model reasoning capabilities than benchmarks that either overemphasize memorization or abstract reasoning alone."
The paper introduces Balance of Benchmarks (BoB), a framework that improves task-conditioned model comparison by weighting benchmark evidence based on semantic density, equating scores across varying difficulty levels, and pooling task-relevant residuals. BoB retains all eligible benchmark data while adjusting its influence, outperforming uniform averaging on the WildScores dataset with higher Spearman correlation, lower MAE, and better shortlist hit rates. The method also reduces ranking instability when benchmarks are repeated or paraphrased, and lowers retrospective regret in model selection.