cs.LGOct 28, 2026

Automatic Construction of Strong Simple Baselines

Authors: Caleb N. Ellington, Jiaqi Wang, Sohan Addagudi, Elijah Cole

Organizations: Carnegie Mellon University · Sunday Scientific · University of Washington

Abstract

To rigorously establish the value of a novel deep learning method, it must be compared against appropriate baselines. As deep learning has become increasingly common in the physical and life sciences, there have been numerous cases in which apparent progress has turned out to be an artifact of omitted or missing baselines. In this work, we investigate whether this problem can be mitigated by using large language models to automate the construction of baselines. We assemble 36 supervised classification and regression tasks from peer-reviewed studies in Nature, Science, and Cell-family journals that investigate novel deep learning methods for scientific applications. We then use a language model to construct simple baselines for each task. These automatically constructed baselines are strong, recovering a median of 95.5% (mean 85.7%) of the gap between a trivial predictor and the score of the deep learning model. The median simple baseline matches or exceeds the performance of the deep learning model in 12 studies. We further examine the variation across scientific domains and publication venues and we characterize the effects of test-time compute and access to different levels of study context. As human peer review has struggled to keep up with the rapidly increasing volume of research articles in recent years, our results suggest that automated baseline construction is a practical and scalable check on deep learning contributions in the physical and life sciences.

Figures & tables

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

May 28, 2026

SoundnessBench: Can Your AI Scientist Really Tell Good Research Ideas from Bad Ones?

Autonomous AI research agents aim to speed up scientific discovery by automating the research pipeline from hypothesis generation to peer review. However, existing benchmarks rarely test a fundamental bottleneck whether LLMs can judge the methodological viability of a research idea before\textbf{before} investing time and computational resources. Without a reliable ``first-gate'' filter, autonomous agents risk scaling flawed methodology rather than accelerating meaningful science. We introduce SoundnessBench\textbf{SoundnessBench}, a curated benchmark of 1,099 real research proposals, with source verified extraction, reviewer-derived proxy labels, and proposal-only expert judgments on a subset of 327 proposals. Across 12 frontier LLMs, our evaluation reveals a pervasive optimism bias\textbf{optimism bias}: standard prompting yields a 74.0%74.0\% mean false-positive rate. Aggressive prompting reduces false approvals but substantially increases false rejections. Additional analyses show that optimism persists under controls for public corpus contamination, and surface features, as well as in literature-grounded agent evaluation. These results suggest a limitation in judging scientific soundness, indicating that current LLMs are not yet reliable as standalone gatekeepers for scientific rigor.
May 19, 2026

SciCustom: A Framework for Custom Evaluation of Scientific Capabilities in Large Language Models

Large language models (LLMs) are increasingly applied to scientific research, yet existing evaluations often fail to reflect the fine-grained capabilities required in practice. Most benchmarks are manually curated or domain-generic, limiting scalability and alignment with real scientific use cases. In this paper, we propose a new framework named SciCustom to address the problem. It enables the custom construction of benchmarks from large-scale scientific data to evaluate application-specific scientific capabilities in LLMs. SciCustom first organizes scientific knowledge into ontology-grounded knowledge units with controlled granularity and trains a tagger to map large-scale data instances into this knowledge space. Given a custom requirement, relevant knowledge units are identified via voting-based multi-model consensus. These units enable relevance-aware benchmark retrieval via binary search, followed by proxy subset selection and data-grounded benchmark generation for efficient evaluation. Experiments in chemistry and healthcare demonstrate that SciCustom reveals fine-grained differences in LLM scientific capabilities that standard benchmarks overlook, while requiring neither expert annotation nor synthetic question generation. This work provides a scalable and application-aware foundation for benchmarking scientific capabilities in LLMs. The source code is available at https://github.com/yjwtheonly/SciCustom.
Jun 23, 2026

LLM-Based Scientific Peer Review: Methods, Benchmarks, and Reliability Challenges

The rapid growth of scientific submissions has pushed traditional peer review toward its scalability limits, motivating the exploration of large language models (LLMs) as intelligent automated evaluation assistants. Although recent studies show that LLMs can generate fluent critiques and approximate reviewer scores, their reliability, robustness, and security as decision-support systems remain insufficiently understood. This survey offers a systems-level analysis of LLM-based scientific peer review, focusing on two core evaluative functions: critique generation and score prediction. We present a structured taxonomy of modeling approaches (including prompt-based, supervised, retrieval-augmented, and alignment-optimized approaches), and synthesize empirical findings across existing benchmarks. We analyze dataset constraints, evaluation shortcomings, and domain concentration biases that limit current assessment practices. Beyond performance metrics, we identify emerging robustness risks, including prompt injection, data poisoning, retrieval vulnerabilities, and reward hacking, which expose automated review pipelines to strategic manipulation. From a data mining perspective, we outline key open challenges in modeling subjective disagreement and cross-domain generalization. By reframing automated peer review as a high-stakes, multi-objective decision problem, this survey provides a roadmap for developing robust, transparent, and trustworthy AI-assisted scientific evaluation systems.