arXiv subjects

Dan Klein

Publications and source records attributed to Dan Klein.

At least 19 recordsLinked to original sources

arXiv2026

Open-Ended CT Volume Segmentation with Weak Supervision from Language

We introduce a method for training a text-conditioned segmentation model for CT scans, which combines voxel-level supervision with coarse but scalable slice-level supervision from reports. We extract, from a large database of scan-report pairs, descriptions of findings with indices of slices where those findings occur. We then finetune a general-purpose 2D image segmentation model, SAM3, with standard segmentation losses from strongly labeled data and with a slice-level classification loss from the extracted weak supervision. Our results on the ReXGroundingCT dataset illustrate that this strategy improves the segmentation dice score: from an 8% relative gain when there are 1000 fully labeled volumes to 22% when there are 250 fully labeled volumes.

cs.CV↗

arXiv2026

optimize_anything: A Universal API for Optimizing any Text Parameter

Can a single LLM-based optimization system match specialized tools across fundamentally different domains? We show that when optimization problems are formulated as improving a text artifact evaluated by a scoring function, a single AI-based optimization system-supporting single-task search, multi-task search with cross-problem transfer, and generalization to unseen inputs-achieves state-of-the-art results across six diverse tasks. Our system discovers agent architectures that nearly triple Gemini Flash's ARC-AGI accuracy (32.5% to 89.5%), finds scheduling algorithms that cut cloud costs by 40%, generates CUDA kernels where 87% match or beat PyTorch, and outperforms AlphaEvolve's reported circle packing solution (n=26). Ablations across three domains reveal that actionable side information yields faster convergence and substantially higher final scores than score-only feedback, and that multi-task search outperforms independent optimization given equivalent per-problem budget through cross-task transfer, with benefits scaling with the number of related tasks. Together, we show for the first time that text optimization with LLM-based search is a general-purpose problem-solving paradigm, unifying tasks traditionally requiring domain-specific algorithms under a single framework. We open-source optimize\_anything with support for multiple backends as part of the GEPA project at https://github.com/gepa-ai/gepa .

cs.CL↗

arXiv2025

GenSIaC: Toward Security-Aware Infrastructure-as-Code Generation with Large Language Models

In recent years, Infrastructure as Code (IaC) has emerged as a critical approach for managing and provisioning IT infrastructure through code and automation. IaC enables organizations to create scalable and consistent environments, effectively managing servers and development settings. However, the growing complexity of cloud infrastructures has led to an increased risk of misconfigurations and security vulnerabilities in IaC scripts. To address this problem, this paper investigates the potential of Large Language Models (LLMs) in generating security-aware IaC code, avoiding misconfigurations introduced by developers and administrators. While LLMs have made significant progress in natural language processing and code generation, their ability to generate secure IaC scripts remains unclear. This paper addresses two major problems: 1) the lack of understanding of security weaknesses in IaC scripts generated by LLMs, and 2) the absence of techniques for enhancing security in generating IaC code with LLMs. To assess the extent to which LLMs contain security knowledge, we first conduct a comprehensive evaluation of base LLMs in recognizing major IaC security weaknesses during the generation and inspection of IaC code. Then, we propose GenSIaC, an instruction fine-tuning dataset designed to improve LLMs' ability to recognize potential security weaknesses. Leveraging GenSIaC, we fine-tune LLMs and instruct models to generate security-aware IaC code. Our evaluation demonstrates that our models achieve substantially improved performance in recognizing and preventing IaC security misconfigurations, e.g., boosting the F1-score from 0.303 to 0.858. Additionally, we perform ablation studies and explore GenSIaC's generalizability to other LLMs and its cross-language capabilities.

cs.CR↗

arXiv2025

Balancing Quality and Variation: Spam Filtering Distorts Data Label Distributions

For machine learning datasets to accurately represent diverse opinions in a population, they must preserve variation in data labels while filtering out spam or low-quality responses. How can we balance annotator reliability and representation? We empirically evaluate how a range of heuristics for annotator filtering affect the preservation of variation on subjective tasks. We find that these methods, designed for contexts in which variation from a single ground-truth label is considered noise, often remove annotators who disagree instead of spam annotators, introducing suboptimal tradeoffs between accuracy and label diversity. We find that conservative settings for annotator removal (<5%) are best, after which all tested methods increase the mean absolute error from the true average label. We analyze performance on synthetic spam to observe that these methods often assume spam annotators are more random than real spammers tend to be: most spammers are distributionally indistinguishable from real annotators, and the minority that are distinguishable tend to give relatively fixed answers, not random ones. Thus, tasks requiring the preservation of variation reverse the intuition of existing spam filtering methods: spammers tend to be less random than non-spammers, so metrics that assume variation is spam fare worse. These results highlight the need for spam removal methods that account for label diversity.

cs.CL↗

arXiv2025

Composing Policy Gradients and Prompt Optimization for Language Model Programs

Group Relative Policy Optimization (GRPO) has proven to be an effective tool for post-training language models (LMs). However, AI systems are increasingly expressed as modular programs that mix together multiple LM calls with distinct prompt templates and other tools, and it is not clear how practitioners can best leverage online RL algorithms like GRPO to improve these systems. We begin to address this challenge by investigating whether it is possible to effectively instantiate GRPO for arbitrary multi-prompt programs and whether it can work robustly as an off-the-shelf optimizer for LM programs using the same abstractions and constraints typically involved for prompt optimization. Our main variant of multi-module GRPO constructs groups from module-level invocations, and we also consider trajectory-level grouping as another natural instantiation. We find for the first time that GRPO (and its multi-module counterpart) empirically composes well with automatic prompt optimization, and together they improve accuracy by 11% on average across classification, many-hop search, and privacy-preserving delegation tasks against the post-trained LM - with 5% gains against prompt optimization on its own. We open-source multi-module GRPO in the DSPy library at https://dspy.ai .

cs.CL↗

arXiv2025

GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning

Large language models (LLMs) are increasingly adapted to downstream tasks via reinforcement learning (RL) methods like Group Relative Policy Optimization (GRPO), which often require thousands of rollouts to learn new tasks. We argue that the interpretable nature of language often provides a much richer learning medium for LLMs, compared to policy gradients derived from sparse, scalar rewards. To test this, we introduce GEPA (Genetic-Pareto), a prompt optimizer that thoroughly incorporates natural language reflection to learn high-level rules from trial and error. Given any AI system containing one or more LLM prompts, GEPA samples trajectories (e.g., reasoning, tool calls, and tool outputs) and reflects on them in natural language to diagnose problems, propose and test prompt updates, and combine complementary lessons from the Pareto frontier of its own attempts. As a result of GEPA's design, it can often turn even just a few rollouts into a large quality gain. Across six tasks, GEPA outperforms GRPO by 6% on average and by up to 20%, while using up to 35x fewer rollouts. GEPA also outperforms the leading prompt optimizer, MIPROv2, by over 10% (e.g., +12% accuracy on AIME-2025), and demonstrates promising results as an inference-time search strategy for code optimization. We release our code at https://github.com/gepa-ai/gepa .

cs.CL↗

arXiv2025

An ontological lens on attack trees: Toward adequacy and interoperability

Attack Trees (AT) are a popular formalism for security analysis. They are meant to display an attacker's goal decomposed into attack steps needed to achieve it and compute certain security metrics (e.g., attack cost, probability, and damage). ATs offer three important services: (a) conceptual modeling capabilities for representing security risk management scenarios, (b) a qualitative assessment to find root causes and minimal conditions of successful attacks, and (c) quantitative analyses via security metrics computation under formal semantics, such as minimal time and cost among all attacks. Still, the AT language presents limitations due to its lack of ontological foundations, thus compromising associated services. Via an ontological analysis grounded in the Common Ontology of Value and Risk (COVER) -- a reference core ontology based on the Unified Foundational Ontology (UFO) -- we investigate the ontological adequacy of AT and reveal four significant shortcomings: (1) ambiguous syntactical terms that can be interpreted in various ways; (2) ontological deficit concerning crucial domain-specific concepts; (3) lacking modeling guidance to construct ATs decomposing a goal; (4) lack of semantic interoperability, resulting in ad hoc stand-alone tools. We also discuss existing incremental solutions and how our analysis paves the way for overcoming those issues through a broader approach to risk management modeling.

cs.CR↗

arXiv2025

Measuring General Intelligence with Generated Games

We present gg-bench, a collection of game environments designed to evaluate general reasoning capabilities in language models. Unlike most static benchmarks, gg-bench is a data generating process where new evaluation instances can be generated at will. In particular, gg-bench is synthetically generated by (1) using a large language model (LLM) to generate natural language descriptions of novel games, (2) using the LLM to implement each game in code as a Gym environment, and (3) training reinforcement learning (RL) agents via self-play on the generated games. We evaluate language models by their winrate against these RL agents by prompting models with the game description, current board state, and a list of valid moves, after which models output the moves they wish to take. gg-bench is challenging: state-of-the-art LLMs such as GPT-4o and Claude 3.7 Sonnet achieve winrates of 7-9% on gg-bench using in-context learning, while reasoning models such as o1, o3-mini and DeepSeek-R1 achieve average winrates of 31-36%. We release the generated games, data generation process, and evaluation code in order to support future modeling work and expansion of our benchmark.

cs.AI↗

arXiv2025

Why Do Multi-Agent LLM Systems Fail?

Despite enthusiasm for Multi-Agent LLM Systems (MAS), their performance gains on popular benchmarks are often minimal. This gap highlights a critical need for a principled understanding of why MAS fail. Addressing this question requires systematic identification and analysis of failure patterns. We introduce MAST-Data, a comprehensive dataset of 1600+ annotated traces collected across 7 popular MAS frameworks. MAST-Data is the first multi-agent system dataset to outline the failure dynamics in MAS for guiding the development of better future systems. To enable systematic classification of failures for MAST-Data, we build the first Multi-Agent System Failure Taxonomy (MAST). We develop MAST through rigorous analysis of 150 traces, guided closely by expert human annotators and validated by high inter-annotator agreement (kappa = 0.88). This process identifies 14 unique modes, clustered into 3 categories: (i) system design issues, (ii) inter-agent misalignment, and (iii) task verification. To enable scalable annotation, we develop an LLM-as-a-Judge pipeline with high agreement with human annotations. We leverage MAST and MAST-Data to analyze failure patterns across models (GPT4, Claude 3, Qwen2.5, CodeLlama) and tasks (coding, math, general agent), demonstrating improvement headrooms from better MAS design. Our analysis provides insights revealing that identified failures require more sophisticated solutions, highlighting a clear roadmap for future research. We publicly release our comprehensive dataset (MAST-Data), the MAST, and our LLM annotator to facilitate widespread research and development in MAS.

cs.AI↗

arXiv2025

Enough Coin Flips Can Make LLMs Act Bayesian

Large language models (LLMs) exhibit the ability to generalize given few-shot examples in their input prompt, an emergent capability known as in-context learning (ICL). We investigate whether LLMs use ICL to perform structured reasoning in ways that are consistent with a Bayesian framework or rely on pattern matching. Using a controlled setting of biased coin flips, we find that: (1) LLMs often possess biased priors, causing initial divergence in zero-shot settings, (2) in-context evidence outweighs explicit bias instructions, (3) LLMs broadly follow Bayesian posterior updates, with deviations primarily due to miscalibrated priors rather than flawed updates, and (4) attention magnitude has negligible effect on Bayesian inference. With sufficient demonstrations of biased coin flips via ICL, LLMs update their priors in a Bayesian manner.

cs.CL↗

arXiv2025

LangProBe: a Language Programs Benchmark

Composing language models (LMs) into multi-step language programs and automatically optimizing their modular prompts is now a mainstream paradigm for building AI systems, but the tradeoffs in this space have only scarcely been studied before. We introduce LangProBe, the first large-scale benchmark for evaluating the architectures and optimization strategies for language programs, with over 2000 combinations of tasks, architectures, optimizers, and choices of LMs. Using LangProBe, we are the first to study the impact of program architectures and optimizers (and their compositions together and with different models) on tradeoffs of quality and cost. We find that optimized language programs offer strong cost--quality Pareto improvement over raw calls to models, but simultaneously demonstrate that human judgment (or empirical decisions) about which compositions to pursue is still necessary for best performance. We will open source the code and evaluation data for LangProBe.

cs.CL↗

arXiv2024

Predicting Emergent Capabilities by Finetuning

A fundamental open challenge in modern LLM scaling is the lack of understanding around emergent capabilities. In particular, language model pretraining loss is known to be highly predictable as a function of compute. However, downstream capabilities are far less predictable -- sometimes even exhibiting emergent jumps -- which makes it challenging to anticipate the capabilities of future models. In this work, we first pose the task of emergence prediction: given access to current LLMs that have random few-shot accuracy on a task, can we predict whether future models (GPT-N+1) will have non-trivial accuracy on that task? We then discover a simple insight for this problem: finetuning LLMs on a given task can shift the point in scaling at which emergence occurs towards less capable models. To operationalize this insight, we can finetune LLMs with varying amounts of data and fit a parametric function that predicts when emergence will occur (i.e., "emergence laws"). We validate this approach using four standard NLP benchmarks where large-scale open-source LLMs already demonstrate emergence (MMLU, GSM8K, CommonsenseQA, and CoLA). Using only small-scale LLMs, we find that, in some cases, we can accurately predict whether models trained with up to 4x more compute have emerged. Finally, we present a case study of two realistic uses for emergence prediction.

cs.LG↗

arXiv2024

Explaining Datasets in Words: Statistical Models with Natural Language Parameters

To make sense of massive data, we often fit simplified models and then interpret the parameters; for example, we cluster the text embeddings and then interpret the mean parameters of each cluster. However, these parameters are often high-dimensional and hard to interpret. To make model parameters directly interpretable, we introduce a family of statistical models -- including clustering, time series, and classification models -- parameterized by natural language predicates. For example, a cluster of text about COVID could be parameterized by the predicate "discusses COVID". To learn these statistical models effectively, we develop a model-agnostic algorithm that optimizes continuous relaxations of predicate parameters with gradient descent and discretizes them by prompting language models (LMs). Finally, we apply our framework to a wide range of problems: taxonomizing user chat dialogues, characterizing how they evolve across time, finding categories where one language model is better than the other, clustering math problems based on subareas, and explaining visual features in memorable images. Our framework is highly versatile, applicable to both textual and visual domains, can be easily steered to focus on specific properties (e.g. subareas), and explains sophisticated concepts that classical methods (e.g. n-gram analysis) struggle to produce.

cs.AI↗

arXiv2024

Efficacy of Language Model Self-Play in Non-Zero-Sum Games

Game-playing agents like AlphaGo have achieved superhuman performance through self-play, which is theoretically guaranteed to yield optimal policies in competitive games. However, most language tasks are partially or fully cooperative, so it is an open question whether techniques like self-play can effectively be used to improve language models. We empirically investigate this question in a negotiation game setting known as Deal or No Deal (DoND). Crucially, the objective in DoND can be modified to produce a fully cooperative game, a strictly competitive one, or anything in between. We finetune language models in self-play over multiple rounds of filtered behavior cloning in DoND for each of these objectives and evaluate them in self-play and in collaboration with humans. We find that language models improve substantially in self-play, achieving 14-17x higher scores in task reward after finetuning. Further, the trained models generalize to both cooperation and competition with humans, scoring 2.5-6x higher than base models. We view these results as an early promising sign for language model self-play in cooperative settings, despite a lack of theoretical guarantees.

cs.CL↗

arXiv2024

Linguistic Bias in ChatGPT: Language Models Reinforce Dialect Discrimination

We present a large-scale study of linguistic bias exhibited by ChatGPT covering ten dialects of English (Standard American English, Standard British English, and eight widely spoken non-"standard" varieties from around the world). We prompted GPT-3.5 Turbo and GPT-4 with text by native speakers of each variety and analyzed the responses via detailed linguistic feature annotation and native speaker evaluation. We find that the models default to "standard" varieties of English; based on evaluation by native speakers, we also find that model responses to non-"standard" varieties consistently exhibit a range of issues: stereotyping (19% worse than for "standard" varieties), demeaning content (25% worse), lack of comprehension (9% worse), and condescending responses (15% worse). We also find that if these models are asked to imitate the writing style of prompts in non-"standard" varieties, they produce text that exhibits lower comprehension of the input and is especially prone to stereotyping. GPT-4 improves on GPT-3.5 in terms of comprehension, warmth, and friendliness, but also exhibits a marked increase in stereotyping (+18%). The results indicate that GPT-3.5 Turbo and GPT-4 can perpetuate linguistic discrimination toward speakers of non-"standard" varieties.

cs.CL↗

arXiv2024

American Sign Language Handshapes Reflect Pressures for Communicative Efficiency

Communicative efficiency is a key topic in linguistics and cognitive psychology, with many studies demonstrating how the pressure to communicate with minimal effort guides the form of natural language. However, this phenomenon is rarely explored in signed languages. This paper shows how handshapes in American Sign Language (ASL) reflect these efficiency pressures and provides new evidence of communicative efficiency in the visual-gestural modality. We focus on hand configurations in native ASL signs and signs borrowed from English to compare efficiency pressures from both ASL and English usage. First, we develop new methodologies to quantify the articulatory effort needed to produce handshapes and the perceptual effort required to recognize them. Then, we analyze correlations between communicative effort and usage statistics in ASL or English. Our findings reveal that frequent ASL handshapes are easier to produce and that pressures for communicative efficiency mostly come from ASL usage, rather than from English lexical borrowing.

cs.CL↗

arXiv2024

The Perspectivist Paradigm Shift: Assumptions and Challenges of Capturing Human Labels

Longstanding data labeling practices in machine learning involve collecting and aggregating labels from multiple annotators. But what should we do when annotators disagree? Though annotator disagreement has long been seen as a problem to minimize, new perspectivist approaches challenge this assumption by treating disagreement as a valuable source of information. In this position paper, we examine practices and assumptions surrounding the causes of disagreement--some challenged by perspectivist approaches, and some that remain to be addressed--as well as practical and normative challenges for work operating under these assumptions. We conclude with recommendations for the data labeling pipeline and avenues for future research engaging with subjectivity and disagreement.

cs.LG↗

arXiv2024

Pose Priors from Language Models

Language is often used to describe physical interaction, yet most 3D human pose estimation methods overlook this rich source of information. We bridge this gap by leveraging large multimodal models (LMMs) as priors for reconstructing contact poses, offering a scalable alternative to traditional methods that rely on human annotations or motion capture data. Our approach extracts contact-relevant descriptors from an LMM and translates them into tractable losses to constrain 3D human pose optimization. Despite its simplicity, our method produces compelling reconstructions for both two-person interactions and self-contact scenarios, accurately capturing the semantics of physical and social interactions. Our results demonstrate that LMMs can serve as powerful tools for contact prediction and pose estimation, offering an alternative to costly manual human annotations or motion capture data. Our code is publicly available at https://prosepose.github.io.

cs.CV↗