arXiv subjects

Seraphina Goldfarb-Tarrant

Publications and source records attributed to Seraphina Goldfarb-Tarrant.

At least 19 recordsLinked to original sources

arXiv2026

An Actionable Diagnosis of Multilingual, Multi-Agent Planning Failures

Multilingual multi-agent systems exhibit substantial degradation beyond English, yet prior work rarely identifies how task-critical information is lost when user requests are converted into executable plans. We study the planner in a multi-agent system as the request-to-action interface and derive an actionable taxonomy of planning-grounding failures from failed real-world task executions. LLM-based analysis shows that these failures constitute an increasing share of unsuccessful executions as language-resource availability declines, with the strongest effects in low-resource languages. To test whether the taxonomy supports mitigation, we introduce TART, Taxonomy-Guided Actionable Representation, that makes the taxonomy's key aspects explicit to the planner and downstream sub-agents. Across multiple languages, three LLM backbones, two datasets, and two agentic configurations, TART consistently improves performance. On multilingual GAIA, it raises a state-of-the-art system's accuracy by 5.6 percentage points averaged across eleven languages spanning low- to high-resource settings.

cs.MA↗

arXiv2026

Safety is Contextual, LLM-Judges Are Not: Navigating the Rigid Priors of Evaluators

LLMs-as-judges are the only way to evaluate safety at scale. Despite their importance, LLM-judges themselves are rarely evaluated beyond human agreement in simple, static benchmarks. We therefore investigate two under-explored but crucial properties of LLMs-as-judges: their susceptibility to relying on in context-information, and their steerability to differing safety definitions, which may not align with their internal safety priors. We evaluate the safety judging abilities of many generalist LLMs and safety-specific judges, and investigate the impact of task demonstrations, novel in-context information, and changing safety definitions. We find that while LLM-judges can learn from new information, they are broadly unlikely to adjust their evaluations if the context or safety definition contradicts their prior.

cs.AI↗

arXiv2026

Large Language Models Generate Harmful Responses Using a Distinct Mechanism, Shared Across Harm Types

Large language models remain vulnerable to jailbreaks that elicit harmful responses, yet the mechanism behind harmful response generation is poorly understood. Here, we investigate how this capability is organized within model parameters. We identify and prune parameters that specifically support harmful compliance, providing a direct mechanistic analysis at the parameter level. We find that this capability depends on a sparse set of critical parameters: pruning these parameters substantially reduces harmful compliance while causing only limited degradation in benign capabilities, suggesting that key components of harmful generation are separable from those of general utility. Parameters identified from one harm category also reduce harmful responses in others, indicating components shared across harm types. This separability appears primarily in aligned models, suggesting that alignment training internally reshapes the harmful response mechanism even when behavioral safeguards remain brittle. We further show that harmful response generation is dissociable from the ability to recognize and reason about harmfulness. Finally, we extend our analysis to emergent misalignment and identify a sparse set of parameters contributing to it, with substantial sharing across fine-tuning domains. Together, these results reveal a consistent parameter-level organization underlying unsafe behaviors and point toward more principled interventions for improving model safety.

cs.CL↗

arXiv2026

Learning is Forgetting: LLM Training As Lossy Compression

Despite the increasing prevalence of large language models (LLMs), we still have a limited understanding of how their representational spaces are structured. This limits our ability to interpret how and what they learn or relate them to learning in humans. We argue LLMs are best seen as an instance of lossy compression, where over training they learn by retaining only information in their training data relevant to their objective(s). We show pre-training results in models that are optimally compressed for next-sequence prediction, approaching the Information Bottleneck bound on compression. Across an array of open weights models, each compresses differently, likely due to differences in the data and training recipes used. However even across different families of LLMs the optimality of a model's compression, and the information present in it, can predict downstream performance on across a wide array of benchmarks, letting us directly link representational structure to actionable insights about model performance. In the general case the work presented here offers a unified Information-Theoretic framing for how these models learn that is deployable at scale.

cs.LG↗

arXiv2026

Lost in Simulation: LLM-Simulated Users are Unreliable Proxies for Human Users in Agentic Evaluations

Agentic benchmarks increasingly rely on LLM-simulated users to scalably evaluate agent performance, yet the robustness, validity, and fairness of this approach remain unexamined. Through a user study with participants across the United States, India, Kenya, and Nigeria, we investigate whether LLM-simulated users serve as reliable proxies for real human users in evaluating agents on {\tau}-Bench retail tasks. We find that user simulation lacks robustness, with agent success rates varying up to 9 percentage points across different user LLMs. Furthermore, evaluations using simulated users exhibit systematic miscalibration, underestimating agent performance on challenging tasks and overestimating it on moderately difficult ones. African American Vernacular English (AAVE) speakers experience consistently worse success rates and calibration errors than Standard American English (SAE) speakers, with disparities compounding significantly with age. We also find simulated users to be a differentially effective proxy for different populations, performing worst for AAVE and Indian English speakers. Additionally, simulated users introduce conversational artifacts and surface different failure patterns than human users. These findings demonstrate that current evaluation practices risk misrepresenting agent capabilities across diverse user populations and may obscure real-world deployment challenges.

cs.HC↗

arXiv2025

A Good Plan is Hard to Find: Aligning Models with Preferences is Misaligned with What Helps Users

To assist users in complex tasks, LLMs generate plans: step-by-step instructions towards a goal. While alignment methods aim to ensure LLM plans are helpful, they train (RLHF) or evaluate (ChatbotArena) on what users prefer, assuming this reflects what helps them. We test this with Planorama: an interface where 126 users answer 300 multi-step questions with LLM plans. We get 4388 plan executions and 5584 comparisons to measure plan helpfulness (QA success) and user preferences on plans, and recreate the setup in agents and reward models to see if they simulate or prefer what helps users. We expose: 1) user/model preferences and agent success do not accurately predict which plans help users, so common alignment feedback can misalign with helpfulness; 2) this gap is not due to user-specific preferences, as users are similarly successful when using plans they prefer/disprefer; 3) surface-level cues like brevity and question similarity strongly link to preferences, but such biases fail to predict helpfulness. In all, we argue aligning helpful LLMs needs feedback from real user interactions, not just preferences of what looks helpful, so we discuss the plan NLP researchers can execute to solve this problem.

cs.CL↗

arXiv2025

Scalpel vs. Hammer: GRPO Amplifies Existing Capabilities, SFT Replaces Them

Training large language models (LLMs) for reasoning via maths and code datasets has become a major new focus in LLM post-training. Two particularly popular approaches are reinforcement learning (RL) and supervised fine-tuning (SFT), but their training dynamics are poorly understood. We present a comparative analysis of RL and SFT on the same maths problems with the same model and similar hyperparameters. We find that RL yields minor in-domain gains on maths and slight degradation on knowledge-intensive benchmarks like MMLU, while both trends are more pronounced in SFT. We also analyse model parameters across checkpoints, observing that both algorithms modify query and key weights the most. Meanwhile, SFT exhibits greater updates and also affects mid-layer MLPs more, leading us to hypothesise that this may have caused the out-of-domain degradation. We therefore investigate whether freezing parts of the model during training can mitigate the reduced performance on knowledge-intensive benchmarks. However, our results are inconclusive, with benefits on GPQA:Diamond and degradation on other benchmarks. Taken together, our observations provide a preliminary indication for why RL amplifies existing capabilities, while SFT replaces old skills with new ones.

cs.LG↗

arXiv2025

The Multilingual Divide and Its Impact on Global AI Safety

Despite advances in large language model capabilities in recent years, a large gap remains in their capabilities and safety performance for many languages beyond a relatively small handful of globally dominant languages. This paper provides researchers, policymakers and governance experts with an overview of key challenges to bridging the "language gap" in AI and minimizing safety risks across languages. We provide an analysis of why the language gap in AI exists and grows, and how it creates disparities in global AI safety. We identify barriers to address these challenges, and recommend how those working in policy and governance can help address safety concerns associated with the language gap by supporting multilingual dataset creation, transparency, and research.

cs.AI↗

arXiv2025

MAPS: A Multilingual Benchmark for Agent Performance and Security

Agentic AI systems, which build on Large Language Models (LLMs) and interact with tools and memory, have rapidly advanced in capability and scope. Yet, since LLMs have been shown to struggle in multilingual settings, typically resulting in lower performance and reduced safety, agentic systems risk inheriting these limitations. This raises concerns about the accessibility of such systems, as users interacting in languages other than English may encounter unreliable or security-critical agent behavior. Despite growing interest in evaluating agentic AI and recent initial efforts toward multilingual interaction, existing benchmarks do not yet provide a comprehensive, multi-domain, security-aware evaluation of multilingual agentic systems. To address this gap, we propose MAPS, a multilingual benchmark suite designed to evaluate agentic AI systems across diverse languages and tasks. MAPS builds on four widely used agentic benchmarks - GAIA (real-world tasks), SWE-Bench (code generation), MATH (mathematical reasoning), and the Agent Security Benchmark (security). We translate each dataset into eleven diverse languages, resulting in 805 unique tasks and 9,660 total language-specific instances - enabling a systematic analysis of the Multilingual Effect on AI agents' performance and robustness. Empirically, we observe a degradation in both performance and security when transitioning from English to other languages, with severity varying by task and correlating with the amount of translated input. This work establishes the first standardized evaluation framework for multilingual agentic AI, encouraging future research towards equitable, reliable, and accessible agentic AI. MAPS benchmark suite is publicly available at https://huggingface.co/datasets/Fujitsu-FRE/MAPS

cs.DB↗

arXiv2025

Command A: An Enterprise-Ready Large Language Model

In this report we describe the development of Command A, a powerful large language model purpose-built to excel at real-world enterprise use cases. Command A is an agent-optimised and multilingual-capable model, with support for 23 languages of global business, and a novel hybrid architecture balancing efficiency with top of the range performance. It offers best-in-class Retrieval Augmented Generation (RAG) capabilities with grounding and tool use to automate sophisticated business processes. These abilities are achieved through a decentralised training approach, including self-refinement algorithms and model merging techniques. We also include results for Command R7B which shares capability and architectural similarities to Command A. Weights for both models have been released for research purposes. This technical report details our original training pipeline and presents an extensive evaluation of our models across a suite of enterprise-relevant tasks and public benchmarks, demonstrating excellent performance and efficiency.

cs.CL↗

arXiv2025

Safer or Luckier? LLMs as Safety Evaluators Are Not Robust to Artifacts

Large Language Models (LLMs) are increasingly employed as automated evaluators to assess the safety of generated content, yet their reliability in this role remains uncertain. This study evaluates a diverse set of 11 LLM judge models across critical safety domains, examining three key aspects: self-consistency in repeated judging tasks, alignment with human judgments, and susceptibility to input artifacts such as apologetic or verbose phrasing. Our findings reveal that biases in LLM judges can significantly distort the final verdict on which content source is safer, undermining the validity of comparative evaluations. Notably, apologetic language artifacts alone can skew evaluator preferences by up to 98\%. Contrary to expectations, larger models do not consistently exhibit greater robustness, while smaller models sometimes show higher resistance to specific artifacts. To mitigate LLM evaluator robustness issues, we investigate jury-based evaluations aggregating decisions from multiple models. Although this approach both improves robustness and enhances alignment to human judgements, artifact sensitivity persists even with the best jury configurations. These results highlight the urgent need for diversified, artifact-resistant methodologies to ensure reliable safety assessments.

cs.CL↗

arXiv2025

Small Changes, Large Consequences: Analyzing the Allocational Fairness of LLMs in Hiring Contexts

Large language models (LLMs) are increasingly being deployed in high-stakes applications like hiring, yet their potential for unfair decision-making remains understudied in generative and retrieval settings. In this work, we examine the allocational fairness of LLM-based hiring systems through two tasks that reflect actual HR usage: resume summarization and applicant ranking. By constructing a synthetic resume dataset with controlled perturbations and curating job postings, we investigate whether model behavior differs across demographic groups. Our findings reveal that generated summaries exhibit meaningful differences more frequently for race than for gender perturbations. Models also display non-uniform retrieval selection patterns across demographic groups and exhibit high ranking sensitivity to both gender and race perturbations. Surprisingly, retrieval models can show comparable sensitivity to both demographic and non-demographic changes, suggesting that fairness issues may stem from broader model brittleness. Overall, our results indicate that LLM-based hiring systems, especially in the retrieval stage, can exhibit notable biases that lead to discriminatory outcomes in real-world contexts.

cs.CL↗

arXiv2024

Mix Data or Merge Models? Optimizing for Diverse Multi-Task Learning

Large Language Models (LLMs) have been adopted and deployed worldwide for a broad variety of applications. However, ensuring their safe use remains a significant challenge. Preference training and safety measures often overfit to harms prevalent in Western-centric datasets, and safety protocols frequently fail to extend to multilingual settings. In this work, we explore model merging in a diverse multi-task setting, combining safety and general-purpose tasks within a multilingual context. Each language introduces unique and varied learning challenges across tasks. We find that objective-based merging is more effective than mixing data, with improvements of up to 8% and 10% in general performance and safety respectively. We also find that language-based merging is highly effective -- by merging monolingually fine-tuned models, we achieve a 4% increase in general performance and 7% reduction in harm across all languages on top of the data mixtures method using the same available data. Overall, our comprehensive study of merging approaches provides a useful framework for building strong and safe multilingual models.

cs.CL↗

arXiv2024

The Multilingual Alignment Prism: Aligning Global and Local Preferences to Reduce Harm

A key concern with the concept of "alignment" is the implicit question of "alignment to what?". AI systems are increasingly used across the world, yet safety alignment is often focused on homogeneous monolingual settings. Additionally, preference training and safety measures often overfit to harms common in Western-centric datasets. Here, we explore the viability of different alignment approaches when balancing dual objectives: addressing and optimizing for a non-homogeneous set of languages and cultural preferences while minimizing both global and local harms. We collect the first set of human annotated red-teaming prompts in different languages distinguishing between global and local harm, which serve as a laboratory for understanding the reliability of alignment techniques when faced with preference distributions that are non-stationary across geographies and languages. While this setting is seldom covered by the literature to date, which primarily centers on English harm mitigation, it captures real-world interactions with AI systems around the world. We establish a new precedent for state-of-the-art alignment techniques across 6 languages with minimal degradation in general performance. Our work provides important insights into cross-lingual transfer and novel optimization approaches to safeguard AI systems designed to serve global populations.

cs.CL↗

arXiv2024

A SMART Mnemonic Sounds like "Glue Tonic": Mixing LLMs with Student Feedback to Make Mnemonic Learning Stick

Keyword mnemonics are memorable explanations that link new terms to simpler keywords. Prior work generates mnemonics for students, but they do not train models using mnemonics students prefer and aid learning. We build SMART, a mnemonic generator trained on feedback from real students learning new terms. To train SMART, we first fine-tune LLaMA-2 on a curated set of user-written mnemonics. We then use LLM alignment to enhance SMART: we deploy mnemonics generated by SMART in a flashcard app to find preferences on mnemonics students favor. We gather 2684 preferences from 45 students across two types: expressed (inferred from ratings) and observed (inferred from student learning), yielding three key findings. First, expressed and observed preferences disagree; what students think is helpful does not always capture what is truly helpful. Second, Bayesian models can synthesize complementary data from multiple preference types into a single effectiveness signal. SMART is tuned via Direct Preference Optimization on this signal, which resolves ties and missing labels in the typical method of pairwise comparisons, augmenting data for LLM output quality gains. Third, mnemonic experts assess SMART as matching GPT-4 at much lower deployment costs, showing the utility of capturing diverse student feedback to align LLMs in education.

cs.CL↗

arXiv2024

MultiContrievers: Analysis of Dense Retrieval Representations

Dense retrievers compress source documents into (possibly lossy) vector representations, yet there is little analysis of what information is lost versus preserved, and how it affects downstream tasks. We conduct the first analysis of the information captured by dense retrievers compared to the language models they are based on (e.g., BERT versus Contriever). We use 25 MultiBert checkpoints as randomized initialisations to train MultiContrievers, a set of 25 contriever models. We test whether specific pieces of information -- such as gender and occupation -- can be extracted from contriever vectors of wikipedia-like documents. We measure this extractability via information theoretic probing. We then examine the relationship of extractability to performance and gender bias, as well as the sensitivity of these results to many random initialisations and data shuffles. We find that (1) contriever models have significantly increased extractability, but extractability usually correlates poorly with benchmark performance 2) gender bias is present, but is not caused by the contriever representations 3) there is high sensitivity to both random initialisation and to data shuffle, suggesting that future retrieval research should test across a wider spread of both.

cs.CL↗

arXiv2023

Cross-lingual Transfer Can Worsen Bias in Sentiment Analysis

Sentiment analysis (SA) systems are widely deployed in many of the world's languages, and there is well-documented evidence of demographic bias in these systems. In languages beyond English, scarcer training data is often supplemented with transfer learning using pre-trained models, including multilingual models trained on other languages. In some cases, even supervision data comes from other languages. Does cross-lingual transfer also import new biases? To answer this question, we use counterfactual evaluation to test whether gender or racial biases are imported when using cross-lingual transfer, compared to a monolingual transfer setting. Across five languages, we find that systems using cross-lingual transfer usually become more biased than their monolingual counterparts. We also find racial biases to be much more prevalent than gender biases. To spur further research on this topic, we release the sentiment models we used for this study, and the intermediate checkpoints throughout training, yielding 1,525 distinct models; we also release our evaluation code.

cs.CL↗

arXiv2023

This Prompt is Measuring : Evaluating Bias Evaluation in Language Models

Bias research in NLP seeks to analyse models for social biases, thus helping NLP practitioners uncover, measure, and mitigate social harms. We analyse the body of work that uses prompts and templates to assess bias in language models. We draw on a measurement modelling framework to create a taxonomy of attributes that capture what a bias test aims to measure and how that measurement is carried out. By applying this taxonomy to 90 bias tests, we illustrate qualitatively and quantitatively that core aspects of bias test conceptualisations and operationalisations are frequently unstated or ambiguous, carry implicit assumptions, or be mismatched. Our analysis illuminates the scope of possible bias types the field is able to measure, and reveals types that are as yet under-researched. We offer guidance to enable the community to explore a wider section of the possible bias space, and to better close the gap between desired outcomes and experimental design, both for bias and for evaluating language models more broadly.

cs.CL↗