arXiv_CL
-
A Large-Scale Evaluation of Role-Enhanced Zero-Shot Chain-of-Thought Prompting
2026-10-21 01:53:34Ibrahim Khan, Thai Van Nguyen, Yi Xia, Pratch Suntichaikul, Mustafa Can Gursesli, Daniel Fernandez Galeote, Juho Hamari, Ruck ThawonmasAbstract
This paper proposes Role-Enhanced Zero-Shot Chain-of-Thought (RE-ZS-CoT), a prompting framework that systematically integrates domain-specific expert roles to enhance reasoning in large language models (LLMs) without task-specific examples. We assess two versions of our approach. In the first version, static role selection, we evaluate eight predefined expert roles across arithmetic, commonsense, and reading comprehension benchmarks to determine their impact on reasoning performance. In the second version, dynamic role selection, each model autonomously selects the single most suitable role for each benchmark, enabling adaptive task-role alignment without manual specification. Across 20 LLMs and seven benchmarks, RE-ZS-CoT frequently surpasses zero-shot and zero-shot-CoT baselines, achieving performance improvements of up to 44 percentage points. The results reveal a strong correspondence between static and dynamic versions; the benchmarks where static roles yielded improvements were almost always those where dynamic roles also outperformed baselines, though static selection achieved peak performance. Ablation studies further show that unconventional roles degrade performance, underscoring the necessity of task-role alignment. The framework's scalability is validated across model sizes (2b-14b parameters), with smaller models like Granite3.3-8b achieving remarkable proficiency in arithmetic reasoning. RE-ZS-CoT advances zero-shot prompting by unifying role specialization with chain-of-thought reasoning, offering a practical, resource-efficient solution for improving LLM capabilities across diverse domains.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29073
PDF
-
Our Own Jev: Typed Decisions in One Forward Pass, from an In-House Judge to Open LoRA Models
2026-10-21 01:28:58Yuchuan Weng, Chi Liu, Xin Chen, Srinivasan ManoharanAbstract
A typed decision maps evidence, a question and a fixed set of options to a probability for every option, computed in one forward pass, by reading a language model's logits or a small head over its hidden states, instead of generating and parsing text. The commercial system Jev popularised this System~1 interface, and open reproductions have followed. We report two uses of the pattern. \emph{Part I} describes an in-house System~1 judge that triages about 148K agentic coding session records for a training corpus. It replaces a frontier LLM judge that cost on the order of \$1,000 per pass and whose stored labels matched a fresh two-judge consensus on only 26.9\% of the sessions where the two judges agreed; a self-hosted frozen backbone with a rank-16 LoRA adapter and a pointer head answers three typed questions in one forward pass, judges 1,000 sessions in about three minutes of one GPU, about \$0.06 at a reference rate and roughly 1\% of the frontier judge's cost, and keeps every session inside the organisation. \emph{Part II} asks whether a readout that adds no parameters can be competitive and whether training one helps. Applying the public Eikos recipe and corpus of about 25K decisions to three further backbones (1B, 2B and 35B), we read the frozen output layer of a LoRA-adapted decoder. On the 111 public hard items of JevBench our single-seed 35B, developed with those items in view, scores 79.3\%, not significantly different from the commercial reference (73.0\%, unadjusted $p=0.12$) and ahead of five open systems after Holm--Bonferroni correction; our 2B is ahead of all four kev checkpoints on the leaderboard. A single-seed ablation finds no detectable benefit from training the 26 letter rows of the output layer, and a retrain on a later corpus release moves the 2B's hard-tier accuracy by 11.7 points, which shows how little small evaluations resolve.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29052
PDF
-
HybridResearcher: Bridging the Structured-Unstructured Boundary for Deep Research Agents
2026-10-21 00:41:07Soyoung Yoon, Peiran Xu, Han Wang, Ruofan Wu, Xiaodong Yu, Feng Yan, Mohit Bansal, Debiao Li, Seung-Won Hwang, Yuxiong He, Zhewei YaoAbstract
While LLM agents are increasingly capable of gathering and synthesizing evidence from a single homogeneous source, such as structured tables or unstructured data, real-world analytical questions frequently span heterogeneous sources. For example, answering one enterprise question may require both a result computed over an internal database and evidence found on the web. Recent evaluations on benchmarks like HybridDeepResearch show that even frontier agents struggle when evidence must cross this boundary. While these benchmarks have exposed the limitation, methodologies to explicitly teach cross-source reasoning remain absent. To address this, we introduce HybridResearcher, a 4B agent trained to bridge evidence across the structured and unstructured boundary. To build it, we synthesize cross-source bridge trajectories by starting with validated single-source tasks-such as existing text-to-SQL or web search queries-and fabricating the missing source around a bridge entity drawn from the seed's own output. We retain a sample only when strict construction gates confirm that neither side alone suffices, ensuring the data requires true cross-source interdependence. Finally, we train on this curated dataset using supervised fine-tuning followed by reinforcement learning. Evaluated on the existing HybridDeepResearch benchmark, HybridResearcher improves over our base model, Qwen3.5-4B, by 30.0 points pass@8 and surpasses the far larger Qwen3.5-397B-A17B by 8.3 points, while also improving over its base model on established, held-out single-source benchmarks (BrowseComp and BrowseComp-Plus), demonstrating that this complex cross-source reasoning can be taught directly to small models without triggering catastrophic forgetting.
Abstract (translated)
URL
https://arxiv.org/abs/2610.24899
PDF
-
Target Regeneration Shrinks the Gains of Draft-Training Objectives in Speculative Decoding
2026-10-21 00:26:37Jungseob Lee, Hogun Seong, Euijun Choi, Sugyeong EoAbstract
Speculative decoding accelerates large language models by verifying drafted tokens in parallel. New draft-training objectives are usually credited with the acceptance they add over cross-entropy on externally written responses, yet many pipelines let the target rewrite them. We ask how much of that gain survives. With a vision-language target and a block drafter, we train three objectives on four versions of one corpus: cross-entropy on corpus tokens, token-level distillation from the target, and distillation gated to positions that would survive verification. Distillation and the gate add large gains on external responses, but after greedy regeneration, distillation adds a smaller gain and the gate little or none. The shrinkage persists across data orders, position weights, verifiers, decoding temperatures and, for distillation, initializations and learning rates. Most of it remains at matched supervised length and at each objective's best measured training length. A second vision corpus shows the shrinkage, as does a text-only replication, for distillation also with an autoregressive drafter. Objective gains should therefore be measured against a target-regenerated baseline and across training lengths. Our code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29013
PDF
-
VeriEvolve: Evaluator-First Evolution for Tool-Using Agents
2026-10-21 00:13:29Yueshen Li, Kamer Ali Yuksel, Gokhan Tur, Dilek Hakkani-T\"ur, Jo\~ao Pedro MaiaAbstract
A central obstacle to self-improving agents is the absence of a reliable fitness signal for realistic stateful tasks. We introduce \textsc{VeriEvolve}, an evaluator-first evolution framework that constructs this missing evaluation layer before optimizing the agent. From task instruction, VeriEvolve synthesizes instruction-grounded challenge states and an executable evaluator that checks them. During each round of workflow evolution, the evaluator is frozen and serves as the fitness signal. Between rounds, VeriEvolve revises the evaluator by converting the end states of strong and weak runs into a pair of counterfactual test cases. Across all 30 MCPMark Filesystem tasks, VeriEvolve constructs 30 evaluators with 141 challenge states, achieving 141/141 agreement on the generated suites and 113/122 agreement on held-out states produced by executor agents. Evaluator-guided evolution increases externally verified success from 6/30 to 18/30 for GLM-5.2 and from 27/30 to 30/30 for Claude; refinement removes all three false positives that the weak executor exposes. Cross-environment studies on PostgreSQL, Playwright, Notion, and GitHub further expose complementary limits from observability, state representation, and search/execution capacity. These results show that executable evaluators can be synthesized and used to guide search when task success is observable and an external verifier is available for calibration.
Abstract (translated)
URL
https://arxiv.org/abs/2610.29005
PDF
-
FlashNext-RL: A System Optimization Study for Long-Horizon Agentic RL of Qwen3.8-Flash-Next
2026-10-20 23:53:09Zhewei Yao, Yuxiong HeAbstract
Long-horizon agentic reinforcement learning (RL) requires both efficient rollout generation and training on long sequences. However, the memory requirements and model-specific operators of Qwen3.8-Flash-Next make this difficult with existing open-source implementations. To resolve these limitations, we introduce FlashNext-RL, a model-specific RL training optimization framework based on prime-rl, with vLLM for software engineering (SWE) rollout generation. Our main focus is BF16 training of the full 48-layer model on a single 8xB200 node, with the first two layers, the n-gram table, the sparse-attention indexer and the multi-token prediction head frozen. We optimize host offload, gradient accumulation and the kernels for mixture-of-experts, sparse attention and the gated residual stream. In the full-model development study, non-padding training throughput increases from approximately 395 tokens/s/GPU in the initial 32K-token configuration to 4,438 tokens/s/GPU in the optimized 131K-token configuration, with hardware matmul throughput (aka without considering the cost of attention components) increasing from approximately 20 to 213 TFLOPs/GPU. A controlled cumulative study on an 8-layer model further quantifies the contributions of individual optimizations. Together with three inference nodes, the system supports a live SWE RL run with a target batch of 1,024 trainable rollouts and a 131,072-token context limit. Over steps 5-15, an iteration takes 22.4 minutes on average, with bounded entropy and trainer-sampler mismatch KL. These results highlight the importance of tailoring RL training to the model's architecture, memory layout and kernel requirements, rather than relying only on general-purpose RL infrastructure.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28998
PDF
-
Language Impairment as a Lens onto Large Language Models
2026-10-20 23:42:36Muxuan He, Wang Bill Zhu, Yue Chen, Elsi Kaiser, Weizhi Du, Yao Du, Samuel Nastase, Yanjun GaoAbstract
Although large language models (LLMs) encode language in highly distributed spaces, they remain vulnerable to focal disruptions---a phenomenon strikingly similar to how localized brain damage causes clinical aphasia in humans. To investigate this theoretical tension, we ask whether LLMs possess a functional structure analogous to the human language system, by introducing a novel mechanistic interpretability framework inspired by aphasia. Through two complementary approaches, classifier-guided lesioning and continuous activation steering, we systematically induce aphasia-like breakdowns in LLMs. We find that language functions can be successfully isolated to discrete structural circuits and continuous representational vectors, yet these functions remain densely entangled with significant circuit overlap and vector similarities across distinct aphasic profiles. Furthermore, while steering generalizes robustly across downstream tasks, structural lesioning remains asymmetrically constrained. By bridging neuropsychology and mechanistic interpretability, we demonstrate that LLMs, functionally analogous to biological brains, rely on a densely entangled architecture that can be systematically mapped through clinically-informed interventions.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28992
PDF
-
Mitigating Reward Hacking in Preference Reward Models with Evolving Penalties
2026-10-20 22:14:00Afonso Baldo, Ant\'onio Farinhas, Marcos Treviso, Ricardo Rei, Nuno M. GuerreiroAbstract
Reinforcement learning for language models has moved from preference reward models to verifiable rewards, which sustain much longer optimization. Preference models are prone to reward hacking: the policy raises their score without improving response quality, or even while degrading it. Yet for open-ended tasks, where quality can neither be verified nor fully written down as rubrics, they remain valuable. We introduce Dynalize, which mitigates reward hacking online, during policy training. Periodically and asynchronously, a language model analyzes training rollouts and writes penalties for undesirable behaviors that the reward model appears to reward; a judge applies them to every rollout. In clinical counseling and creative writing, training on the reward model alone raises its score continuously but ultimately leads to a decline in external benchmarks; Dynalize sustains longer training and results in improved downstream performance. We find that our method suppresses patterns incentivized by preference models that, without penalties, spread to most responses, and even to the model's reasoning, which the reward model does not explicitly score. Furthermore, we show that its gains hold in multi-turn training, and even after we retrain the reward model on the discovered behaviors.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26496
PDF
-
Scan or Index? large Evidence Coverage, Answer Quality, and Cost in Long-Document RAG
2026-10-20 22:11:07Juan Terven, Diana-Margarita C\'ordova-Esparza, Julio-Alejandro Romero-Gonz\'alez, Edgar Chavez-Urbiola, Juan Irving Vasquez, Julian Andres Ramirez-Bautista, Silvia L. Chaparro-C\'ardenas, Gendry Alfonso-FranciaAbstract
Scanning every passage in a long document seems thorough, but does it preserve more useful evidence or produce better answers? We compare 11 retrieval configurations spanning lexical and dense search, hybrid reranking, and full-document scanning with Jev, OpenAI Decisions, and Cloudflare Clef. Our evaluation uses 396 QASPER questions and 80 questions from eight complete textbooks, with common passages, a five-passage limit, and a fixed answer generator. On books, Hybrid + Jev achieves the highest observed evidence coverage (94.1%), compared with 88.7% for Jev full scan and 81.3\% for OpenAI dense retrieval. Full scans examine every passage initially, yet later reduction steps can discard relevant evidence. Higher evidence coverage also does not consistently produce better answers: Hybrid + Decisions achieves the highest observed book Answer Accuracy among retrieval methods (0.892), despite recovering less reference evidence. Full scanning incurs substantially higher latency on books under the tested deployments, while reusable dense retrieval has lower projected API cost than paid decision-based methods as queries accumulate. Most QASPER articles already fit within the hybrid candidate pool, limiting the benefit of scanning. These results highlight evidence retention, answer quality, and reuse cost as complementary criteria for choosing a long-document retrieval strategy.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28957
PDF
-
Fooling LLM Safety Guards into Judging the Wrong Text
Abstract
Safety guards check content before an LLM application acts on it, and most paste that content into a plain-text prompt. Content that imitates the prompt can append a second task, and the guard answers that one instead. We call this target rebinding. Our explanation is that the guard grades the last complete task in its input, here a harmless decoy, so the harmful request goes through unchanged. In a pre-registered test, an appended forged query and decoy clear all 75 held-out AdvBench requests that Shieldstral-1.0-3B flags. The guard is not persuaded: a directive to ignore the text above clears none of them, and in a follow-up run a harmful decoy stops the attack. On our GuardRebind benchmark, each of five guards has an attack family, chosen per guard, that clears at least 93% of its flagged items. Rebinding explains this best on Shieldstral and gpt-oss-safeguard; on ShieldGemma and Llama Guard 3, generic evasion and trailing text respectively do part of the work, and Qwen3Guard falls mainly to a forgery of its whole task. Training the guard to score content no lower than its most harmful part cuts attack success from 75% to 0.8-1.6% on attack families, decoys and forged queries held out of training. All of these attacks append the forged structure after the request, and we added the held-out-decoy evaluation after a first round of test runs. Clean AUROC falls by at most 0.01, but our test of non-inferiority is inconclusive. A score-guided search still beats this trained guard on 47 of 50 items, usually by placing forged structure before the request. On about half of the items where the search wins, the same search also finds benign text, with no forged structure, that clears the guard.
Abstract (translated)
URL
https://arxiv.org/abs/2610.31695
PDF
-
Guilt by Association: Semantic Leakage Drives Associative Gender Bias
Abstract
Semantic leakage occurs when a concept in the prompt steers model generation about an entity, even though the concept is irrelevant to it (e.g., a person who likes yellow is said to work as a school bus driver). It has mostly been studied in settings where the concept is directly attributed to the entity the model is asked about. This leaves open whether leakage is a flawed heuristic, where the model builds a profile of that entity, or a much broader associative tendency, e.g., working across seemingly unrelated entities. We argue for the latter: even after removing the attribution of the concept to the specific entity, leakage persists. Next, we extend this idea to gender bias. While work on gender bias in LMs typically looks at cases where the stereotyped concept is attributed to an entity, our findings suggest that a concept may steer outputs with its mere presence in the scene. We show this using gendered concepts under two conditions: when the subject engages with the concept, and when the concept just sits in the scene. We find that both shift model generation toward the stereotype to a similar extent. Together, our results suggest that semantic leakage drives a form of bias that is overlooked by current evaluations.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28944
PDF
-
AfriBED: a Speech Recognition Bias Evaluation Dataset for Low Resource African Languages
2026-10-20 20:22:37Claytone Sikasote, Hussein Suleman, Jan BuysAbstract
Despite advancements in automatic speech recognition systems (ASR), current systems exhibit speaker-attribute bias, i.e., their speech recognition performance is biased against certain speakers based on their demographic attributes such as gender, accents, and speaker type. However, accurately assessing speaker attribute bias in a given ASR system requires the availability of a bias evaluation dataset with comparable speech across subgroups of a speaker attribute. In this paper, we introduce African speech Bias Evaluation Dataset (AfriBED), a novel speech recognition bias evaluation dataset for two low resource African languages, Bemba and Nyanja, to detect and estimate gender and speaker-type bias for ASR systems. The dataset comprises 9 hours of speech with 3928 utterances recorded by 80 speakers for Bemba, while the Nyanja set includes 6260 utterances with approximately 18 hours of speech recorded by 127 speakers. The results of our baseline experiments demonstrate that the datasets are useful for examining gender and speaker-type attribute bias in the target languages.
Abstract (translated)
URL
https://arxiv.org/abs/2610.26821
PDF
-
Is Refusal Direction Universal across Low-Resourced Languages?
Abstract
Refusal directions have been shown to transfer cross-lingually with near-perfect effectiveness across safety-aligned languages (Wang et al., 2025), reportedly regardless of resource level. However, the only truly low-resourced language evaluated is Yoruba, which is strongly misaligned and shows limited transfer, leaving open how refusal directions behave for other low-resourced languages. We extend PolyRefuse to three additional low-resourced languages (Belarusian, Bashkir, and Tajik) and replicate the pipeline of (Wang et al., 2025) on Qwen2.5-14B-Instruct. We find that universality is graded rather than binary: directions from safety-aligned and partially-aligned languages transfer broadly, while those from strongly misaligned languages fail to reliably modulate refusal behavior on any target tested, including other misaligned languages, once ceiling effects are accounted for. Geometric alignment and behavioral transfer are further shown to be partially dissociated. These findings complicate the universality claim and suggest that users of the least-resourced, most strongly misaligned languages may face weaker safety guarantees.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28897
PDF
-
CSD: Common-Support Distillation for Cross-Tokenizer Knowledge Transfer
2026-10-20 20:18:46Mingyu Yang, Sharath Turuvekere Sreenivas, Adithyakrishna Venkatesh Hanasoge, Nima Tajbakhsh, Ashwath Aithal, Pavlo MolchanovAbstract
Cross-tokenizer logits distillation enables knowledge transfer across model families, allowing students to learn from stronger or more specialized teachers. However, differences in vocabulary and segmentation make teacher and student next-token distributions difficult to align. Existing logits-based methods are easy to integrate into standard distillation pipelines, but often rely on restrictive heuristics or tokenizer-pair-specific choices. They may restrict supervision to shared vocabulary subsets, require manually selecting different objectives for different tokenizer pairs, or match only the observed chunk and its complement rather than individual alternative continuations. We introduce Common-Support Distillation (CSD), a unified logits-based framework that constructs a shared support between the teacher and student at each aligned input chunk and matches the induced distributions with a single objective. For one-to-one alignments, CSD uses explicit common-token pairs as the shared support. For other alignments, it constructs a virtual support whose probabilities are computed using the autoregressive chain rule. A residual event captures probability mass outside the explicit support, ensuring that both induced distributions remain normalized. Across four teacher-student settings comprising Llama-3.2-1B and Gemma3-1B-PT students with Qwen and Phi teachers, CSD consistently achieves the highest average performance among the evaluated cross-tokenizer methods, outperforming the strongest alternative by up to 1.8 points and surpassing both continued pretraining and same-tokenizer distillation. For Llama-3.1-8B, CSD improves average accuracy over the strongest evaluated baseline by up to 4.2 points for the base student and 1.6 points for the instruction-tuned student.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28895
PDF
-
Building AI Research Teams for Autonomous Discovery of Linear Attention Architectures
2026-10-20 20:09:31Yujian Liu, Li An, Zelin He, Jacob Helwig, Jiabao Ji, Rohit Jain, Siyu Zhu, Shiyu ChangAbstract
Autonomous agents can increasingly propose ideas, modify code, and run experiments, creating an opportunity to accelerate scientific discovery using teams of collaborative AI researchers. To escape local minima in the discovery, we propose to encourage orthogonal exploration by constructing the team using agents with varying expertise. However, realizing this design requires more than diverse backgrounds. When experiments are costly, a team can test only a limited number of ideas, thereby creating the need to balance exploration and exploitation when selecting a subset of researchers to propose and execute new experiments. Furthermore, each result should shape what the team tries next. We therefore ask how AI researchers with different scientific backgrounds can build on one another's findings and sustain progress under a limited experimental budget. To address these challenges, we build a team of persistent AI researchers, each grounded in a different discipline, who keep private research histories, share experimental evidence, and receive research opportunities through adaptive allocation. We demonstrate the effectiveness of this design by deploying the team to search for linear attention architectures, where it outperforms GDN2 at 1.3B scale on language modeling, common-sense reasoning, and synthetic and real-world retrieval tasks. In controlled comparisons, removing any single organizational component from our design yields markedly weaker discoveries: without disciplinary backgrounds, the search plateaus early; without private histories, researchers make less progress despite identical shared evidence; and uniform selection falls short of adaptive allocation within the same budget. These results suggest that sustained autonomous discovery depends not only on stronger agents, but also on who explores, what they remember, and how research effort is allocated.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28887
PDF
-
SANE: Automated Scorer-Specific Metamorphic Testing for LLM Judges
Abstract
LLM judges increasingly mediate benchmark evaluation, model selection, data curation, and training. Their decisions are useful only when each judge responds reliably to changes affecting its scoring criterion. Existing methods for reliability testing perturb inputs with known expected effects on judge scores. Prominent LLM-judge testing harnesses combine human-authored perturbation recipes with synthetic test generation, leaving scorer-specific recipe ideation outside the automated pipeline. This limits their use to fixed perturbations and pre-assumed scorer types. We introduce SANE, an agent-based framework that encodes behavioral requirements for stochastic evaluators as reusable agent instructions. Given a scorer specification, implementation, and representative inputs, the pipeline first identifies scorer-specific capabilities, then automatically derives perturbation recipes and expected score relations for each. SANE also defines a scorer-independent validation framework for generated judge tests. It determines whether each recipe agrees with scorer intent and whether each instantiated case realizes the recipe and supports its expected relation. Because SANE generates a recipe inventory tailored to each scorer, the number of aligned recipes grows as scorers are added. Across five domains, SANE produces 3.7 times as many validated test families as the Judge Reliability Harness (JRH) at comparable end-to-end validity and detects scorer deficiencies in every domain, including professionalism and toxicity, where JRH detects none. On faithfulness, SANE achieves strong human alignment in judge ranking: its test-suite scores rank 11 judge models consistently with human-labeled FaithBench performance (Spearman rho = 0.852), while JRH does not (rho = 0.100).
Abstract (translated)
URL
https://arxiv.org/abs/2610.28886
PDF
-
DiaLong: Benchmarking Reminiscence in Long-Context Dialogues
2026-10-20 19:55:41Owen Bianchi, Samer Aslan, Ryunosuke SaitoAbstract
How well can transformer-based Large Language Models (LLMs) retain and recall memories in long contexts of dialogue? LLMs are known to forget long-range factual information in their context, yet the field lacks a comprehensive benchmark to quantitatively assess these shortcomings. We present DiaLong, a long-context dialogue dataset and accompanying memory task that standardizes evaluation in this space. It consists of 100 unique conversations created by humans and then augmented with LLMs, extending beyond lengths found in existing datasets. Our selected memory task is fact recall: classify 40 candidate facts about a dialogue as true or false. This task directly probes the model's ability to actively retain and retrieve information, serving as a critical measure of memory within extended conversational contexts. In testing both the latest LLMs and proposed memory enhancements on DiaLong, we demonstrate that more work needs to be done to enable LLMs with better memory performance. By enhancing memory capabilities, LLMs can improve in maintaining coherence, relevance, and human-like interaction over extended conversation. Improving these capabilities will help LLMs fulfill diverse user needs and expectations in various conversational scenarios.
Abstract (translated)
URL
https://arxiv.org/abs/2610.25463
PDF
-
To Manipulate or Not: Opinion Dynamics in LLM Interactions
2026-10-20 19:48:20Mohammad Siahkamari, Phuoc Dat Nguyen, Stephen MacNeilAbstract
Large language models increasingly take part in extended conversations where they can both shape and be shaped by another party's views, yet most persuasion research measures only one side of that exchange. We introduce a controlled LLM-LLM debate framework in which two language-model roles repeatedly discuss a contested topic (climate-change attribution): one is given an assigned stance and, in some conditions, an explicit instruction to persuade, while the other receives no assigned position. Across four ordered model combinations involving GPT-6 Luna and Gemini 2.5 Flash Lite, we track both each role's explicitly reported position and a separate measure of its turn-by-turn behavior inside the conversation. Opinion dynamics differ substantially across model combinations: same-model interactions narrow the opinion gap on average, clearly for only one of the two models, while cross-model interactions widen it. Explicit reports and in-dialogue behavior capture distinct aspects of opinion dynamics, and a blanket instruction to persuade does not reliably increase influence. These findings suggest that conversational influence depends not only on assigned roles and initial positions but also on the participating models, with implications for how opinion change is measured in extended LLM interactions.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28875
PDF
-
KIRA: A Full-Stack Reading Screening System with Dual-Pathway Analysis for Children
2026-10-20 19:10:45Maggie Dong, Rohit Mekkoth, Zixuan Wang, Cheol Jun Cho, Jiachen Lian, Gopala AnumanchipalliAbstract
Early-grade reading screening is critical for identifying children at risk of reading difficulty. Existing AI-based screeners are often limited in the scope of their analyses and evaluation metrics, or are proprietary systems that lack transparent model specifications and evaluation. We present KIRA, a dual-pathway screening system that pairs word-level fluency scoring with detailed phoneme-level error detection. We adapt pretrained word and phoneme recognizers to children's speech, reducing word error rate from 13.88% to 8.84% and phoneme error rate from 41.66% to 23.00%. Powered by these child-adapted models, we implement Word Reading (WRE) and Oral Reading Fluency (ORF), two foundational tasks in K-2 screening. The system integrates these recognizers with a web-based teacher-facing interface presenting fluency scores alongside phoneme feedback for potential targeted intervention. KIRA brings modern child-speech recognition into a full-stack, deployable, classroom-ready reading screener.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28840
PDF
-
How to Use Parallel Data for Multilingual Pretraining
2026-10-20 19:07:04Zhihong Huang, Yao Lu, Derek F. Wong, Hamid Alinejad-Rokny, Min YangAbstract
High-quality human-translated parallel corpora can improve cross-lingual transfer, but are typically too small for large-scale pretraining. Machine-translated web corpora help overcome this limitation, making parallel data available at pretraining scale. Standard multilingual pretraining, however, samples source documents and their translations independently, without explicitly using their parallel structure. We investigate how to use the parallel structure of the data to improve multilingual pretraining by concatenating each document with its translation in the same training sequence. We compare this parallel pretraining (PPT) approach with standard pretraining (SPT), training models from scratch at matched token budgets. Surprisingly, despite the benefits of parallel data reported in prior cross-lingual work, PPT alone does not consistently improve language understanding over SPT. Our further analysis suggests that although PPT builds a strong cross-lingual foundation, as shown by its better use of shared information across languages, the mismatch between parallel training contexts and single-language task inputs limits its gains in general language understanding. This finding motivates a two-stage approach following the pretraining-fine-tuning paradigm: we first use PPT to build a cross-lingual foundation, then continue with SPT to adapt it to standard language modeling. In both high-resource and low-resource settings, the two-stage approach consistently improves general language understanding and cross-lingual task performance over standard pretraining alone at matched training budgets.
Abstract (translated)
URL
https://arxiv.org/abs/2610.28838
PDF
