paperlist.ai

AI conference papers

AI conference paper library

Search papers from 30+ top conferences, including NeurIPS, ICLR, ICML, CVPR, ACL and AAAI. Filter by keyword, then open any paper for its abstract, authors and links.

arXiv

SMTrap: Cost-Effective DoS Attacks Against Large Reasoning Models via SMT Conflict Guidance

Jian Yang, Zhenqi Feng, Zhaoyang Yu, Zhaoxin Fan, Kejian Wu, Xiaofeng Wang, Zheng Zhu, Jianjun Huang, Wei You, Bin Liang

Existing LRM-DoS methods rely heavily on model feedback to synthesize attack queries, requiring either repeated queries to the target model or training a dedicated attack model. These expensive operations severely weaken attack leverage. In this paper, we propose \emph{search amplification}, a novel, model-feedback-free LRM-DoS paradigm. It employs the conflict count derived from an Satisfiability Modulo Theories (SMT) solver as a low-cost external signal to guide the synthesis of inference-heavy Constraint Satisfaction Problem (CSP) instances. Our key observation is that LRMs depend on trial-and-backtracking search when solving CSPs, where higher SMT conflict counts on a given CSP instance positively correlate with more extensive LRM backtracking search and substantially longer output trajectories. Building on this finding, we propose \textsc{SMTrap}, a lightweight, CPU-only framework. Guided by SMT conflict counts, \textsc{SMTrap} generates inference-heavy CSP queries without model queries, attack-model training, or GPU computation. Evaluations across seven frontier models demonstrate the state-of-the-art LRM-DoS capability of \textsc{SMTrap}, producing DoS effects multiple times stronger than existing baselines. To mitigate the threat of \textsc{SMTrap}, we demonstrate a tool-based mitigation that significantly cuts token usage.

arXiv

Assessing Quality of Experience in Natural Language Generation of German Text

Dinh Nam Pham, Shushen Manakhimova, Vivien Macketanz, Sebastian Möller

The rapid advancement of Natural Language Generation (NLG) has made the reliable evaluation of generated text increasingly critical, as these systems, such as large language models (LLMs), are now widely deployed in real-world applications. However, traditional automatic metrics fail to capture the multifaceted nature of perceived quality. In this paper, we introduce TextQ-German, a novel dataset suite for human-centered evaluation of German NLG from a Quality of Experience (QoE) perspective, covering automatic text summarization and machine translation. Through crowdsourcing studies with German speakers, we collect human quality ratings and identify relevant perceptual quality dimensions for each task. We develop automatic QoE prediction models, including transformer-based, linguistic feature-based, and hybrid approaches. Hybrid models outperform pure transformer baselines in almost all experimental settings, while linguistic features alone can approach the performance of fine-tuned language models. The dataset is extended with LLM-generated outputs annotated with overall QoE scores. Final validation on held-out sets indicates generalization to unseen data. Our work contributes a publicly accessible resource for NLG evaluation and baselines for automatic QoE prediction, providing a foundation for developing NLG systems that better align with human quality perception.

arXiv

MLREF: Efficient Module Reuse for Reward Design in Reinforcement Learning via Large Language Models

Chenglin Liu, Xun Wang, Ruishuo Chen, Zhuoran Li, Longbo Huang

Reward function design remains a bottleneck in reinforcement learning. While large language models (LLMs) have enabled automated reward generation, existing methods generate and revise reward functions as monolithic programs, making it difficult to reliably preserve and reuse effective components discovered in earlier iterations, leading to unstable performance across iterations. To address this, we propose Module Level Reward Evolution Framework (MLREF). At the core of MLREF is a module pool, a persistent repository of reusable reward components. MLREF treats the module pool as the primary optimization object: the pool evolves across iterations by accumulating successful modules, refining underperforming ones, and reusing proven components; while reward functions are constructed as linear combinations of modules drawn from this pool. To drive this evolution, MLREF integrates three mechanisms: reflection-based refinement, hybrid credit assignment, and a merge strategy with rollback, which together improve the effectiveness and robustness of reward optimization. Experiments on 17 tasks show that MLREF outperforms strong baselines by 25.2% in locomotion and 6.6% in manipulation, with more stable optimization dynamics.

arXiv

Understanding Multilingual Medical ASR Adaptation Through Layer-Wise Analysis

Souranil Kahali, Rituparna Bose, Abner Hernandez, Tomas Arias-Vergara, Andreas Maier, Ning Ma, Paula Andrea Perez-Toro

Medical automatic speech recognition (MedASR) requires adaptation to specialised terminology, limited annotated clinical data, and multilingual use cases. Although large-scale pretrained ASR models such as Whisper achieve strong generalisation, their behaviour after medical and multilingual adaptation remains insufficiently understood beyond word error rate (WER). This paper investigates how multilingual medical adaptation reshapes the internal representations of Whisper models through layer-wise encoder analysis. We compare zero-shot decoding, English-only fine-tuning, German-only diagnostic fine-tuning, two-stage EN->EN+DE continuation, and direct EN+DE fine-tuning across Whisper model sizes. Fine-tuning substantially improves MedASR performance, but the best model depends on the adaptation setting: Whisper-Medium gives the lowest English WER (7.72%) and the lowest combined EN+DE WER under direct EN+DE training (26.30%); German-only Whisper-Large-v3 gives the lowest German WER (44.96%), but as a within-corpus diagnostic on 86 single-speaker training utterances rather than robust generalisation. Layer-wise analysis of the two-stage Whisper-Small trajectory shows that English medical fine-tuning produces the dominant encoder shift, whereas multilingual continuation largely preserves the adapted representation space. Domain and language information remain highly recoverable across layers, while linearly recoverable error-predictive cues weaken as WER improves.

arXiv

Identifying Implicit Premises for Logical Reconstruction of Argument Graphs

Xuyao Feng, Anthony Hunter

The logical reconstruction of argument graphs from natural language text is challenging because of the prevalence of enthymemes (i.e., arguments with implicit premises). There are natural language processing methods for identifying enthymemes in text, and there are symbolic methods based on abduction for identifying missing premises in a logical representation of enthymemes. However, there is a need for methods to generate implicit premises to logically show a known entailment or contradiction relationship between a pair of statements. To address this, we propose a neuro-symbolic pipeline that uses large language models (LLMs) to generate intermediate implicit premises that are translated into logical formulae and used with logical formulae representing explicit premises and explicit claims to show the logical relationships between them (entailment, contradiction, or neutrality). Our approach is evaluated on the Microtext Argumentative Corpus.

arXiv

Do Large Language Models Hallucinate Electric Fata Morganas?

Kristina Šekrst

AI hallucinations - that is, outputs which are made up, cannot be verified, or contradict the source material - are generally regarded as an engineering flaw to be dealt with. This paper contends that they also have philosophical significance when it comes to the question of machine consciousness. We examine the known causes of hallucinations in large language models - such as source-target divergence, discrepancies between training and inference, and overfitting - and we present two empirical investigations. In the first, we apply successive generations of the GPT model to ambiguous factual questions under different temperature settings, finding that higher temperatures result in plausible but incorrect answers while lower temperatures lead to factually accurate ones. The sampling parameters that cause a model to seem creative or spontaneous and thus more likely to pass behavioral tests of intelligence are the same ones that increase its hallucination rate. In the second, we look at an encoder-only model that has been trained on encyclopedic data and which answers questions of the same type factually and without embellishment, indicating that hallucinations are due to exposure to subjective and socially diverse training data rather than to the development of any cognitive ability. Using references to Turing, Searle's Chinese Room, the frame problem, and the cybernetic tradition of Wiener and Ashby, we claim that a model's self-reports of emotion or sentience come within the definition of hallucination, and that any future occurrence of machine consciousness might remain epistemically inaccessible since it would be indistinguishable from a sufficiently advanced hallucination.

arXiv

Decomposing Wrong-Consensus Agreement in LLM Self-Consistency

Lizhuo Zhang, Mengmeng Tang, Chenfeng Long, Xiaoyong Tang, Xiang Luo

Agreement among repeated samples of a language model is routinely read as evidence about answer reliability, yet wrong answers can agree just as strongly as right ones. This paper asks what information wrong-consensus agreement actually contains, and answers with a quantitative decomposition. A pluralistic agreement index Gamma, normalized by the reference scale d=(1-p)/(C-1), is split into a mechanical component (agreement delivered by a per-case answer preference alone) and a preference-unexplained residual. The mechanical reference is leak-free: each case's preference and accuracy are estimated from its other runs only. On public GPT-4.1 per-run data, coverage phi (the mechanical/empirical ratio) shows a benchmark-associated direction: 0.81-0.93 on multiple-choice GPQA-Diamond against 0.59-0.78 on open-domain AIME, where a residual of 1.54-2.80 Gamma units survives, more than absorbed by a calibrated run-level preference-heterogeneity reference. A controlled replication under one fixed protocol (four runs per question, K=32 votes) on five open-weights checkpoints (Qwen3.5-9B/122B, Qwen3.8-27B, Gemma4-26B/31B) finds near-complete mechanical coverage in all ten cells (phi approximately 1, with a small overshoot consistent with a quantified finite-donor plug-in bias), robust to a two-run design; the largest cell (qwen3.5-122b, p=0.222) sits inside the GPT-4.1 AIME accuracy range and still saturates (phi=1.041). A cross-system contrast at comparable aggregate accuracy contrasts near-complete mechanical agreement in the open-weights models against a larger preference-unexplained residual in the frontier family. This contrast is confounded with sampling protocol by design. Agreement is graded evidence, not certification. No new voting method is proposed; code and evidence are committed.

arXiv

Readable, Faithful, Used: Three Dissociable Properties of Demographic Identity in a Language Model

Fathin Difa Robbani

Large language models are widely used to simulate survey respondents, yet their answers are homogeneous and unfaithful to real inter-group differences. We ask where demographic group identity lives inside an LLM, how faithfully its geometry mirrors real inter-group opinion structure, and whether it uses what it encodes. Using representational similarity analysis against Pew ground truth over 169 demographic cells, we score 1,089 read-out locations in Mistral-7B and intervene causally across six attribute types. Four results. (1) The standard last-token residual read-out understates the model: attention-head read-outs dominate it in five of six types, with selection-corrected fidelity up to rho=0.63 -- roughly 70% of the measurement-reliability ceiling -- surviving a lexical-similarity control. (2) A single head (L11 H16) is significantly faithful in all six types as a fixed location, while race-based types stay weak and prompt-fragile. Both phenomena replicate -- the analogous head significant in five of six types, weakest on the same race type -- across three checkpoints of a second model family, where ten billion training tokens barely move the map. (3) Causal use does not follow fidelity: the clearest causal pathway sits in one of the least faithful types (p=0.002, cluster-robust, fixed depth), the most faithful type shows no correction-surviving single-layer effect, and replacing the entire identity moves predictions by under 2% of their error. (4) A 128-dimensional probe of the single head lands 21-31% closer to survey truth than the model's own answers -- yet recovers almost none of the per-question group ordering, no better than the answers themselves. Readable, faithfully arranged, and causally used are three dissociable properties of the same model; treating them as one claim is what keeps the "can LLMs simulate populations" debate unresolved.

arXiv

Gradient Mirage: Trainable yet Label-Unidentifiable Gradients in Large Language Model Split Learning

Shiyu Miao, Yunlong Mao, Zirui Huang, Liang Yao, Tianshuo Zheng, Yanhui Gu, Fan Liu, Sheng Zhong

Gradient matching attacks (GMAs) in LLM split learning (SL) rely on a critical yet underexplored assumption: the gradient exposed at the split interface is a faithful derivative of the client's full-label training objective. This gradient-objective consistency allows a curious server to recover private labels by searching for a sequence whose induced gradient explains the observation. We propose Gradient Mirage, a defense that breaks this consistency without discarding the optimization utility of the backward signal. Our key idea is to induce the adversary to solve a misspecified inverse problem, in which no plausible label sequence in the sequence space can explain the observed gradients. Concretely, Gradient Mirage achieves this by inducing inconsistency across three dimensions: objective, direction, and scale. Selective Autoregressive Supervision derives the exposed gradient from a masked surrogate loss rather than the full-label objective assumed by the attacker; Scale Blinding then applies randomized multiplicative rescaling, obscuring the gradient's natural magnitude; and Directional Privatization further randomizes the gradient direction while preserving its magnitude through the von Mises-Fisher (vMF) mechanism under a directional metric differential privacy guarantee. Crucially, utility is preserved: the Top segment still learns from all target tokens via Dual-Track Backpropagation, the exposed gradient remains informative since each supervised token retains its complete autoregressive context, and Bottom-Gradient Recovery restores the effective gradient for Bottom-segment optimization. Extensive experiments show that Gradient Mirage provides substantially stronger protection than existing defenses under comparable fine-tuning performance, achieving a better privacy-utility trade-off.

arXiv

Learning Canonical Register Automata over Ordered Data Domains

Yong Li, Qiyi Tang, Di-De Yen

Register automata are finite automata equipped with memory that recognize data languages over infinite alphabets. In this work, we investigate active learning algorithms for deterministic register automata (DRAs) over ordered data domains--covering both dense domains, such as the rationals, and non-dense domains such as the integers. We show that the active learning problem for DRAs over both dense and non-dense ordered domains can be treated within a single unified framework. More specifically, we develop and implement a polynomial-time active learning procedure for DRAs over ordered domains, using oracles for membership, equivalence and memorability queries. The memorability queries were originally introduced for learning DRAs over domains with identity tests. Our unified framework also leads to a new consequence: minimization of DRAs over the non-dense ordered domain of integers is decidable, extending a result previously known only for dense domains. Finally, we give improved complexity bounds of several decision problems for DRAs over ordered domains that are closely related to the queries used in active learning.

arXiv

Metrics That Write Themselves: Evolving an Evaluator from Its Own Blind Spots

Xing Zhang, Yanwei Cui, Guanghui Wang, Zhihao Lin, Peiyang He

Agents improve quickly against a reliable automatic metric and stall without one, and the applications that need them most, report generation among them, are the ones nobody knows how to score. Can the metric write itself? Saying what makes an answer good is hard; pointing at something wrong with one is easier, so the metric we evolve is a pool of small Python operators that each flag a candidate for one named defect, or abstain, and vote. Asking a model for operators directly does not work: 183 candidates realise only 96 distinct behaviours, from one narrow region of an enormous space. EvalCEGAR instead borrows counterexample-guided abstraction refinement from program verification. It reads the pool as an abstraction and searches for a collision, two answers the operators score identically, one correct and one not. That pair, not a prompt, is the authoring request, and when a collision defeats every attempt the loop widens what an operator may read rather than resampling. On MBPP+ and HumanEval+, a sandbox whose hidden unit tests give exact ground truth, the loop writes a 55-line operator that closes 15.4% of the gap between flagging nothing and a perfect filter on 428 unseen tasks (+0.0065, p=0.0010) at a quarter of our best hand-written operator's flags. On the benchmark it never saw it matches that operator's effect exactly on a third of the flags. Six of eight runs admit such an operator and all six help out of sample; our 15 hand-written operators applied together as one filter lose accuracy. An LLM judge on the same information ties that delta on a nearly disjoint set of candidates, and charges a model call per candidate forever where the operator charges none.

arXiv

Execution-grounded evaluation reveals hidden failures in language-model calculations for environmental science

Maohao Ran, Chendong Ma, Yanting Zhang, Dailing Jiang, Yusen Huang, Meng Gao, Jun Song

Large language models are increasingly used for quantitative work in the environmental sciences, yet existing evaluations score only final answers, leaving calculation process unobserved. Here we introduce AtmosCoder-Bench, an execution-grounded benchmark that makes the calculation process visible. Built through a transferable semi-automated pipeline (436 problems, 3,910 variants, 7,029 graded quantities), every problem is validated to be unambiguous and human-solvable, with uniquely verifiable answers. We find that (i) multiple-choice formats inflate measured accuracy by at least 12 percentage points; (ii) many failures arise not from missing knowledge but from models failing to apply known formulas and constraints consistently throughout multi-step computation; and (iii) even frontier models remain weak when task-specific conditions invalidate familiar methods, often reverting to canonical solution patterns rather than adapting methods to the relevant physical regime, leaving expert oversight essential.

arXiv

Budget-First Tariff Recommendation (BFTR): A Complete Algorithmic Framework for Telecom Plan Recommendation without Overcharging

Ghislain Dorian Tchuente Mondjo

Telecom operators traditionally offer predefined tariff grids, forcing users to choose from a limited set of plans. This paper proposes BFTR (Budget-First Tariff Recommendation), a complete algorithmic framework integrating eight Budget-First strategies, including two original hybrid approaches: Recursive Hybrid (conditional interpolation) and Knapsack-First Hybrid (priority knapsack). Unlike existing approaches that adjust prices upward to guarantee a minimum margin, BFTR guarantees the absence of overcharging by systematically aligning the final price with the catalog reference price. We mathematically formalize each strategy, prove the existence of an offer for any positive budget, and prove that the price deviation (surcharge) is zero for all strategies that do not use interpolation with correction. A detailed comparative analysis confronts BFTR to ten main existing tariff models on ten dimensions. Experiments on a dataset of 974 customers inspired by the Nigerian MTN market show that: (i) Recursive Hybrid is optimal for the customer (100% budget used, 29.9 GB volume, utility 0.946, 0% overcharging), (ii) Piecewise offers the highest volume (39.7 GB) with 0% overcharging, (iii) Power Law provides an excellent compromise (99.9% budget, 38.1 GB, 0% overcharging). All strategies achieve a zero surcharge, confirming the theoretical guarantees. A sensitivity analysis on the weighting parameter alpha (0.2 - volume priority, 0.5 - balance, 0.8 - budget priority) shows that utility rankings evolve logically. Execution times (< 10 ms) and very low failure rates (0% for robust strategies) confirm the operational viability of the system. The formal proof of the absence of overcharging constitutes a major theoretical contribution.

arXiv

MemFuse: Multi-Source Memory Fusion from Fragmented Observations

Chao Li, Yuanfa Li, Wenhao Wu, Xule Liu, Zhi Wang, Kun Shao

Long-term memory is essential for agents that operate across extended interactions, yet existing memory systems and benchmarks predominantly focus on single-source textual histories. In realistic settings, however, relevant information is often fragmented across applications and devices, as well as across users and time, requiring agents to integrate dispersed observations into coherent episodic memories while preserving their source provenance. To address these gaps, we introduce **MemFuseBench**, a benchmark for *multi-source memory fusion*. MemFuseBench is built with a Scene-to-Sensor pipeline that synthesizes controllable scenarios into source-tagged observations, evidence-grounded questions, and adversarial distractors. It enables systematic evaluation of temporal reasoning, cross-source evidence fusion, and robustness to noise. We further propose **MemFuse**, a structured memory system that preserves source-level evidence in event-layer atomic memory and organizes related atomic events into cluster-layer fused memory within a causal fusion graph. During retrieval, MemFuse retrieves and organizes related evidence fragments while maintaining traceability to original source events. Experiments on MemFuseBench show that MemFuse achieves the best overall performance among the evaluated memory systems under all three LLM settings and consistently improves performance on questions requiring cross-source evidence fusion.

arXiv

Learning What to Fail On: Failure-Mode Contextual Bandits for Adversarial Data Curation

Roie Kazoom, Ofir Cohen, Rami Puzis, Asaf Shabtai, Ofer Hadar

We introduce a failure-aware adversarial retrieval-augmented framework for improving robustness in natural language understanding. Rather than selecting synthetic examples with a fixed reward threshold, our method formulates adversarial data curation as a failure-mode contextual bandit problem. Candidate examples are generated with retrieval-augmented prompting, filtered by the current target model, automatically validated by an LLM judge ensemble, and clustered into recurring failure modes. A stochastic policy then selects which failure modes to sample for retraining, and is updated using validation-based reward that balances robustness gains, forgetting, and data cost. This makes the data curator itself the learning agent, enabling adaptive selection of the most useful model failures across training rounds. On standard benchmarks, our approach improves RoBERTa-base accuracy from 88.48% to 92.60% on SNLI, from 75.04% to 80.95% on ANLI, and from 54.67% to 71.99% on MultiNLI, while consistently outperforming prior adversarial augmentation methods. We further demonstrate transfer to FEVER fact verification, achieving up to 79.86\% FEVER score and 82.45\% accuracy with RoBERTa-large. Finally, we provide a theoretical interpretation showing that, under stated assumptions, failure-mode sampling can reduce shortcut-aligned gradient contributions while inducing bounded distributional drift. By combining retrieval, automated validation, contextual-bandit failure selection, and controlled adversarial retraining, our framework enables scalable robustness improvement without additional human annotation.

arXiv

X2Streaming-TTS: Causal Token-Level Text-to-Speech from Streaming Text with Speech-State Inheritance

Rime Wen, Zehan Liu, Shawn Qin, Lights Shi, Roy Gan, Hao Wang, Qian Wang

Streaming text-to-speech is essential for low-latency spoken dialogue systems, yet many systems wait for sentence-level text and are therefore only pseudo-streaming. True token-level synthesis must generate speech from uncertain prefixes while maintaining perceptual continuity over an unbounded stream with bounded context. We present X2Streaming-TTS, a causal TTS framework that consumes asynchronously arriving text tokens and emits speech without accessing future input. To handle uncertain prefixes, we introduce causal commitment, which keeps ambiguous expressions provisional through uncertainty-aware buffering and performs capacity-adaptive, punctuation-aware segmentation. To preserve acoustic continuity, we further introduce causal speech-state inheritance, which carries the complete Code2Wav state and selected historical Talker states across segment boundaries. Together with an attention prior constraint, it blocks access to future positions while retaining bounded acoustic context. Experiments show that X2Streaming-TTS outperforms existing pseudo-streaming models on most subjective and objective metrics. Further analysis shows that causal commitment stabilizes online segmentation and reduces failures caused by insufficient context, while speech-state inheritance improves boundary continuity without degrading naturalness or speaker identity. X2Streaming-TTS thus achieves strict token-level synthesis with quality comparable to the evaluated offline baselines, a median time to first audio token (TTFT) of 15.8 ms for a single request, and a median TTFT of 260.8 ms at 128 concurrent requests. Our implementation is publicly available at https://github.com/X-Square-Robot/X2Streaming-TTS .

arXiv

TranslatePsy-AfriSLM: High-Quality Data Scaling For Low-Resource Machine Translation

Milan Gritta, Patrik Lambert, Jihye Back, Amril Nazir

The rapid progress in Artificial Intelligence has largely bypassed African languages, creating a digital divide that limits AI adoption on the continent. Recent open-source LLMs systematically underperform on African language machine translation, while the lack of large-scale, high-quality, open-source parallel data has constrained the development of competitive small language models (SLMs). We introduce TranslatePsy-AfriSLM, a collection of open-source machine translation resources for 19 Sub-Saharan African languages, including curated parallel data, African-specialized synthetic data, and a family of fine-tuned SLMs. Our empirical study shows that unified quality-estimation filtering removes up to 96% of training tokens without degrading quality, and that filtered synthetic data dominates the quality-efficiency Pareto frontier. Fine-tuned on the resulting data mixture, TranslatePsy-AfriSLMs outperform substantially larger systems, including TranslateGemma-27B and Qwen3.5-122B-A10B, with as few as 0.8B parameters.

arXiv

When Safety Overrides Vision: Exploring Dynamics between Vision Influence and Safety Alignment in Vision-Language Models

Mehak Gupta, Tanmoy Chakraborty

Aligned vision-language models (VLMs) are designed to balance grounded visual reasoning with safe generation behavior. However, we observe a striking phenomenon: under safety-constrained instruction, models frequently abstain from answering questions that remain correctly answerable under default instruction despite receiving identical image-question inputs. This raises a fundamental question: does safety alignment suppress perceptual grounding itself, or does visual evidence remain internally available while generation is redirected toward abstention? In this work, we investigate the internal decoding dynamics underlying safety-induced abstention in aligned VLMs. Across multiple architectures and multimodal benchmarks, we show that abstained generations remain consistently influenced by visual evidence throughout decoding, indicating that perceptual grounding is largely preserved despite refusal behavior. We further demonstrate that, although the representational organization of refusal differs substantially across architectures, safety-constrained instruction consistently alters late-stage hidden-state dynamics toward refusal-oriented decoding. Finally, through targeted activation-level interventions, we show that suppressing refusal-related representations reliably restores grounded answering behavior across models without retraining or modifying visual inputs. Together, these findings reveal a previously underexplored failure mode in aligned VLMs: safety alignment can override grounded visual expression even when perceptual evidence remains internally preserved.

arXiv

Can a Lightweight Multimodal Model Estimate LLM Reasoning Performance? A Study for Compute-Optimal Document Inference

Zishan Ahmad, Vishal Vaddina

Uniformly allocating inference reasoning budgets to LLMs is expensive and prone to over-thinking penalties; especially in document tasks where visual layouts drive complexity. To address this, we introduce BudgetDoc, the first multimodal benchmark providing explicit supervision for model-budget-performance trade-offs across three document tasks. Using BudgetDoc, we train DRB (Document-Reasoning Balancer), an approx. 1B-parameter pre-flight estimator (SigLIP-2 + Qwen3-0.6B) that predicts ordinal model performance across budget levels, achieving a 0.753 weighted F1. When dynamically allocating reasoning budgets across five frontier models and three datasets, DRB matches or improves F1 scores compared to always-maximum-budget baselines in 9 of 15 configurations while drastically reducing cost. Finally, preliminary evaluations demonstrate DRB's potential to generalize to cross-model selection.

arXiv

From Storage to Access: Verifiable Activation of Parametric Knowledge in LLMs via Explicit Priming and Implicit Reasoning

Zuocheng Ying, Yang Yang, Yumou Wu, Chuanbo Zhu, Jiarui Wang, Ziqi Wu, Jingming Cai, Junqing Yu, Zikai Song

Although Large Language Models (LLMs) encode rich factual knowledge in their parameters, reliably recalling and verifying such knowledge remains a key bottleneck in factual question answering. Existing end-to-end methods entangle knowledge elicitation with reasoning, making it difficult to determine whether correct answers arise from parametric knowledge or the input context. To address this challenge, we propose VAKE (Verifiable Activation of Parametric KnowledgE), a two-stage reinforcement-learning framework that externalizes latent parametric knowledge through explicit Priming and transfers the acquired elicitation capability to implicit Reasoning. Given a query and an insufficient retrieved subgraph, the Priming policy explicitly inserts bridging triples as verifiable evidence, with supervision provided by rewards derived from answers generated by a separate frozen model over the augmented subgraph. Building on the policy learned during Priming, the Reasoning stage trains the model to answer from the original input, testing whether the capability acquired through explicit knowledge elicitation transfers to implicit reasoning. Experiments across seven benchmarks and models from 3B to 14B show that VAKE consistently outperforms standard baselines, including when transferring directly from HotpotQA to OOD datasets. LLM-based evaluation further shows that over 80% of the inserted triples provide factual bridging knowledge not derivable from the retrieved context, while more than half elicit knowledge inaccessible through direct prompting. These results suggest that VAKE activates latent parametric knowledge rather than copying the input context or memorizing dataset-specific associations.

Keep browsing

More papers from top AI conferences

30+ top conferences, by conference, year, and topic.