# Research Papers — 2026-10 · bundle page 35 of 156

> Full-text markdown of papers #1701–1750 of 7799 published in 2026-10, newest first. AI-readable mirrors of the full 7799-paper 2026-10 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-10/page/35
[← Previous bundle](https://aipod.dev/papers/archive/2026-10/page/34.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-10/page/36.md)

### Back to the Future: Reverse-Consequence Modeling for Long-Horizon Particle Dynamics

Authors: Leyan Zhang, Nantian He, Jiajun Ma, Yanpeng Zhao, Yunbo Wang
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18168v1

Accurate prediction of object trajectories is fundamental to mechanical understanding and robotic control. However, autoregressive models inherently suffer from cumulative errors, leading to severe drift in long-horizon rollouts. Observing that in many physical systems with dissipative or convergence-to-rest behavior, the final states are often more predictable than transient dynamics (e.g., a sliding object resting under friction or a pendulum stabilizing at equilibrium), we propose ReStep, a cascaded approach that reverses the conventional forward-prediction paradigm. ReStep first employs a stochastic model to directly predict the final state, then leverages Anchored Diffusion to generate goal-conditioned intermediate trajectories. By conditioning on both ''boundary states'', it effectively constrains dynamics and reduces uncertainty, producing physically plausible and temporally coherent transitions. We further integrate ReStep dynamics as the physical engine within a 3D Gaussian Splatting pipeline, enabling a unified physical-visual framework for intuitive inference from images. Extensive experiments on diverse rigid and non-rigid collision datasets show that ReStep achieves superior long-term accuracy, substantially mitigating error accumulation via reverse-consequence modeling.

---

### Lifting Finite-Sample Gaussian Approximation of Wide Neural Networks to Functional Convergence

Authors: Luis Wiederhold, Nil Ayday, Debarghya Ghoshdastidar
Published: 2026-10-13
Categories: stat.ML, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.18166v1

Randomly initialised neural networks of width $m$ converge to Gaussian processes as $m \to \infty$, with two types of quantitative bounds characterising this convergence. Finite-sample bounds compare the network with its Gaussian limit at a fixed finite sample set, on an arbitrary domain, in the quadratic Wasserstein distance $\mathcal W_2$, achieving a convergence rate of $m^{-1/2}$ with explicit constants. However, these constants grow with the number of samples, so the bound becomes uninformative as the sample set is refined. Functional bounds remove that dependence by comparing the two as random functions, at the price of spherical input, metrics weaker than $\mathcal W_2$, or slower rates. In this paper, we therefore lift a finite-sample bound to a functional one. The lifting refines the sample set until it is dense in the domain, so the bound must first be made independent of the number of samples. We achieve this by measuring the error per sample, which makes the constant uniform. For a polynomial activation and any compact input set, this gives a rate of $m^{-1/2}$ in the supremum norm, with an explicit constant.

---

### Random Probing for Structural Self-Interactions in Graph Neural Networks

Authors: Saku Peltonen, H. Çağrı Bilgi, Kubilay Atasu
Published: 2026-10-13
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.18164v1

Graph Neural Networks (GNNs) fundamentally lack the ability to capture structural properties defined by node self-interactions, such as cycle counts, subgraph centrality, and heat kernel signatures. Existing solutions often rely on computationally expensive higher-order architectures or hand-crafted features. In this work, we introduce DIAL, a message-passing layer that gives nodes access to graph structure through the diagonal of graph-derived operators. Our approach uses randomized probing to provide nodes with learnable, permutation equivariant access to diagonal entries. We prove that our approach can express important structural features, including cycle counts, subgraph centrality, and heat kernel signatures. The proposed method can be integrated directly into message passing layers. Experiments on molecular property prediction, cycle counting, and pattern detection demonstrate the effectiveness of DIAL across a range of graph learning tasks.

---

### LoOPSD: On-Policy Self-Distillation with Looped Teachers

Authors: Wei Fan, Qingyue Yang, Chao-Hong Tan, Qian Chen, Wen Wang, Xiangang Li, Kejiang Chen, Weiming Zhang, Nenghai Yu
Published: 2026-10-13
Categories: cs.LG, cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.18162v1

On-policy self-distillation improves a model using guidance from the model itself, but it still requires a teacher that provides more useful predictions than the student. Existing approaches obtain such guidance by giving the teacher additional context or instructions. We ask whether it can instead come from changing how the same model uses its own layers. To address this, we introduce LoOPSD, which searches over a model's layer execution order to construct a stronger teacher and distills its advantage into the student. Specifically, we keep the model weights unchanged while searching for layers to repeat, and use the resulting teacher's predictions on student-generated prefixes to guide learning. We then alternate search and distillation, rebuilding the teacher from the updated student in each round so that the student progressively absorbs the gains from alternative layer execution. Experiments on mathematical and logical reasoning tasks show improvements across models. On GSM8K with a 400-token thinking budget, LoOPSD raises Qwen3-1.7B accuracy from 42.30% to 79.61%; subsequent distillation rounds yield further gains beyond the first round. Together, these findings show that by searching layer execution to construct teachers, we can turn a model's additional computation into improved student capabilities, opening a new avenue for self-distillation to derive supervision from the model's own computation.

---

### Benchmarking Multimodal Large Language Models for Zero-Shot Human-Object Interaction

Authors: Oriol Rabasseda, Zenjie Li, Kamal Nasrollahi, Sergio Escalera
Published: 2026-10-13
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.18161v1

Multimodal Large Language Models (MLLMs) have demonstrated strong zero-shot reasoning across diverse domains, but their potential for Human-Object Interaction (HOI) remains unexplored. Unlike traditional HOI approaches that rely on extensive labeled datasets and fixed interaction taxonomies, MLLMs offer flexible reasoning, making them well-suited for dynamic, real-world scenarios. In this work, we present the first systematic study of MLLMs for HOI recognition and detection within a pipeline that combines an external object detector for entity localization with zero-shot multimodal reasoning for interaction understanding. We introduce a benchmark framework for principled zero-shot evaluation, consisting of (i) a unified taxonomy of zero-shot prompting strategies, and (ii) a methodology that transforms token-level logits into interaction scores, enabling fair comparison with specialized HOI models under standard metrics. To support prompt-based evaluation, we additionally construct new question-answer annotations for standard HOI datasets, transforming them into a format suitable for MLLMs. Building on this, we propose two visual prompting strategies, namely Objects as Context and Black Other Objects, that guide MLLMs to target objects while preserving scene context. Experiments on HICO, HICO-DET, and V-COCO demonstrate state-of-the-art zero-shot performance, achieving 53.50 mAP on HICO and 23.69 mAP on HICO-DET. Overall, our work provides a principled benchmarking framework for MLLMs in HOI, opening the door to new models and datasets and enabling flexible open-world interaction understanding.

---

### Claim Framing as a Driver of LLM Sycophancy

Authors: Jean Seo, Pepa Atanasova, Isabelle Augenstein
Published: 2026-10-13
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2610.18159v1

Sycophancy in Large Language Models, the tendency to align responses with user preferences over ground truth, has largely been studied using questions without a clear ground truth and with a single, plainly stated user opinion. We present a systematic study of which framings of a user's claim trigger sycophancy in factual question answering, where every answer is objectively verifiable without extra context. Evaluating 14 models on 42K questions, we express correct and incorrect user opinions through various claim framings over 8 dimensions and 21 types, and measure a broad set of opinion-driven behaviors. We find that all models are susceptible to user opinions regardless of size. Models are most influenced by how confidently an opinion is stated, whom it is attributed to, and whether it is presented as shared between the user and the model. In contrast to prior work, framings based on the user's expertise, authority over the model, or social pressure matter little in our factual setting. Larger models resist incorrect opinions better but are no better at recovering from their own errors. Our results call for sycophancy evaluations that probe how an opinion is framed, not merely whether one is present, and that measure multiple behaviors beyond a single flip rate.

---

### When and How Order-Flow Interventions Work After Dealers Adapt

Authors: Guanghao Wu, Zhuo Cai, Zhidong Li, Shoujin Wang
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18158v1

Learning dealer agents can settle on persistently wide quotes, raising what customers pay to trade. A venue can route more orders to tighter quotes, but a wide quote may reflect real risk, and once dealers adapt, an end-to-end comparison cannot show why a rule works: because it intervenes more, picks better moments, is announced in advance, or rewards the right quote. We propose SOFA (State-aligned Order-Flow Allocation), a public order-flow contract with three parts: a State Scorer ranks market states from lagged public data, a Commitment Scheduler announces a budgeted intervention strength before dealers quote, and a Quote-to-Flow Allocator shifts a bounded share of fixed customer demand toward tighter quotes. Our paired re-adaptation protocol gives competing rules the same intervention strengths, only re-timed, lets dealers re-adapt to each rule from a shared checkpoint, and breaks one link at a time. Across 20 independently trained dealer populations in an ABIDES market, concentrating the intervention in states that the score ranks as low-risk lowers low-risk execution cost relative to random timing in every population, by 0.040 tick on average, while a stale announcement or a broken quote-to-flow link raises this cost. In the primary setting the gain comes at the expense of high-risk trades, whose costs rise, so state alignment pays off only for a venue that weighs low-risk costs more than 1.38 times as heavily as high-risk costs. Our code is available at https://github.com/moderncavemann/state-aligned-order-flow-contracts.

---

### Fan Out or Relay? Ownership-Aware Routing for LLM Agents Querying Siloed Tables

Authors: Zhexi Feng
Published: 2026-10-13
Categories: cs.CL, cs.AI, cs.MA
arXiv: https://arxiv.org/abs/2610.18157v1

Inside organizations, a question's data often sit with several owners, who query their own tables but never hand them over. An LLM coordinator can fan out, sending the question to every owner at once, or relay, querying owners in turn and reusing earlier rows. We show that neither is safe in advance, because data ownership decides which one is right. Across thirteen settings on PeopleJoin and BIRD with three model families, fan-out is more accurate in twelve on questions one owner can answer, yet answers at most 4.3% of those joining two owners' tables. The flip follows ownership, not query shape: on BIRD two-table joins, fan-out answers 32-44% when one owner holds both and 1-4% otherwise. Relay wins by conditioning on returned rows, and hiding them from an otherwise identical planner cuts two-owner accuracy from 23.2% to 5.1%. Since ownership is public, we propose Ownership-Aware Routing (OAR), which drafts a schema-only SQL query, counts the owners whose tables it joins, and routes the question before any private row is read. In ten settings OAR is never significantly less accurate than the best fixed protocol, and beats every fixed protocol on PeopleJoin, by up to 9.3 points on held-out organizations.

---

### When Do Learned Blocks Preserve Causal Meaning? Block Identifiability and Constructive Causal Abstractions

Authors: Karim Zaghw, Wieland Brendel, Patrik Reizinger
Published: 2026-10-13
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.18155v2

Standard identifiability theory assumes one-dimensional latent factors, but learned representations can encode latent information across multidimensional blocks whose internal coordinates are not uniquely determined. When such blocks are used to represent and intervene on higher-level concepts, it is unclear whether their causal meaning survives the transformations left unresolved by block-identifiability. We connect block-identifiability to constructive causal abstraction. We show that block-diagonal invertible transformations followed by compatible coordinate permutations form the maximal class that preserves every constructive causal abstraction built on those blocks. This guarantee assumes that the correspondence between blocks and concepts is updated with the representation. A small synthetic experiment with additive autoencoders shows that high-level interventions can be transferred between independently learned representations while approximately preserving their intended effects.

---

### Learning a Group Element from Its Action: Galois Correspondences, Optimal ERM, and an Abel-Ruffini Threshold for the Natarajan Dimension

Authors: Xiaoyu Li
Published: 2026-10-13
Categories: cs.LG, math.GR, stat.ML
arXiv: https://arxiv.org/abs/2610.18154v1

We study learning an unknown element $g$ of a group acting with finite orbits on a countable set, from examples $(x,gx)$. After a sample on $A$, the consistent hypotheses are the restrictions of a coset of the pointwise stabilizer of $A$, and the sample forces exactly the labels of points this stabilizer fixes; for a Galois group, they form the field generated by the data. Through this correspondence, multiclass dimensions become group invariants: the DS and graph dimensions equal the height, the Littlestone dimension the longest irredundant sequence length, and the teaching dimension the minimal base size. Among deterministic learners, symmetry makes empirical risk minimization optimal: on nondegenerate classes every consistent learner, whatever its tie-breaking, attains the optimal realizable sample complexity up to constant factors, and every consistent online learner the optimal mistake bound. Agnostically, every ERM's sample complexity exceeds the known lower bound by at most the DS-to-Natarajan ratio, up to constants. For finite Galois extensions of countable fields, the Natarajan dimension equals a group invariant, the coset-cube number, which is at least the number of complemented factors of any chief series of any subgroup. Hence the dimensions coincide for every solvable Galois group, while for the groups $A_5$ and $S_5$ of the quintic they are $2<3$ and $3<4$, and $A_5$ is the smallest group with a gap. For the Galois group $S_n$ of the general equation of degree $n$, they coincide exactly when $n\le 4$, i.e., in characteristic zero, exactly when the equation is solvable by radicals; the implication is one-way, as the simple group $\mathrm{PSL}(2,8)$ has no gap. For linear, orthogonal and unitary maps of $\mathbb R^d$ or $\mathbb C^d$ both dimensions equal $d$, and a two-fibered pseudo-cube is a disjoint union of Natarajan cubes exactly when its monodromy group is abelian.

---

### Attacking Only Half the Data: Two-Sided Evaluation of AI-Text Detector Robustness

Authors: Apu Chandraw Shill, Rabindra Nath Nandi
Published: 2026-10-13
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.18151v1

We investigate the robustness of widely used AI text detectors under attacks applied to both human and machine text. Prior evaluations typically attack only machine text, overlooking that human text is routinely modified through grammar correction, formatting, or invisible character insertion. We introduce a two-sided protocol that applies attacks to both classes. Across two corpora and nine detectors, reported AUROC gains vanish under two-sided evaluation, with attacked human text misclassified at rates up to 100%. While simple preprocessing reverses character-level attacks, word-level synonym substitution remains effective meaning-preserving, and irreversible.

---

### RadixFlow: Filling Execution Bubbles in Prefix-Cached Hybrid LLM Serving

Authors: Huihao Jing, Wenbin Hu, Shaojin Chen, Haochen Shi, Haoyu Huang, Sirui Zhang, Yuxuan Liu, Zhongwei Xie, Haoran Li
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18143v1

Prefix reuse improves LLM inference by sharing the KV cache of common prompt prefixes. This helps only if the scheduler can execute a matched branch in the current worker round. Schedulers assume that a matched prefix is ready, which holds for full attention but fails for hybrid LLMs. There, a branch (the consumer) may need recurrent state that another branch (the producer) is still computing in the same round. The native scheduler stops its waiting-queue scan at this blocked consumer, and the resulting head-of-line blocking creates \emph{execution bubbles}: idle batch slots that independent work could fill. We introduce RadixFlow, which decouples prefix matching from state readiness. Its state-ready traversal turns a global scan barrier into a local wait: only the consumer is blocked until the next round, while feasible independent branches keep filling the batch under the native state check. On five hybrid models, RadixFlow lowers mean latency by 3\% to 24\% and worker rounds by 11\% to 44\% at almost unchanged reuse. On the larger Qwen3.5 models, ablations show that most of the gain comes from traversal, which adds little on a full-attention control. Heavier rounds limit how much of this saving reaches latency.

---

### FlowCred: Hierarchical Credit Assignment and Functional Coordination for Joint Text-to-Video Post-Training

Authors: Mingxuan Zhu, Yanwei Huang, Shiyuan Liu, Xinzhi Zhang, Shujie Li, Yuanxing Zhang, Lu Chen
Published: 2026-10-13
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.18142v1

Prompt optimization and generator post-training are two complementary ways to improve text-to-video (T2V) generation, and recent systems increasingly train both modules jointly. However, optimizing a prompt policy and a video generator from the same terminal reward creates two ambiguities: prompt quality is confounded with stochastic generation, and the two modules may realize overlapping local improvements. We introduce FlowCred, a two-level framework for joint T2V post-training. First, hierarchical credit assignment uses matched prompt-seed rollout grids to separate prompt-level credit from prompt-conditioned generator residuals under shared randomness. Second, functional coordination probes alternative current-policy prompts at states from actual generation trajectories, providing generator-aware evidence for prompt optimization while selectively attenuating generator updates already supported by conditioning changes. Across multiple T2V benchmarks, FlowCred consistently improves matched joint-training baselines. On LTX-2.5, FlowCred improves a strictly matched joint-training control from 61.47 to 64.46 on T2V-CompBench and from 55.58 to 58.08 on StoryEval, without increasing decoded-video or reward-model-call budgets. Factorial and realized-update analyses show complementary contributions from prompt shaping and generator routing, while human evaluation and transfer across backbones and updater families provide further supporting evidence.

---

### Mitigating Cross-Modal Hallucinations by Correcting Modality Misrouting in Omnimodal LLMs

Authors: Yu Mi, Licheng Pan, Jiayi Zhu, Qirui Liu, Yitong Qiao, Changtong Xu, Tao Wei, Fei Sun, Zhixuan Chu, Kui Ren
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18141v1

Omnimodal large language models (OLLMs) are susceptible to cross-modal hallucinations under audio-visual interference. We find that, for a substantial fraction of incorrect joint predictions, the question-relevant unimodal branch can still recover the correct answer, suggesting that task-relevant evidence can remain recoverable from the relevant modality even when it is not reflected in the joint prediction. We refer to this recoverable failure pattern as \emph{modality misrouting}. We propose CaRe (Conflict-aware Residual Intervention), a training-free inference-time method for identifying and correcting modality misrouting. CaRe uses a text-only modality probe to identify a candidate modality-specific counterfactual branch, and flags potential misrouting when its prediction conflicts with the original joint prediction. For such cases, CaRe computes a residual direction from the discrepancy between the counterfactual and fused representations and uses it to steer the original multimodal state toward the relevant modality evidence, while retaining both modalities in the original audio-visual inference path. The intervention strength is adapted based on the confidence of the counterfactual branch. Experiments on CMM and AVHBench across two OLLMs show that CaRe consistently outperforms the evaluated inference-time hallucination-mitigation baselines.

---

### SparPR: Runtime KV-Cache Recovery for Resource-Constrained LLM Inference on Consumer Hardware

Authors: Jieyun Deng, Keming Hu, Yanghe Liu, Li Feng
Published: 2026-10-13
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.18140v1

Running large language models (LLMs) on consumer hardware is bottlenecked by the key-value (KV) cache, which grows with every decoded token and quickly exhausts the memory and bandwidth of a single consumer GPU. We present SparPR (Sparse Partial Recomputation), a runtime KV-cache recovery mechanism that does not modify model weights and requires no offline training. SparPR couples three online decisions: Query-Aware key-block selection scores historical blocks against the current query; dynamic budget control estimates per-block copy and recompute costs to decide how many blocks to retain; and on-demand local recomputation with cost-aware fallback merges retained blocks into continuous spans and switches between sparse recomputation and full recovery according to a runtime cost comparison. Implemented on FlexGen/FlexLLMGen, SparPR is evaluated on a single consumer laptop (RTX 3070 Ti 8GB, Ryzen 9 6900HX, 16GB RAM). It cuts end-to-end latency by 31%, 52%, and 63% at 1k, 2k, and 4k prompt tokens respectively, with 1.4x-2.7x higher throughput, and completes an 8k-token workload in 44 minutes where the FlexGen baseline times out. Under a fixed cache budget, SparPR retains a Needle Accuracy that stays well above StreamingLLM and H2O at every keep ratio, and ablation studies confirm that the three mechanisms act jointly.

---

### Revisiting Error Feedback for Model Quantization

Authors: Wai-Yiu Keung, Chi Cui, Wing-Kin Ma, Jesús Cid-Sueiro, Hei Victor Cheng
Published: 2026-10-13
Categories: cs.LG, eess.SP
arXiv: https://arxiv.org/abs/2610.18139v1

We study causal error-feedback coefficients as design variables for model quantization. Modeling instantaneous quantization errors as independent, zero-mean variables with a common fixed variance, we formulated a metric-dependent trace problem for designing feedback coefficients. The formulation resembles fixed-order GPTQ/LDLQ feedback and supports Jacobian-aware designs that account for downstream error propagation. For a positive-definite metric $\boldsymbol{H}$, we prove that the joint rotation-feedback trace minimum is $n\det(\boldsymbol{H})^{1/n}$ and present an attaining pair using the geometric mean decomposition. Experiments on synthetic networks, trained MLPs, and language models show benefits from downstream-sensitive feedback, but also reveal that trace optimality need not improve finite-level performance. Small QAT experiments further illustrate direct loss-driven feedback tuning with fixed or trainable network weights.

---

### Human-Inspired Asymmetric Collaboration for Dual-Student Semi-Supervised Medical Image Segmentation

Authors: Huiyi Ren, Ziming Fang, Jiaming Cai, Rui Xu, Haojie Li
Published: 2026-10-13
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.18133v1

Dual-student learning enriches predictive diversity in SSMIS by introducing a strongly augmented view alongside a weakly augmented view. Insufficient strong augmentation may homogenize the students, whereas excessive augmentation can corrupt anatomical evidence and destabilize collaboration. This challenge echoes feedback-regulated human instruction, where learners receive differentiated tasks and the task difficulty is continually adjusted according to their evolving responses. Inspired by these, we propose Human-Inspired Asymmetric Collaboration (HAC), a closed-loop dual-student framework that maintains weak--strong learning through adaptive difficulty calibration and geometry preservation. Specifically, Reinforcement-guided Difficulty Calibration (RDC) formulates strong-augmentation control as a window-level reinforcement learning process, using augmentation history, realized perturbations, and training responses to dynamically adjust the augmentation magnitude, operation number, and magnitude range. Meanwhile, the Gramian Geometry Preservation Constraint (G2PC) preserves the early-stage three-branch geometry through global volume and pairwise area constraints, preventing shared supervision from homogenizing the two students. Comprehensive evaluations across LA, Pancreas, and BraTS-2019 validate both the segmentation effectiveness and capability-aware difficulty calibration of HAC.

---

### Distance Prediction for Quantum Codes

Authors: Nadav Schneider, Eliya Nachmani
Published: 2026-10-13
Categories: quant-ph, cs.LG
arXiv: https://arxiv.org/abs/2610.18130v1

The distance of a quantum code is a fundamental metric that dictates its error-correction capability. While quantum low-density parity-check (qLDPC) families such as Bivariate Bicycle (BB) codes offer a promising pathway to low-overhead fault-tolerant quantum computing, computing their distance is challenging. Current exact solvers and even search heuristics that provide a distance upper bound suffer from severe computational bottlenecks, restricting the discovery of optimal codes. In this paper, we construct a dataset of BB codes with corresponding distance upper bounds obtained through high-budget heuristic searches. We introduce BB-Net, an attention-based neural network architecture for predicting the distance of BB codes. It achieves 96.39% accuracy on the reference distances in the test set, compared with 94.86% for state-of-the-art heuristics with a budget of up to $10^{4}$ times more computational runtime. On a subset with certified exact distances, BB-Net achieves 98.94%. By relieving the distance-calculation bottleneck, BB-Net enables faster screening of large code spaces, particularly in ML-driven search pipelines, and thereby can accelerate the discovery of novel quantum error-correcting codes, an important capability for advancing scalable fault-tolerant quantum computing. Our code is publicly available: https://github.com/NadavSc/BBNet

---

### Monitoring latent reasoning in looped language models

Authors: Pawel Olszowiec, Paulina Tomaszewska, Maciej Żelaszczyk, Michal Byra, Neo Christopher Chung
Published: 2026-10-13
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.18127v1

Looped language models (LMs) reuse one block of layers several times per token generation, promising more latent computation at no additional storage cost. Despite their rising popularity, the analysis of looped LMs is still in its infancy: little is known about whether the iterative transformations actually improve their capabilities, or whether these models remain stable under prompt perturbation. We construct an inspection protocol rooted in Mechanistic Interpretability comprising five diverse lenses. We apply it to examine eight looped LMs from four families. Models differ in many aspects but the key one is whether the prompt is transmitted only within the looped state (carry loops: Ouro, LoopUS) or re-supplied at every iteration (re-read loops: Parcae, and Rec-OLMo & Huginn from the Raven family). The results of our study provide valuable insights concerning safety and efficient use of looped language models.

---

### Fresh at Decision, Stale at Action: Routing New Information to LLM Decisions Already Made

Authors: Ayoub Merimi, Souhail Bakkali, Abdeljalil Sakat
Published: 2026-10-13
Categories: cs.AI, cs.GT, cs.IR, cs.MA
arXiv: https://arxiv.org/abs/2610.18126v1

Freshness shapes an LLM decision at the moment it is made; the action it commits to is latent, carried out later, in a changing world. We therefore introduce a timestamp theory of decision validity and a decision ledger that realises it. Six timestamps define a decision; a freshness ceiling bounds what up-to-date evidence guarantees at action time, placing the remaining risk in information published afterwards. Accuracy then factorises into two capabilities: routing, which delivers that information to affected decisions before their revision deadlines, and temporal judgment, which reads it for the time of action. The ledger is persistent state holding each decision's times, evidence and background knowledge, indexed by watch conditions written at decision time. Controlled decision streams with exact labels confirm four predictions across five LLMs, from 8B to frontier scale: freshness alone yields exactly the accuracy the ceiling predicts; models act on decisive news once it arrives, so accuracy follows routing; the ledger's index sets what routing reaches, and watch conditions reach indirect dependencies at low exposure; and the remaining errors concern when a change takes effect. Freshness makes a decision right when it is made; routing and temporal judgment keep it right when it is carried out.

---

### H-DDCM: Hadamard Codebooks for Fast Generative Image and Video Compression

Authors: Or Ronai, Vladimir Kulikov, Tomer Michaeli
Published: 2026-10-13
Categories: cs.CV, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.18125v1

Training-free diffusion-based compression methods repurpose the generative priors of a diffusion or a flow model to compress data without retraining them as task-specific codecs. Recent codebook-based approaches such as Turbo-DDCM achieve strong perceptual quality at ultra-low bitrates, but their reliance on dense Gaussian codebooks remains costly in both memory and computation, and scales poorly with the large latent dimensions of modern image and video generators. Their combinatorial bitstream encoding further becomes computationally heavy at this scale. We introduce H-DDCM, a scalable generative compression framework that replaces Gaussian codebooks with implicit randomized Hadamard dictionaries. The fast Walsh-Hadamard transform allows for implicitly synthesizing full-dimensional dictionaries with $O(d \log d)$ compute for projection and reconstruction and $O(d)$ memory, without materializing the full-dimensional dictionary. We further introduce an efficient table-free lexicographic coding scheme that makes combinatorial encoding practical at larger dimensions. Beyond images, we adapt the framework to an autoregressive video flow model and develop a streaming compression scheme that periodically re-encodes recently reconstructed frames into a fresh KV-cache prefix, preserving temporal context across segments at no additional transmitted bits. This enables long-form, close to real-time generative video compression on a single GPU. Across image and video benchmarks, H-DDCM achieves competitive rate-distortion-perception performance while substantially reducing the memory and computational overhead of codebook-based training-free generative compression.

---

### Cheap, Consistent, Calibrated, but Biased? Typed Decision Models as Judges

Authors: Fabian Lukassen, Christoph Weisser, Alexander Silbersdorff
Published: 2026-10-13
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.18122v1

Typed decision models (TDMs) return a probability for each allowed answer instead of generating text, which makes them candidates for cheaply replacing LLM judges that compare two answers. Whether they share the biases of LLM judges is untested. We run the CALM benchmark's bias tests on eight TDMs and seven LLM judges of 2024 and 2026. Every TDM is biased, in three to eight of twelve tests. Emotional tone affects them most, and rewriting the better answer in such a tone lowers the accuracy of a typical TDM from 99% to 42%. TDMs are more biased than 2026 LLM judges. All eight change verdicts when the answers swap places, which the 2026 LLM judges hardly do, and several side with an answer that most people allegedly prefer or that a famous model allegedly wrote. Yet both kinds of judge are tripped up by the same prompts. The TDMs' probabilities show that most tests mainly flip verdicts the TDM was unsure of, and that good calibration does not protect against bias. As TDMs are more consistent and 37 times cheaper than 2026 LLM judges, our results suggest they can be attractive judges if their biases are tested for the task.

---

### Narrative Lookahead in LLMs: Event-Level Decodability, Predictive Advantage, and Steering

Authors: Donghoon Jung, Min Kim, Seojin Choi, Songeun Chae, Seohyon Jung
Published: 2026-10-13
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.18121v1

Do language models carry information about narrative events before writing them, and how far ahead? We examine narrative lookahead through salient, state-changing events rather than tokens or response-level attributes. We extract 23,355 LLM-annotated events from 2,000 stories generated by four open-weight LLMs, labeling them along three nonexclusive dimensions of change: within agents, between agents, and in the story world, with human validation on a subset. Event segmentation and plot-independent labels enable comparisons across stories at horizons measured in events rather than in tokens. At event ends, we probe hidden states for upcoming event types and compare predictions with text-only predictors using matched prefixes. Across four models, future event types are decodable above prevalence baselines for at least four subsequent events, but probes significantly outperform both primary text-only predictors only for the next event. In exploratory steering with OLMo-3.1-32B, a probe-derived direction shifts the proportion of the target event type in both intended directions at higher strengths. Our event-level approach links the prediction of future narrative developments to interventions that reshape subsequent event composition.

---

### Certifying Almost-Sure Reachability in Unknown MDPs

Authors: S Akshay, Suguman Bansal, Gaurav Mahla
Published: 2026-10-13
Categories: cs.LO, cs.AI
arXiv: https://arxiv.org/abs/2610.18119v1

{\em Specification-Guided Reinforcement Learning (RL)} has largely concerned itself with \emph{learning} an optimal policy against a formal specification in an unknown {\em Markov Decision Process (MDP)}. We take up the dual question. Given a policy already in hand, can one \emph{verify} that it is correct against a formal specification in an unknown MDP, using sampling access alone? Such a question arises naturally in safety-critical systems. We investigate it for the {\em almost-sure reachability property}, whose correctness is witnessed by a {\em ranking supermartingale (RSM)} certificate. Our first finding is negative. In the absence of any further information about the MDP, one can neither learn such a certificate nor verify a candidate one with probabilistic guarantees. On the positive side, we show that, under a mild restriction, a known lower bound $p_{\min}$ on nonzero transition probabilities, a correct certificate \emph{can} be learned with PAC-style guarantees. To the best of our knowledge, this is the first study of learning and verifying certificates for almost-sure reachability in unknown MDPs from sampling access alone. Finally, our experiments show that the $p_{\min}$-based sufficient condition, while sound, is conservative: the learned certificate often remains valid on the true MDP even when the condition fails.

---

### When Missingness Follows the Graph: Beyond Feature Propagation for Missing Features

Authors: Enrico Chialà, Francesco Ferrini, Antonio Longa, Bruno Lepri, Filippo Maria Bianchi, Manfred Jaeger, Veronica Lachi
Published: 2026-10-13
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.18113v1

Graph Neural Networks (GNNs) commonly assume that node features are fully observed, despite missing features being pervasive in real-world applications. By leveraging the dependence between node features and graph structure, existing graph-specific methods handle missingness by propagating observed feature values along graph edges. However, these methods have mainly been evaluated in overly optimistic settings, where feature entries are removed independently and uniformly across the graph. This overlooks realistic scenarios in which missingness propagates among connected nodes, as can occur during localized sensing failures in sensor networks. We formalize this complementary notion as graph-structured mask dependence, where the joint distribution of missingness indicators is shaped by the graph, and introduce a controlled mechanism that instantiates this setting. We study propagation-based methods theoretically and empirically in this setting, showing that their effectiveness changes substantially compared with the uniformly random regime. Motivated by this limitation, we propose PEMix, designed to remain effective under challenging graph-structured missingness. Rather than propagating observed values, PEMix uses global structural representations to condition a probabilistic model of incomplete node features before message passing. Experiments under controlled graph-structured missingness and naturally occurring missingness show that PEMix remains robust across complex missingness regimes, with particularly strong performance as missingness becomes severe.

---

### SYNPRED: A Synergistic Approach to Multimodal Learning for Clinical Prediction

Authors: Ivana Janíčková, Yen Y. Tan, Thomas H. Helbich, Philipp Seeböck, Julieta Di Marco, Zsuzsanna Bago-Horvath, Ulrike Heber, Thomas Spiegel, Georg Langs
Published: 2026-10-13
Categories: cs.LG, cs.CV
arXiv: https://arxiv.org/abs/2610.18110v1

Multimodal medical data captures complementary aspects of disease. It provides a comprehensive basis for diagnosis and the identification of biological associations linked to progression or treatment response. Established contrastive learning approaches exploit shared information across modalities, but fall short in exploiting complementary, modality-specific signals relevant for a task. Here, we propose a synergistic multimodal variational autoencoder (VAE) framework that separates task-relevant unimodal and multimodal information from instance-specific variation task-irrelevant, allowing complementary signals across modalities to be leveraged for prediction. The method yields benefits even if modalities are missing at test time. We evaluate the method on a treatment response prediction task using medical imaging and RNA data. Empirical comparison with alignment-based and multimodal VAE state-of-the-art approaches demonstrates improved multimodal prediction performance, and consistent benefits for prediction from a single modality after multimodal training. The full code can be found here: https://github.com/cirmuw/SYNPRED.

---

### A CaLLM Judge is a Better Judge: Mitigating Bias in LLM-as-a-judge

Authors: Cesare Barbera, Luca Stradiotti, Lorenzo Cascioli, Lorenzo Perini, Niek Tax, Daniel Haimovich, Jesse Davis, Andrea Passerini, Andrea Pugnana
Published: 2026-10-13
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.18108v1

Large language models (LLMs) are increasingly used as judges to predict which of two candidate answers a human evaluator would prefer, providing a scalable alternative when human evaluation is too costly. Unfortunately, LLM judgments can exhibit systematic biases. Prior work has documented position, verbosity or model family of the candidate answers influencing the judge decision rather than their quality, while additional biases may remain undocumented. $\textit{Multicalibration}$ provides a natural framework to address this problem: each bias corresponds to miscalibration on an identifiable group, $\textit{e.g.}$, family bias manifests as overconfidence in answers generated by models belonging to the same family of the judge. The challenge is to fix miscalibration simultaneously across many, potentially unknown groups. To this end, we introduce $\mathrm{CaLLM}$, a post hoc framework that multicalibrates an LLM judge, thereby bounding all such biases simultaneously. We prove that $\mathrm{CaLLM}$ bounds the bias of every group whose error structure is captured by an available representation, and that augmenting it with explicit bias groups further reduces the residual bias. We then derive theoretical conditions under which $\mathrm{CaLLM}$ improves the judge's accuracy, rather than trading it for calibration. Finally, we evaluate $\mathrm{CaLLM}$ across pairwise preference benchmarks, showing that it reduces the multicalibration error on both specified and unspecified bias groups, while improving predictive performance of the judge.

---

### Chain of Thought Monitoring Can Fail Even When CoT Is Necessary

Authors: Lennart Bürger, Piotr Komorowski, Moritz Weckbecker, Evgenii Kortukov, Aakriti Jain, Sebastian Lapuschkin, Wojciech Samek
Published: 2026-10-13
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2610.18105v1

Chain of Thought (CoT) monitoring is used by frontier labs to detect misbehaviour of AI systems. Yet, its reliability remains poorly understood. We attribute this to the lack of a successful organizing principle that leads to generalizable conclusions about monitoring reliability in different regimes. Recent work attributes monitoring failures to the CoT not being necessary for the answer, and argues that CoT becomes monitorable when it is necessary. We show that this binary necessary/not-necessary organizing principle is too coarse by demonstrating that monitoring failures persist even if the CoT is necessary for the answer. In response, we introduce a more fine-grained organizing principle based on a hierarchy of four CoT necessity levels. Building on this hierarchy, we develop a framework for measuring CoT monitorability at the three lowest levels via intervention experiments. We demonstrate monitoring failures at all three of these levels on realistic, hard prediction tasks. On a difficult medical-diagnosis task, patient sex and a clinical attribute influence the answer in at least 19% and 24% of correct responses with a necessary CoT, respectively, yet their influence is almost never articulated. These results show that even when CoT is necessary, it is not guaranteed to be monitorable.

---

### The Effect of Layer Ordering in Hybrid Architectures

Authors: John Cooper, Mingchen Ma, Jiaqi Mo, Shengqi Qiu, Nathaniel Bastian, Ilias Diakonikolas, Frederic Sala
Published: 2026-10-13
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.18104v1

Hybrid attention is rapidly becoming the dominant architectural choice for frontier models. Despite their successes, such architectures complicate model design by introducing additional choices, including the types and locations of layers. This combinatorially large design space has been studied empirically, but given the vast expense of training and evaluating models with different configurations, a principled understanding of the effects of layer selection and ordering is needed. We provide the first such analysis, showing that the dominant hybrid architecture is highly task-specific. Concretely, for any layer configuration, we prove that there exists a task where we can construct a model with this configuration that provably outperforms any other model of any different configuration under fixed resource constraints. Empirically, we observe that for standard training, learned hybrid architectures exhibit separations between layer orderings that match our theoretical findings. For instance, under the same memory budget, the optimal architecture predicted by our theory for a segmented retrieval task consistently achieves perfect accuracy, whereas the next strongest plateaus below 0.5. These results extend to deeper models, where ``containing'' the dominant architecture for a task indicates whether the model attains high performance. In addition, we investigate the mechanisms learned by hybrids, observing that Transformer layers contribute the largest share of performance on associative recall across architectures. This work establishes insights into task-dependent hybrid architecture design relevant to the training and scaling of frontier-scale hybrids.

---

### From Spectral Saliency to Finite Deletion: Evaluating Selective Language Model Edits

Authors: Ibne Farabi Shihab, Joyanta Jyoti Mondal
Published: 2026-10-13
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.18098v1

Deleting singular components offers a precise way to remove a capability from a language model, yet the first-order spectral saliency used to rank components predicts only a local loss change, whereas deletion is a finite intervention. In this paper, we study how prediction error affects selective editing, where selective editing requires degradation of a target task while preserving others. We decompose the error into derivative transfer across item splits and nonlinear response along the deletion path. We establish that outer interval screening retains every feasible candidate when every candidate's intervals contain its finite validation loss changes on every task. We evaluate these effects using one answer-only negative log-likelihood objective for search, validation, and test, with disjoint calibration and evaluation masks. The primary configuration targets arithmetic in Mistral-7B-Instruct-v0.3 while preserving code and question answering; extensions cover Llama-3.1-8B-Instruct and reversed target roles. In the primary study, which deletes 1024 coordinates from a support of 7168, calibrated intervals cover 968 of 1024 evaluation masks. Outer screening reaches the direct-selected edit after 25 verification checks, whereas direct verification abstains at a 32-check cap, but it rejects 4 of 74 feasible masks. All three selected edits pass validation, yet only one meets every test constraint. Interval coverage, feasible-candidate retention, and independent test preservation are therefore distinct properties and require separate assessment.

---

### APT: Feedback-Driven Agentic Post-Training for Text-to-Image Generation

Authors: Sirui Zhang, Fangyun Wei, Dong Chen, Jinjing Zhao, Yitong Wang, Yunuo Chen, Baining Guo, Zheng-Jun Zha
Published: 2026-10-13
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2610.18097v1

Post-training is crucial for improving text-to-image models, yet existing pipelines typically rely on manually curated training data and fixed training distributions. We introduce APT (Agentic Post-Training), a fully automated framework that adaptively post-trains a base model according to its evolving capability profile. APT coordinates three agents to construct evaluation sets from user requirements, diagnose model-specific strengths and failure modes, and generate reinforcement learning data with an adaptive prompt distribution that emphasizes underperforming capabilities. By repeatedly evaluating, diagnosing, and updating the generator, APT lets the current model's behavior shape its next training tasks, forming a feedback loop inspired by recursive self-improvement. For general-purpose settings without user-specified evaluation scenarios, we further develop an evaluation harness that enables broad and diverse evaluation set generation. Experiments across multiple base models, benchmarks, and generation architectures demonstrate that APT is model-agnostic, benchmark-agnostic, architecture-agnostic, and fully automated.

---

### CompanionSafe: From Multi-turn Safety Evaluation to Guardrails for AI Companions

Authors: Xinyang Liao, Lingyu Li, Kaicheng Shen, Yang Yao, Na Lu, Yan Teng, Xia Hu
Published: 2026-10-13
Categories: cs.AI, cs.CR, cs.HC
arXiv: https://arxiv.org/abs/2610.18095v1

AI companions provide users with emotional support and ongoing companionship, but their sustained relational interactions also introduce distinct safety risks. Prior work largely targets general harms or specific companion platforms and risk categories, while relying on manually collected conversations, limiting scalable and user-grounded evaluation and mitigation. To address this, we introduce CompanionSafe, a unified evaluation-to-mitigation framework for AI companion safety. Its evaluation component, CompanionEval, provides scalable multi-turn evaluation by constructing 1,300 relationally grounded settings from diverse personas, matched companion characters, and dialogue scenarios across four turn tiers. Its mitigation component, CompanionGuard, converts benchmark-exposed failures into turn-localized supervision for companion-specific risk detection. Our results show that even strong LLMs and existing guard models remain vulnerable to companion-specific relational harms, while both closed- and open-source models exhibit increasing safety risks as interactions become longer. Together, CompanionEval and CompanionGuard form a practical evaluation-to-mitigation loop for AI companion safety, turning relational safety failures exposed during multi-turn interaction into supervision for companion-specific guardrails.

---

### LifeSearch: Benchmarking Deep Research Agents for Real-World Life and Productivity

Authors: Yibo Wang, Senjie Jin, Yansong Ning, Xinbing Liang, Haotian Luo, Jiazheng Zhang, Yuxuan He, Will Wu, Clarenceai
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18092v1

Deep research (DR) agents address open-ended information needs through multi-step reasoning, long-horizon web search, and evidence synthesis. However, existing benchmarks primarily focus on academic research or synthetic tasks, leaving DR agents' capabilities in real-world life and professional productivity settings insufficiently evaluated. To bridge this gap, we introduce LifeSearch, a challenging benchmark comprising 203 Chinese and English tasks derived from real-user query seeds across 7 concrete life and productivity domains. With more than 4,500 hours of human annotation, we construct over 4,000 expert-written, fine-grained rubrics, covering both report-content criteria for coverage and accuracy and concrete criteria specifying values, units, dates, and URLs. We conduct extensive experiments on 12 advanced DR agents under a unified evaluation harness, systematically analyzing their performance patterns and failure modes. We find that even the best-performing frontier agent, GPT-5.6 Sol, achieves a score of only 63.4%, leaving substantial room for improvement on realistic life and professional DR tasks. We select the grading configuration by comparing candidate LLM judges against human rubric labels, and a human ranking study shows strong alignment between aggregate benchmark scores and human judgments (Spearman's $ρ= 0.89$, Kendall's $τ= 0.81$). LifeSearch provides a rigorous evaluation framework and foundation for advancing DR agents toward real-world life and productivity.

---

### AM-Bench: Understanding Agents under Task Underspecification across Models and Harnesses

Authors: Kaiyu Huang, Yiting Li, Angyang Gu, Pinjia He, Minchen Yu, Qingjiang Shi
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18090v1

Ambiguous or incomplete task specifications complicate both execution and clarification for language-model agents. Yet agents' responses to these information gaps remain poorly characterized across models and execution harnesses. We present AM-Bench, a benchmark built from 48 tasks adapted from Terminal-Bench 2.0 and Harness-Bench. Each task includes a fully specified reference and six variants spanning ambiguous or missing goals, context, and constraints. We evaluate every pairing of five models and four harnesses over 20,160 runs. Our results reveal distinct execution deficits: goal gaps hinder completion, context gaps increase token use, and constraint gaps weaken compliance without necessarily preventing completion. Missingness lowers success more than ambiguity in all three domains. These effects also reshape system comparisons: advantages under complete information need not persist, and rankings depend on model $\times$ harness pairings and the evaluation criterion -- completion, success, or repeat consistency. Estimates based on observed clarification requests and matched fully specified runs further suggest that clarification could recover unfinished work while leaving most constraint violations unresolved. Together, these findings motivate joint model-harness evaluation across information conditions and targeted clarification of unresolved requirements before dependent actions. To support future research on agents under task underspecification, we provide the AM-Bench project repository at https://github.com/s4-lab-cuhksz/AM-Bench.git.

---

### One-Sided Verification Under Limited Expert Review

Authors: Fan Nie, Haotian Ye, Shurui Liu, Zihao Wang, Stefano Ermon, James Zou
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18089v1

AI has become increasingly helpful in scientific discovery, yet human expert review remains essential for those problems without reliable automated evaluation, posing a fundamental challenge as candidate generation scales beyond available review capacity. This paper studies verification under limited human effort, where rare valid solutions need to be prioritized while invalid candidates should be filtered out to reduce human workload. We adopt the one-sided verification objective that maximizes the rejection rate of invalid solutions (TNR) under a constraint on the false-rejection rate of valid ones (FNR), making human review cost an explicit term to minimize. Drawing on the intuition that verifying a specific error can be easier than discovering one, we introduce OSVerify, an algorithmic framework that decouples error proposal from error validation. The error proposer searches for errors in solutions across different granularities by hierarchical decomposition, while the validator distinguishes substantial errors from minor or hallucinated ones. Experiments on competition- and research-level benchmarks show that OSVerify reduces the number of human reviews of invalid solutions by 46% on average compared to voting baselines under the same empirical low FNR constraint. Our post-training of the validator further improves the reduction to 57%. As high-impact real-world use cases, we apply OSVerify to identify errors in published research papers that are subsequently confirmed by the original authors and help resolve three Langlands open conjectures that are confirmed by experts, demonstrating its utility to facilitate human-AI research collaboration.

---

### ResearchFlow-Eval: Tracing the Execution Gap in End-to-End Multi-Claim Autoresearch

Authors: Peng Jiang, Dongyang Li, Quanying Liu, Xiaoxuan Jia
Published: 2026-10-13
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.18088v2

Autoresearch systems increasingly undertake entire research projects. In practice, scientific research requires systems to pursue multiple interdependent claims through a continuous, multi-stage workflow and revise experiments as evidence accumulates. Existing evaluations of autoresearch systems, however, typically score isolated tasks or final artifacts, leaving these process-level capabilities largely unmeasured. We therefore introduce ResearchFlow-Eval, a fine-grained framework for assessing continuous scientific execution and iteration at both the stage and claim levels. Each evaluation is grounded in a published anchor paper, from which we derive stage-specific rubrics that decompose each research stage into dimensions and separately scored rubric items. Using these rubrics, we score each research stage and examine experimental execution in detail, focusing on multi-claim completion and revisions across attempts. We apply the evaluation framework to general-purpose agent harnesses, skill-augmented harnesses, and systems with research-specific orchestration. Across all evaluated systems, experimental execution consistently lags behind design and planning. Within execution, systems struggle to produce high-quality results and meet claim requirements, and repeated attempts rarely yield measurable improvements. Moreover, neither skill augmentation nor research-specific orchestration reliably yields higher scores. ResearchFlow-Eval thus extends autoresearch evaluation beyond isolated, single-objective runs to continuous workflows in which systems pursue multiple claims and refine their work across iterations.

---

### Same Function, Different Trajectory: No Standard Optimizer Is Equivariant to Transformer Reparameterization

Authors: Bum Jun Kim
Published: 2026-10-13
Categories: cs.LG, cs.NE
arXiv: https://arxiv.org/abs/2610.18085v1

We show that two Transformers representing the same function can follow different training trajectories. For example, doubling a query matrix and halving its key matrix preserves the model's function but changes the gradient scales seen by the optimizer. Our theoretical analysis establishes that standard block-local first-order optimizers fail to scale updates with parameters under a shared learning rate. For AdamW, transforming the first- and second-moment estimates alone does not correct the mismatch between parameter and update scales. To make runs restarted from equivalent checkpoints follow the same trajectory, we balance matrix pairs and transform their moment estimates together, bringing distinct but equivalent training states to the same starting point. To correct scale imbalance between paired Transformer matrices without changing predictions, we propose Symmetry-Aware Gradient Geometry Stabilization (SAGGS), which periodically equalizes the Frobenius norms of exact symmetry pairs, and its damped variant, Soft SAGGS. SAGGS reduces perplexity (PPL) by $0.374\pm0.026$ at 101M parameters on WikiText-103, and Soft SAGGS reduces PPL by $0.780\pm0.040$ in a 31M modern-architecture replication, both under scale imbalance. In a 123.8M checkpoint-resume experiment, reciprocally rescaling matrix pairs barely changes validation loss at restart but increases final PPL by $3.78\pm0.42$; hard and soft rebalancing recover to within $0.02$ PPL of the control.

---

### AReaL-RSI: Adaptive Scaling and Coordination of Recursive Self-Improvement

Authors: Wenhao Zhou, Jiajun Song, Haitao Wang, Zipeng Qiu, Yingqi Peng, Ruida Xu, Jiale Li, Wenshuang Li, Shusheng Xu, Wei Dong, Hao Dai, Tongkai Yang et al.
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18083v1

Scaling recursive self-improvement requires additional computation to produce useful research progress in reaching persistent objectives. Intuitively, more researcher agents can explore more hypotheses, but early evaluation scores might misrepresent their promise, shared findings could lose their applicability conditions, and useful procedures could remain confined to completed investigations. In this paper, we present AReaL-RSI, a framework for organizing a persistent research team of agents without updating backbone LLM weights. We formulate the scaling problem in terms of evaluated outcomes and whole-system resource expenditure. Concretely, a leader agent selects task-appropriate research orientations and guides ongoing investigations; conditional memory sharing distinguishes unresolved problems from supported findings and enables recipient adaptation. During recursive evolution, after each campaign, evidence-based review revises researcher orientation, knowledge, and skills for later campaigns. Experiments compare AReaL-RSI with Codex-Goal on downstream harness optimization, GPU kernel optimization, and AI research tasks using two backbones. Under the reported resource allocations, AReaL-RSI improves Terminal Bench performance by 2.2 percentage points for each backbone and increases mean KernelBench speedup by 32.2-36.8%; further scalability comparisons favor the design of AReaL-RSI over native scaling: combining experience sharing with specialized researcher roles yields 1.9-27.5% lower kernel latency than independent exploration and 1.3-33.9% lower latency than sharing alone across all seven comparable task-backbone pairs.

---

### Failure Recovery via Belief Updates and World Modeling for Long-Horizon Agents

Authors: Xiyang Wu, Zhichao Liu, Shengxin zhang, Zongxia Li, Ming Li, Tianyi Zhou, Dinesh Manocha
Published: 2026-10-13
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.18082v1

Long-horizon agents often struggle to recover from failures whose causes lie several steps earlier, beyond what local reflection or backtracking can resolve. We introduce BELIEFWEAVER, a framework that combines persistent evidence across attempts with prospective reasoning over repair directions. It maintains a relational belief state over frontiers, attempts, observations, and checkpoints. Execution evidence updates related frontier priorities across attempts and rollback, while diagnostic evidence gates execution. An implicit world model (IWM) evaluates repair hypotheses, predicts progress, and identifies relations among frontiers to guide exploration. We further distill these recurring IWM reasoning components into a 9B student through supervised fine-tuning and on-policy distillation (OPD). Across three terminal-based benchmarks, BELIEFWEAVER improves aggregate final score by 5.1 percentage points over native Terminus-2 continuation, with positive trends in exploration efficiency. The distilled 9B student improves final score by 9.8 percentage points over the reported native baseline while reducing estimated teacher API cost by 54.8% relative to the matched teacher. These results show that persistent evidence and prospective reasoning can improve recovery in long-horizon agent tasks while substantially reducing reliance on teacher-model inference.

---

### Causal Discovery Under Hard Selection Bias: A New Robust Score-Matching Approach

Authors: Yiwen Qiu, Francesco Montagna, Shimeng Huang, Francesco Locatello
Published: 2026-10-13
Categories: cs.LG
arXiv: https://arxiv.org/abs/2610.18075v1

Selection bias is a difficult yet widespread problem in causal discovery, occurring whenever non-random selection processes lead to data that is not representative of the underlying populations. Due to its practical importance, related works have been proposed to address the problem under local (Versteeg et al., 2022) or interventional settings (Dai et al., 2025a), but general algorithms for observational data remain elusive. In this paper, we focus on truncated data (hard selection), and provide a general result for Additive Noise Models (ANM). We show that score-based methods are a surprisingly suitable candidate due to a special property of the score function ($\nabla \log p(x)$): deterministic truncation preserves both the score of the joint distribution as well as the Jacobian of the score. While established score-matching algorithms still fail in most settings, we actively leverage this observation to propose a new method for causal discovery that is robust under truncated data with ANMs. Overall, our approach offers a general, robust solution that is agnostic to external information about the selection process and still achieves comparable performance to the state-of-the-art approaches when selection is not present.

---

### LensErase: A Foundation Model for Erase Anything

Authors: Yunuo Chen, Xiuyu Wu, Qinhong Yang, Jinjing Zhao, Yanchen Dong, Sirui Zhang, Dong Chen, Fangyun Wei, Lei Shi, Ji Li, Dongdong Chen, Guo Lu et al.
Published: 2026-10-13
Categories: cs.CV
arXiv: https://arxiv.org/abs/2610.18072v1

We introduce LensErase, a general-purpose mask-based Erase Anything foundation model designed to cover a broad spectrum of real-world erase scenarios. LensErase supports diverse objects, visual text, watermarks, associated visual effects, multiple-object removal, and a wide range of object scales, aspect ratios, and image resolutions. First, to enable such broad coverage, we construct LensErase-100M, a large-scale dataset containing 101.07 million paired source-mask-target examples spanning diverse erase scenarios. Second, we pretrain LensErase from the text-to-image model Lens, adapting its strong generative prior to mask-based erasing. To efficiently incorporate the source image and erase mask, we propose Mask-Aware RoPE, a 3D rotary positional encoding that directly injects mask information into the source-image tokens without introducing a separate mask token sequence. Finally, we refine the model through supervised fine-tuning and reinforcement learning with contract-guided, region-aware rewards, and apply TraceFusion at inference to preserve high-fidelity content outside the effective edit region. Across six public benchmarks and the proposed LensErase-Bench, LensErase achieves substantial improvements under both reference-based objective metrics and reference-free VLM evaluation. We hope LensErase can serve as a foundational model for erase-related applications and systems.

---

### From Agents to Edges: Paired Optimization for Multi-Agent Collaboration

Authors: Feng Lin, Xiaotian Yu, Yifan Yang, Aijun Li
Published: 2026-10-13
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.18069v1

Strong individual agents do not guarantee effective collaboration. Communication must align what one agent conveys with how another interprets and uses it. This dependency makes the communication edge a natural unit of optimization. We introduce Paired Automated Communication Optimization (PACO), which jointly optimizes the instructions at both ends of each edge and their shared communication convention in systems built on black-box language models. PACO keeps model weights, agent roles, and workflow topology fixed. A state-aware discounted bandit selects edges for refinement, while execution feedback guides paired revisions evaluated through staged end-to-end execution. The optimized instructions integrate into existing agent prompts without additional model calls at deployment. On a four-agent workflow evaluated across diverse multi-agent benchmarks, PACO consistently outperforms leading prompt-optimization and communication baselines. Configurations learned on HotpotQA retain their advantage in cross-dataset transfer without target-specific optimization. These results demonstrate that optimizing how agents communicate can substantially improve collaboration within an otherwise fixed system.

---

### WorkRover: Scaling Long-Horizon Cowork Agent Training Via Recursive Environment Evolution

Authors: Hang He, Linting Wang, Guanming Liu, Bo Mao, Jinxiu Liu, Yiran Suo, Maosen Zhou, Zhiheng Xi, Guanting Dong, Zhou Shao, Rui Zheng, Jie Zhou et al.
Published: 2026-10-13
Categories: cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2610.18067v1

Scaling interactive environments is important for training long-horizon agents through reinforcement learning. These environments often require further testing and refinement after initial generation, and correcting defects in a task may also require updating its work context and verifier. However, existing construction pipelines offer limited support for such iterative refinement while keeping these components aligned. We introduce WorkRover, a framework for training long-horizon cowork agents through recursive environment evolution. Its harness-driven synthesis enables construction agents to jointly build and validate semantically aligned work contexts, executable tasks, and evidence-grounded verifiers, adapting their actions to workspace state and execution feedback rather than following a fixed construction workflow. To support environment evolution, we introduce Factorial Compatibility Replay (FCR), which uses rollout feedback to evaluate combinations of component revisions through dependency-closed replay and selects the smallest validated repair among tested candidates while preserving task requirements and evaluation constraints. To provide finer-grained training signals, we propose Verifier-to-Turn Credit Assignment (VTCA), which augments trajectory-level GRPO advantages with signals from deterministic workspace-progress checks and assigns a bounded share of positive-progress credit to prerequisite turns using recorded tool-output and artifact dependencies. Using WorkRover, we post-train Qwen3.5-35B-A3B to obtain WorkRover-35B-A3B and evaluate it on several cowork-agent benchmarks. Its reported scores exceed those of Qwen3.5-35B-A3B by 26.27 percentage points on JobBench, 29.32 percentage points on OfficeQA Pro, and 413.25 Elo points on GDPval v2. These results suggest that WorkRover offers a promising foundation for scalable training of long-horizon cowork agents.

---

### Uncertainty Evaluation: Bridging Machine Learning and Measurement Science

Authors: Dario Petri, Luca Mari
Published: 2026-10-13
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.18066v1

Reliable uncertainty evaluation is essential for assessing the trustworthiness of machine learning (ML) predictions, particularly in safety-critical applications. This paper explores conceptual analogies between uncertainty evaluation in ML and measurement science, drawing on the evolution of metrology from error-based approaches to the framework established by the Guide to the Expression of Uncertainty in Measurement (GUM). Particular attention is devoted to the distinction between process-level and result-level quality and to the limitations of the conventional classification of ML uncertainty into aleatoric and epistemic components. It is argued that this dichotomy, similarly to the traditional distinction between random and systematic measurement errors, is inherently model-dependent and provides an insufficient basis for a unified uncertainty evaluation framework. Further analogies between machine learning and measurement field of knowledge are identified concerning distribution shifts, anomalous operating conditions, and the validity of uncertainty evaluation. Building on the proposed considerations, the paper advocates a metrology-inspired approach based on the systematic identification, quantification, and propagation of uncertainty contributions. By extending the methodology already established in measurement science to ML, this approach could offer a potentially rigorous, consistent, and transferable framework for evaluating the trustworthiness of individual predictions.

---

### A Benchmark for Bangla Task-Oriented Dialogue Understanding

Authors: Afrina Alam Lamia, Sumiaya Sharif, Prottoy Modak, Ohiduzzaman Rayhan, Nafis Sadeq
Published: 2026-10-13
Categories: cs.CL
arXiv: https://arxiv.org/abs/2610.18060v1

Task-oriented dialogue understanding remains underexplored for Bangla, particularly in multilingual settings involving English and romanized Bangla (Banglish). We introduce BanglaTOD, a parallel tri-lingual benchmark comprising 209 multi-turn dialogues across nine Bangladesh-specific service domains and 64 intents. Each conversational turn is aligned across Bangla, English, and Banglish, yielding 4,689 language-specific turn records annotated with intents, token-level slots, state updates, and accumulated dialogue states. We establish context-aware baselines for joint intent classification and slot filling using mBERT, BERT-base-uncased, and BanglaBERT. Cross-language analysis reveals substantial performance differences across models and linguistic representations, while removing dialogue history reduces intent accuracy by 19.0--25.7 percentage points. BanglaTOD provides a foundation for multilingual dialogue understanding and future dialogue state tracking research in Bangladesh-specific conversational applications. The dataset is publicly available at \href{https://huggingface.co/datasets/afri7here/BanglaTOD}{Hugging Face}.

---

### From Policy Evaluation to Off-Policy Control: Provable Emergence of In-Context Q-Learning

Authors: Yingjun Mo, Yajie Zhao, Fang Kong
Published: 2026-10-13
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2610.18057v1

Transformers can implement reinforcement-learning algorithms in context, yet how off-policy control emerges within Transformers remains poorly understood. We study Q-learning, a foundational model-free control algorithm that combines off-policy bootstrapping with greedy policy improvement. We prove that a two-layer attention-only Transformer can implement a finite-temperature form of Q-learning, where the temperature $τ>0$ controls the approximation to greedy action selection, recovering window Q-learning as $τ\to0$. We train the structured Transformer on randomly generated MDP tasks using output-only supervision. Experiments show that the theoretically predicted parameter structure emerges during training, including the internal greedy action-selection mechanism, and that the trained Transformer achieves performance comparable to Q-learning on unseen tasks. Our theoretical analysis further shows that the equivalent realizations form a solution manifold and that, under suitable task-diversity conditions, the training iterates converge locally to this manifold at an exponential rate. For the random-task model analyzed here, the local worst-case number of gradient steps needed to reduce the distance to the manifold by a factor $ε$ is $Θ\left(τ^{-5}\log(1/\varepsilon)\right)$, quantifying the increasing optimization cost as the learned update approaches window Q-learning.

---

### Reliable Scale-Free Bandits via Empirical Likelihood

Authors: Pratham Gupta, Alhad Sethi, Shubhada Agrawal
Published: 2026-10-13
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2610.18055v1

We study stochastic bandits without knowledge of reward scales or tail parameters. Almost-sure regret guarantees describe eventual learning but do not quantify expected performance or the risk of large regret in a finite run. We introduce $\texttt{EL-F}$, an anytime empirical-likelihood (EL) extension of the $g$-Forcing policy of Cowan and Katehakis (2020), and characterize its performance at three levels. First, for integrable rewards and non-decreasing, diverging exploration schedules $g$ with vanishing increments, $\texttt{EL-F}$ preserves the $g$-Forcing asymptotics for pseudo-regret $R(T)$: $R(T)/g(T)\to\sum_{i:Δ_i>0}Δ_i$ almost surely as $T\to\infty$, where $Δ_i$ is the suboptimality gap of arm $i$. Second, on every sub-exponential instance, for non-decreasing $g(t)=ω(\log t)$ satisfying a mild feasibility condition, expected pseudo-regret is at most $g(T)\sum_{i:Δ_i>0}Δ_i+C_{\underline{P}}$, where $C_{\underline{P}}$ is an instance-dependent constant independent of $T$. Moreover, we also establish finite-time tail bounds with super-polynomial decay in regret beyond the forced-exploration budget. For slowly varying $g$, we determine the exact exponent of linear regret deviations on two-armed Gaussian instances with common variance and establish a matching lower bound among policies satisfying the same expected-regret budget on every such Gaussian instance. These quantitative guarantees also cover the greedy variant obtained at zero EL radius. A fixed positive radius introduces local optimism while preserving the leading asymptotic regret. We establish coupled pathwise dominance over the greedy variant in two-armed bandits with a deterministic suboptimal arm. Experiments demonstrate reductions in prolonged suboptimal allocation compared with $g$-Forcing on challenging instances.

---

### Privacy-Preserving GNN Inference on Financial Transaction Graphs via Homomorphic Encryption

Authors: H. Çağrı Bilgi, Kemal Mutluergil, Ceren Yıldırım, Kubilay Atasu, Erkay Savas
Published: 2026-10-13
Categories: cs.CR, cs.LG
arXiv: https://arxiv.org/abs/2610.18054v2

Anti-money laundering (AML) in financial networks requires modeling transaction patterns that span multiple institutions. In practice, however, banks cannot share account and transaction features due to strict privacy regulations. We propose a privacy-preserving graph neural network (GNN) inference framework tailored to this multi-bank setting. A central server holds the graph connectivity and pretrained model weights in plaintext, while each bank keeps its transaction features encrypted. Inference is then carried out entirely on encrypted features under the CKKS homomorphic encryption scheme, so that no plaintext transaction features or predictions are exposed to the server. The key to making this practical is an edge-centric GNN architecture, wherein message passing is expressed through operations on node--edge incidence matrices. This formulation not only preserves the edge-attributed multigraph structure of transaction networks but also reduces every aggregation step to plaintext--ciphertext matrix products. Since CKKS can evaluate only polynomial circuits, we also approximate each nonlinearity with a composite polynomial of the sign function, which we calibrate per activation site. On the IBM AML benchmark, our model achieves performance competitive with plaintext baselines, and with a consumer-grade GPU, classifies each queried transaction end-to-end in as little as 5.31 seconds, while per-transaction latency can be amortized to 1.65 seconds when optimized for throughput.

---

### LADDER: A Long-Horizon Robot Manipulation Benchmark

Authors: Petko Petkov, Giuliano Albanese, Nikolay Nikolov, Luc Van Gool, Jonathan Tremblay, Danda Pani Paudel
Published: 2026-10-13
Categories: cs.RO, cs.LG
arXiv: https://arxiv.org/abs/2610.18053v1

General-purpose robot manipulation policies learn short-horizon skills effectively, but degrade rapidly on long-horizon tasks. Existing benchmarks cannot isolate failures because task horizon is either too short or entangled with scene and object variations. We propose LADDER, a long-horizon robot manipulation benchmark scaling from 1 to 10 subgoals within the same environment over three suites (SimplerEnv-Long, RobotArena-$\infty$-Long, and RoboLab-Long), 145 tasks, eight categories, two embodiments, and 2,024 simulated episodes per system. We evaluate two orchestration paradigms: text subgoals for learned low-level policies (VoLo) and tool selection orchestration (CaP-X, GPT-Policy, PSP-Agent, TiPToP). Given the full instruction directly, low-level policies achieve 85.4% on single-subgoal tasks, but drop to near zero on extended tasks. While orchestration remedies this, it remains bounded by physical execution errors of the low-level policy rather than high-level planning and achieves only ~25%. Tool selection orchestration with GPT-Policy on GPT-6 Astra significantly outperforms learned-policy methods, but incurs significant time and cost budget and succeeds only ~62%. Real-world experiments across two embodiments confirm the collapse of the low-level policy and the gain from orchestration. Project website: https://ladder.insait.ai

---

### Can Agents Forget When Asked? A Benchmark for Memory Unlearning in LLM Agents

Authors: Bofan Xie, Chaokun Wang, Guanchen Ge, Songyao Wang, Fang Niu, Hang Zhang, Wei Wang
Published: 2026-10-13
Categories: cs.AI
arXiv: https://arxiv.org/abs/2610.18052v1

Long-term memory has become a core component of large language model (LLM) agents. As agents accumulate personal information over long-term use, users increasingly ask them to delete sensitive or outdated facts. A memory system should honor such requests. It should remove exactly what was requested and keep the rest of its memory intact. However, existing memory benchmarks evaluate only memory writing, retrieval, and organization, so whether agents can forget at the user's explicit request remains under-evaluated. To address this gap, we propose Bemula, the first benchmark dedicated to memory unlearning. It comprises 50 multi-session dialogue instances, 135 forget requests of six types, and 1,742 test questions. Each dialogue has a control version without the requests, so that only systems that have stored the target can earn credit for forgetting. Across the paired versions, five test suites measure storage, removal of the requested information, preservation of unrelated memories, relearning of similar facts, and resistance to extraction attacks. On Bemula, we evaluate six mainstream memory systems under three backbone LLMs, together with the Full Context LLM on two of them. The results are negative. Across all evaluated systems and backbones, no system passes more than 1.5% of forget requests end to end, and most pass none at all. These systems either leave the requested information extractable under simple attacks or delete far more than requested. Memory unlearning therefore remains an open challenge for agent memory design. We release our data and code at https://github.com/Nboxff/bemula-benchmark.

---