Action
-
Domain adaptation of a video Foundation model for Heart Failure Diagnosis from Echocardiography
2026-11-09 22:04:03Mahyar Bolhassani, Tijana Geroski, Narayana S. Singam, Amir AminiAbstract
The rapid advancement of foundation models has transformed computer vision, yet their adaptation to medical imaging, particularly purpose built foundation models for spatio temporal echocardiography videos, remains limited. In this work, we explore the feasibility of fine tuning a pretrained foundation model for video (TimeSformer) for automated diagnosis of reduced ejection fraction heart failure (HFrEF) from apical four chamber videos. The model is finetuned on the publicly available RVENet A4C dataset with a phase aware clip generation strategy that captures complementary systolic and diastolic dynamics. We employ divided space time attention, layer wise learning rate decay, and multi clip inference to enhance robustness under limited data. Experimental results show that full fine tuning of the proposed model outperforms classifier only tuning and a 3D CNN baseline. These findings highlight the potential of leveraging foundation video models for echocardiography based heart failure diagnosis and motivates further domain adaptation studies in medical imaging.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13297
PDF
-
Large Language Model Explainability for Planning Lunar Rover Trajectories under Uncertainty
2026-11-09 22:01:15Michelle Ho, Maximilian Adang, Zachary Booth, David E. Astle, Margaret Hansen, Edward Balaban, Mykel J. KochenderferAbstract
Planning planetary rover trajectories under probabilistic uncertainty is a multi-objective optimization problem that science teams currently iteratively solve and verify by hand. This process becomes impractical when timelines and constraints frequently shift throughout the mission planning phase. Sampling-based planners such as Monte Carlo tree search can speed up this process, but their outputs do not preserve traceability to the science objectives that justify each plan, which the science team needs to verify. We propose an interactive assistant based on a large language model (LLM) that can explain planner trajectories using domain vocabulary and re-run the planner under user-specified conditions to compare alternatives. Previous work has used LLMs inside planner decision-making loops, but we explore using them for post-planning explanation, where they are more likely to be trusted since they do not directly influence planner decisions. Our framework pairs a simulator inspired by the NASA Volatiles Investigating Polar Exploration Rover (VIPER) mission with a ReAct queryable interface that calls curated action tools and retrieval functions. We evaluate the assistant on a query bank consisting of plan comparisons, feasibility, and lookup queries. Experiments show that 90.9% of the assistant's responses are faithful to the planner's outputs, a nearly 65% improvement over zero-shot prompting of the same LLM without tool access, demonstrating the potential for trustworthy explainability of uncertainty-aware traverses with an LLM-based system.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13292
PDF
-
MRFoundry: Knowledge-guided agent loops for end-to-end MRI research
2026-11-09 21:54:54Ke Wang, Atakan Topcu, Hossein Momeni, Molin ZhangAbstract
AI agents can write code, use scientific software, and revise computational workflows, opening new possibilities for MRI research. Effective assistance requires connecting acquisition physics, data conventions, specialized tools, and scientific validation. We present MRFoundry, an open-source, extensible package that connects MRI knowledge to agent actions and output checks. Project-local memory preserves decisions and evidence, while reviewed contributions allow researchers to improve shared guidance. Researcher-supervised examples demonstrate skill-guided sequence auditing and acquired-data reconstruction, connecting tool use, output checks, and documented decisions. Alongside MRFoundry, we introduce MRI Agent Bench to compare agents with and without the package across models and MRI task domains. Across 240 paired Core evaluations spanning 24 task types and ten models, all-checks-passed completion increased from 67.5% without MRFoundry to 78.3% with it. Across 134 Hard pairs, completion increased from 36.6% to 44.8%. MRFoundry makes MRI workflow expertise available as reusable guidance that researchers can inspect, adapt, and extend within existing agent and scientific software environments. MRFoundry is openly available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13289
PDF
-
Decide, Ask, or Defer: Clinical LLMs under Incomplete Evidence
2026-11-09 21:39:31Mingzhan Yang, Weili WuAbstract
Clinical LLMs must decide not only what diagnosis to produce, but also whether the available evidence is sufficient for autonomous decision making. Binary DECIDE/ABSTAIN formulations merge distinct non decision states and do not explicitly evaluate information acquisition. We introduce a DECIDE/ASK/DEFER formulation together with a blinded protocol that prevents models from using evidence completeness metadata. We evaluate Qwen, Gemini, and GPT on 200 matched clinical evidence states constructed from DDXPlus. The models show substantial differences in action selection under identical evidence, with disagreement in 137 of 200 states. For Qwen, a matched targeted versus random analysis shows that selected information changes the likelihood of a subsequent autonomous decision more clearly than diagnostic correctness. Its matched DECIDE/ABSTAIN baseline further reveals a safety autonomy tradeoff: the three action policy rescues some erroneous autonomous decisions but also removes some correct autonomous deci sions. These results show that separating information acquisition from clinician deferral exposes behavior that binary abstention hides, without yielding a uniformly improved decision policy.
Abstract (translated)
URL
https://arxiv.org/abs/2610.04542
PDF
-
Adaptive Residual Quantization for Memory-Efficient Temporal Action Segmentation
2026-11-09 21:31:31Gerard Donahue, Guven Gergerli, Ayush Gupta, Reza Ghoddoosian, Enna Sachdeva, Faizan SiddiquiAbstract
Temporal action segmentation (TAS) is commonly trained on large banks of pre-extracted frame features, but storing and replaying those features becomes a practical bottleneck for long-form video, especially in incremental TAS where past-task data must remain available to mitigate catastrophic forgetting. Recent TAS replay and condensation methods reduce this burden, but they rely on dense frame-wise annotations and allocate storage through fixed segment-level structures, limiting their flexibility for long procedural videos whose frame complexity varies substantially over time. We introduce Adaptive Residual-Quantized Variational AutoEncoder (ARQ-VAE), a compact and label-free memory representation for TAS features that learns a discrete residual quantization of frame-level video features and then adaptively truncates the residual chain on a per-frame basis, retaining only the residual prefix that best reconstructs each frame feature. This yields a practical compact feature memory that does not depend on frame-wise labels and can therefore support multiple TAS settings with the same stored representation, while a lightweight fine-tuning stage further improves reconstruction quality for the truncated codes that are actually retained. Across standard TAS benchmarks, ARQ-VAE achieves a substantially improved storage-performance trade-off over prior replay and condensation baselines. On Breakfast, our final representation reduces training-set storage from 28 GB for the original I3D feature bank to 10 MB while retaining strong downstream segmentation performance; with ASFormer, it achieves 71.5 Edit and 51.2 F1@50 in the supervised setting. We further show that the same label-free memory remains effective for incremental and unsupervised TAS. We position ARQ-VAE as a practical memory representation for TAS and a promising direction for broader dataset condensation applications.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13262
PDF
-
NCP: Verifier-Guided Optimization of Weight-Space Control Policies
Abstract
We present the Neural Control Plane (NCP), a closed-loop system that observes language-model execution, synthesizes candidate weight-space intervention rules, evaluates them, and maintains a versioned table of admitted policies. NCP separates a pretrained model's inference data plane from the control policy that selects temporary parameter interventions. W-ISA supplies transactional execution and parameter restoration; NCP supplies telemetry, candidate synthesis, admission criteria, priorities, and rule lifetime management. The optimization state consists of active rules and an archive of evaluated candidates, while base-model parameters are restored after each request. In a local SmolLM-135M-Instruct experiment, NCP generated three candidates from development telemetry and evaluated them on separate task examples and a 4,095-target WikiText-2 slice. One projection reduced validation token repetition from 0.4133 to 0.2191, with a perplexity change of -0.0574%. All candidates produced zero newly correct exact answers, so the configured admission gate archived all three and retained the original active policy. Fresh-process repetition and saved-table resumption reproduced the measured outcomes. NCP provides an executable framework for studying recursive self-improvement at the control-policy level through explicit synthesis, verification, and state transitions.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13236
PDF
-
W-ISA: Transactional Execution of Weight-Space Intervention Programs
Abstract
We present W-ISA, a software instruction set and transactional runtime for executing weight-space intervention programs on pretrained language models. Shape-checked instructions express parameter updates, projections, ablations, and rank-one transformations. The runtime applies these programs within request-scoped execution contexts and restores saved parameters after completion or interruption. Activation-based match-action rules connect measured model states to intervention selection. We evaluate restoration and replay on SmolLM-135M-Instruct and Qwen2.5-1.5B-Instruct, and compare materialized rank-one updates with equivalent PyTorch hooks on the latter. In 45 timed requests, both intervention implementations produce identical greedy sequences at 32, 128, and 512 tokens; their maximum prompt-logit difference is 0.000184. On the complete WikiText-2 raw test split, the combined SmolLM attention-head ablation and MLP projection changes perplexity from 22.2353 to 22.3188, a relative increase of 0.3754%. W-ISA separates intervention specification, execution, and measurement, making parameter-level experiments reproducible and composable.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13232
PDF
-
Deletion Sensitivity Is a Property of the Model: A Three-Condition Measurement across Seven Open Reasoning Models
2026-11-09 20:51:21Hanzhang Jia, Liheng Zeng, Zhiyu Yang, Yi Gao, Bo MaAbstract
Reasoning models write a reasoning trace at every turn of a multi-turn tool session, and later turns read this history. Across 94k ShareGPT conversations, 73 percent of user interactions span multiple turns. Upstream providers bind each trace to its issuer, so a request forwarded across issuers arrives with its history deleted, and the task cost of this stripping remains unmeasured. To the best of our knowledge, this paper presents the first direct, blind, multi-seed measurement of full reasoning-history deletion in multi-turn tool calling, through three rendering conditions, seven open-weight reasoning models, three seeds, 794 tasks, and the official checker. The same deletion spans -4.4 to +51.3 points with both ends excluding zero, and for the one model re-dosed from 0 to 100 percent deletion, the reading varies by at most 1.5 points. Deletion sensitivity is a model-level property. The reading follows the model itself and stays constant across deployments, and a retrained or re-released model is a new model that a two-cell screen, one retention run and one deletion run, re-reads for the cost of two runs. Deletion sensitivity is thus a measurable property of the model, and the reading must enter routing and aggregation as prior information rather than persist as an invisible default cost.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13223
PDF
-
Evaluating Long-Horizon Robot Planning under Uncertainty
2026-11-09 20:46:30Harsh Muriki, Gayathri Aravindan, Yuchen Zheng, Sonia ChernovaAbstract
Household robots must successfully navigate im-perfect circumstances, both in their internal models and the external world. We introduce a benchmark of seven long-horizon household tasks in Habitat that vary initial robot memory and environment configurations. We evaluate adaptations of ReAct, VLM-TAMP, and Tru-POMDP within a shared execution framework across 90 tasks-environment-memory combinations, totaling 1,350 episodes. Our evaluation measures task success, partial completion, and action efficiency, along with rule-based analysis of execution traces. ReAct achieves the highest overall task success (45.3%) and completion (71.7%), while having accurate initial memory does not consistently yield the highest performance for every planner. We identify recurring placement errors, precondition errors, and redundant actions through trace analysis. These findings highlight differences in how the evaluated planners use initial knowledge and respond to uncertainty.
Abstract (translated)
URL
https://arxiv.org/abs/2610.17326
PDF
-
Wrong the Same Way: Cross-Prompt Sharing Both Suppresses and Amplifies Reward Hacking in RLVR
2026-11-09 20:39:36Michael Li, Rohan Pandey, Ray ChenAbstract
Two accounts of how reinforcement learning with verifiable rewards (RLVR) interacts with an imperfect verifier coexist: in one, reward hacking is inherited from the base model; in the other, RL amplifies it. We study both in a controlled environment small enough to compute the entire response distribution exactly. We vary the universality of a leak, or the fraction of prompts that accept the same wrong response, and the verifier's static false-positive rate. At its KL-regularized optimum, RLVR acts as a filter that reweights the accepted set without altering its composition, so the hack rate is inherited exactly. Taking that as a null model, a known decomposition writes the drift from it as a sum over prompts of covariances between reward and the empirical tangent kernel. The terms from other prompts vanish without parameter sharing. With sharing, every prompt that accepts a wrong answer pushes it up and every prompt that rejects it pushes it down through the baseline, in proportion to the model's probability of producing it there. A shared-parameter policy thus drives a leak that few prompts accept below the base rate and amplifies one that most prompts accept. At every convergent learning rate under Adam, SGD, and sampled GRPO, hacking ends higher at high universality than at low, but whether either end crosses the base rate depends on the step size. Independent per-prompt copies remove the dependence. The direction of this push depends on neither the verifier's error rate nor universality alone, but on how much of the model's kernel-weighted probability of a wrong answer falls on prompts that accept it, which frozen base model controls support. Verifier fixes should target the mistakes the model repeats across prompts: two patches that remove equally many errors differ by up to 38 points of true accuracy. We propose this alignment as a hypothesis for why earlier studies disagree.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13218
PDF
-
Before Words, Beyond Speech: Evaluating Nonverbal Social Reasoning in Early Childhood
2026-11-09 20:35:48Marie Amale Huynh, Laura Bravo-S\'anchez, Lauren Klein, Nick Haber, Philip Fisher, Serena Yeung-LevyAbstract
Children's earliest years are a period of intense social learning, during which meaning is built through gaze, gesture, and physical contact. Yet, this critical aspect of social interaction remains under-evaluated in multimodal video understanding, as existing benchmarks prioritize the conversation-driven interactions of adults. We address this gap with \bench ($\textbf{N}$aturalistic $\textbf{E}$arly-childhood $\textbf{S}$ocial in$\textbf{T}$eractions), the first child-centric benchmark for nonverbal social reasoning. \bench comprises \totalvids manually annotated clips ($\sim$12s) curated for diversity across naturalistic settings, cultural contexts, and developmental stages. Beyond a lack of data, targeted model improvement in this domain is hindered by coarse evaluation schemas that conflate basic perception with higher-order reasoning. To resolve this, \bench introduces a compositional schema that builds atomic interaction cues into progressive levels of social reasoning: contextual, behavioral, interpersonal, and open-ended descriptions. Extensive analysis of state-of-the-art Vision-Language Models (VLMs) reveals a stark performance collapse: despite strong contextual recognition, models exhibit substantial gaps in decoding precise behavioral and interpersonal cues. We trace these failures to specific biases, including poor consequence grounding and temporal sensitivity. Finally, we ground \bench in real-world applications by evaluating VLMs using domain adaptation strategies. Ultimately, \bench provides a rigorous testbed for advancing socially grounded artificial intelligence in early childhood. Project page: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13213
PDF
-
Ground Reaction Force-Aware Model Predictive Control for Underactuated Upright Physical Assistance
2026-11-09 20:29:14Darren Biskup, Jeonghan Yu, Zibo Yu, Sunil K. AgrawalAbstract
Many rehabilitation robots execute prescribed motions using reactive tracking controllers that do not explicitly incorporate the user's instantaneous human-ground interaction. This paper presents Ground Reaction Force-Aware Model Predictive Control (GRF-MPC), a framework for upright physical assistance. By integrating live GRF and center of pressure (CoP) measurements directly into a human centroidal dynamics model, GRF-MPC formulates a real-time quadratic program (QP) that optimizes assistive forces and moments over a receding horizon. We deployed GRF-MPC on the Robotic Upright Stand Trainer (RobUST) in an underactuated 4-cable configuration and evaluated it against a fully actuated 8-cable Cartesian impedance baseline in nearly 300 trials across six healthy participants. Across tasks spanning impulsive perturbations, lunge trajectories, and voluntary limits of stability, GRF-MPC reduced peak planar center-of-mass (CoM) error by 19.6% during perturbations, lowered trajectory tracking error by 24.1%, required 28.7% less net mechanical work, and permitted a 45.9% greater normalized voluntary excursion area. The underlying QP solves in under 1 ms, comfortably enabling 100 Hz replanning and demonstrating that incorporating live human-ground interaction into predictive control yields stronger, more transparent physical assistance with half the actuators.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13204
PDF
-
Structured Kernel Surrogate Models for Multimodal Attribution
2026-11-09 20:21:39Ziniu Zhang, Zhenshuo Zhang, Jianglin Lu, Ruoxuan Xiong, Yun Fu, Hongyang R. ZhangAbstract
We study the problem of cross-sample modality attribution, which involves measuring how one modality in one sample affects the learning outcome in other multimodal samples. This problem is crucial in a multimodal system because a data modality (such as a text description) may be redundant for one sample (a self-explanatory image), but the same data modality would otherwise be uniquely informative for another sample, or informative when combined with other modalities. Prior work focuses on influence estimation at the level of one sample, and it remains an open question how one should capture the interactions across both samples and modalities. To capture scenarios like the above, we propose structured kernel surrogate models, which are surrogate mappings from both samples and modalities to the prediction outcome when they are trained together. We design an algorithm, SKS, which applies kernel ridge regression with a structured, compositional kernel on the selection states of individual samples and modalities. Specifically, the kernel function multiplies kernel-based similarity scores between every pair of data modalities for one sample and a pair of two samples. Additionally, we design a gradient-based estimation algorithm to efficiently estimate the structured kernel surrogate model. We evaluate our approach on two multimodal benchmarks, spanning a range of data modalities including satellite images, road networks, text, audio, and tabular data. We find that SKS achieves $57.8$% higher correlation with the true prediction outcomes than existing methods. In downstream data selection, SKS improves the average accuracy by $1.93$% over baselines. Models with data selected by SKS have lower Hessian traces, indicating a flatter loss surface. Using SKS to select satellite images identifies road segments, where accident scenes commonly occur, as opposed to background images.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13198
PDF
-
From Process Models to Composable Agentic Workflows: Combining BPMN, Agents and Enterprise Knowledge
2026-11-09 20:18:32Christian Fillies, Jan Fillies, Frauke Weichhardt, Gay Anne Wood-AlbrechtAbstract
LLM-based agents can reason, use tools and collaborate. However, coordinating increasingly complex agentic applications becomes difficult to understand, monitor and improve. Decisions distributed across prompts, tool calls and application logic make workflows implicit. We introduce **Recursive Capability Composition**, enabling agents to execute business processes and expose them as reusable Agentic Tools. This paper presents a BPMN-based approach to Agentic Workflows, combining explicit orchestration with autonomous reasoning. SemTalk BPMN models are transformed into executable workflows using a LangGraph/Pregel runtime. Agentic Tasks introduce LLM reasoning at selected activities, while conventional BPMN controls process flows, deterministic decisions, system interactions and human approvals. Agentic Participants provide reusable interfaces for agent behavior or other executable processes. BPMN orchestration makes multi-agent systems more understandable and manageable. Explicit process models provide developers and operators with a shared view of workflow structures and execution progress, facilitating debugging, monitoring and optimization. Enterprise agents also need organization-specific context. Process Models and Knowledge Graphs describe activities, participants, information objects, decisions and relationships between processes. They provide structured knowledge of how an organization operates, supporting informed agent decisions and actions. Our contributions are threefold: 1. **Model-Driven Execution:** Transforming BPMN models into executable Agentic Workflows. 2. **Recursive Capability Composition:** Exposing agents and executable business processes as reusable tools for other agents. 3. **Enterprise Context:** Using Process Models and Ontologies to provide organization-specific knowledge for agent reasoning.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13196
PDF
-
AgentScript-Eval: Benchmarking LLMs and Agents for Code Generation in an Enterprise DSL
2026-11-09 20:04:32Dipin Khati, Shubham Mehrotra, Bin Bi, Zhou Yu, Denys Poshyvanyk, James Zhu, Sitaram Asur, Phil L. MuiAbstract
Domain-specific languages (DSLs) are widely used in enterprise software, giving a compact interface to complex tasks, from data access to infrastructure automation. Platforms increasingly extend them to agents, where a developer declares the topics, permitted actions, and guardrails of an agent rather than writing general-purpose code. Large language models (LLMs) write general-purpose code well, but that ability has been measured almost entirely on languages abundant in public pretraining data, and enterprise DSLs are neither abundant nor public. We introduce AgentScript-Eval, an execution-graded benchmark of 100 real tasks for AgentScript, the declarative DSL that Salesforce uses in Agentforce to build conversational agents. Tasks span authoring an agent from a specification, debugging a broken definition, and optimizing a working one. Evaluating five models zero-shot, we find that the regime is hard: the best model reaches only 25.0% pass@3, and every model scores 0% on authoring from scratch. We further find that model scale and price are weak predictors of success -- a 31B open-weight model attains the highest compilation rate of any model tested (48.0%) and matches the strongest frontier model on partial-credit accuracy at the lowest token cost. Adding an agentic compile-and-edit loop is the single largest lever, roughly doubling partial-credit accuracy. Beyond that gain, the wording of domain guidance matters more than its packaging. The same skill pack scores 28% as plain documentation but only 14% as a live plugin. Rewriting one skill inside that plugin then raises it to 37%, our best result, with catalog and tooling untouched. Larger models did not reliably outperform smaller ones; what mattered was the harness surrounding the model and the guidance supplied to it.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13183
PDF
-
Continuous Angular Control in Text-to-Image Generation via Geometric Token Encoding
2026-11-09 19:55:34Furkan G\"uzelant, Basavaraj Sunagad, Aysegul Dundar, Adam KortylewskiAbstract
Text-to-image diffusion models can render a photorealistic analog clock but cannot control the time it displays. Across four pretrained models, fewer than 1% of generated clocks place both hands within 6 degrees of the requested time; the models instead collapse onto visually dominant configurations such as the canonical 10:10 pose. The same failure appears when a specific fraction of a circular object is requested, where models return complete or conventionally sliced objects. We observe that these two failures share a geometric structure: both target quantities can be parameterized by angular variables. We introduce a geometric token encoding framework that conditions generation directly on these angles. Each angular component is encoded as a periodic sine-cosine feature and projected into a role-specific token embedding that is injected into the prompt before a frozen text encoder. To spatially ground these tokens without requiring localization input at inference, we introduce Stochastic Localization Relaxation (SLR), which stochastically relaxes attention supervision from component-level to object-level to unconstrained localization during training. For clocks, we additionally introduce a hand-swap preference objective that discourages assigning the correct angles to the wrong hand roles. On analog clocks our method reaches 63-78% Acc@6 degrees across evaluation methods, compared with below 1% for the baselines. On circular-fraction generation, it achieves 89.5% Acc@0.10, while maintaining 88.8% on unseen object categories and 84.9% on unseen target values, compared with roughly 20% for pretrained baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13172
PDF
-
AI Alignment Is Neither Solvable nor Sufficient
Abstract
Humans and AI systems increasingly act within the same social and economic environments. Alignment, the goal of ensuring that AI systems act in accordance with human values, is widely regarded as a central challenge for ensuring that AI benefits society while avoiding harm. We argue that this endeavor is misguided. First, alignment is not well defined, as it presumes a sufficiently stable and identifiable set of human intentions or values. Second, where a concrete notion of alignment can be specified, it is not solvable. Third, where alignment appears solvable, solving it would not necessarily mitigate the negative consequences of AI. A key reason is that alignment is typically framed as a property of individual AI systems, while the resulting hybrid societies of humans and machines are complex systems shaped by interactions and feedback loops. Emergent risks, and not only those posed by individual misaligned systems, are therefore likely to become a major source of harm. We conclude that AI safety is an ongoing challenge and not a problem that can be solved once and for all by verifying a particular model property.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13167
PDF
-
How Much of Your VLA Is kNN
2026-11-09 19:44:58Denis Grachev, Damien LaRocque, Omar MouredAbstract
Vision-language-action (VLA) models now exceed 95% success on the LIBERO simulation benchmark, yet it remains unclear how much of this performance requires generalization beyond the training demonstrations. In this work, we show that retrieving nearby demonstrations and reusing their action chunks achieves a success rate of 91.7 % on LIBERO without policy training. Under LIBERO-Plus perturbations, retrieval shows the smallest drop of all evaluated policies. Across nine publicly available VLAs, fine-tuning mostly makes the vision-language backbone a better retrieval key, and the actions of the VLAs remain close to the demonstrations. In addition, we introduce Local Action Consistency (LAC), an offline metric that measures whether nearby observations in representation space are also associated with similar future actions. We show that LAC tracks the success rate of retrieval and can guide the training of a retrieval key. Our evaluation and source code are available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13160
PDF
-
VIGOR: Benchmarking Visual Rationale Correctness in Multimodal Large Language Models
2026-11-09 19:30:03Yanlin Chen, Tang Li, Xi PengAbstract
Multimodal large language models (MLLMs) are typically evaluated by whether they produce correct answers, but correctness alone does not reveal whether their outputs are supported by the right visual evidence. We introduce VIGOR, a benchmark for evaluating VIsually GrOunded Rationales in MLLMs. VIGOR represents visual rationales as explicit concepts paired with pixel-level groundings, covering multiple abstraction levels including low-level visual properties, intermediate structures, objects, parts and high-level motion, relations. To construct VIGOR at scale, we develop an automatic annotation framework that reuses existing dense supervision when available and generates missing concept groundings through proposal-based annotation followed by conservative multimodal verification. Using VIGOR, we evaluate whether MLLMs "look at what they say" by comparing model-derived visual evidence for each expressed concept against annotated ground-truth rationales. Our results show that current MLLMs exhibit limited visual rationale correctness and reveal distinct grounding failures across concept levels. VIGOR provides a benchmark and scalable data construction pipeline for studying whether MLLM outputs are supported by appropriate visual evidence. Code is available at this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2611.13150
PDF
-
Teaching the Harness: Student-Aware Skill Transfer for LLM Agents
Abstract
Effective teaching requires reasoning about the learner as well as the task, a capacity that Theory of Mind describes as inferring another agent's latent state from its behavior. We ask whether a stronger LLM teaches a fixed student better when it applies this principle. The student is an LLM agent acting through a harness that fixes its tools, context, and action budget, so a procedure that helps one model can exhaust the budget of another. We introduce Student-Aware Teaching (SAT). The teacher builds a learner model that pairs an empirical profile with hypotheses about the student's limitations, tests these hypotheses with diagnostic probes, and writes reusable skills for the diagnosed failures. A skill enters the library only after it raises success on matched development tasks. With Claude-Sonnet-5 teaching DeepSeek-V4-Flash, SAT reaches 86.46% task success across four primary runs, leading both a profile-free teacher with the same framework (81.04%) and the untaught student (75.00%) in each run. Removing the analytical learner model lowers both prediction and success, and under misleading learner profiles none of the 48 candidate skills passes admission. When DeepSeek-V4-Flash and Gemini-3.8-Flash exchange libraries, each performs better with its own. Saved libraries keep average gains of up to 11.1 points after the teacher leaves. The benefit depends on a student that can follow instructions and a deployment route that delivers them.
Abstract (translated)
URL
https://arxiv.org/abs/2611.13144
PDF
