Chat
-
Learning from Research: Toward Lifelong Agent Harness Evolution
2026-09-30 17:00:38Jingbo Yang, Kwei-Herng Lai, Xiaowen Wang, Yaar Harari, Evgeniy Gabrilovich, Shiyu ChangAbstract
Language agents are expected to solve increasingly complex tasks, creating a growing need for continual improvement. One promising approach is to evolve the agent harness, the software that governs tool use, memory management, and task execution, while keeping the underlying language model fixed. Recent methods automate this process by using a meta coding agent to modify the harness based on execution feedback. However, relying on that agent's existing knowledge and observed failures can restrict exploration and make adaptation reactive. Inspired by how human experts learn from the research literature for new solutions, we introduce ScholarEvolve, a framework that automatically draws on state-of-the-art research to guide harness evolution. ScholarEvolve organizes the harness evolution directions into functional modules and uses topic modeling to identify distinct improvement strategies for each module. It implements these strategies and evaluates their combinations to improve task performance. Moreover, the framework is designed to incorporate new publications over time, allowing research advances to drive proactive lifelong evolution. Experiments demonstrate improvements on AppWorld and Tau2-Bench. ScholarEvolve raises Qwen3.5-27B task goal completion from 49.6% to 63.6% on AppWorld Challenge, and raises GPT-5.4-mini pass@1 from 72.7% to 81.9% on Tau2-Bench Telecom.
Abstract (translated)
URL
https://arxiv.org/abs/2609.40169
PDF
-
Persistent Context Graphs for Efficient Memory Compaction in LLM Agents
2026-09-30 16:43:28Jingbo Yang, Kwei-Herng Lai, Xiaowen Wang, Zhaoxuan Tan, Pei Zhou, Mengting Wan, Yaar Harari, Evgeniy Gabrilovich, Shiyu ChangAbstract
As LLM capabilities advance, agents are tackling increasingly complex tasks over longer horizons. Their growing interaction histories make memory compaction essential for staying within context windows and reducing prefill cost. Existing methods summarize the history or compress its KV cache, often adding model computation to preserve information for future requests. A new user request can change which history matters, but reassessing that history with the model requires re-encoding it if the KV cache has expired. Past attention provides signals of historical importance and dependencies between messages, while relevance to the current task must be assessed using the new user request. We introduce ReCAP, a memory compaction method that stores attention-derived importance scores and dependency links in a lightweight, persistent context graph. For each new request, ReCAP combines stored importance with relevance cues from the request and follows dependency links to select messages and their supporting context, without additional model calls for selection. Compared with Codex's default summarization-based compaction, ReCAP reduces estimated latency for compaction and cold restoration by approximately 95% on both Qwen3-Coder and gpt-oss. It also roughly halves the historical context per call on SWE-Together at comparable task quality and improves accuracy on the code tasks of Lost-in-Conversation over full history by 19.8 and 41.2 points.
Abstract (translated)
URL
https://arxiv.org/abs/2609.40118
PDF
-
Mid-Harness: Scaling Actions Between Model and Harness for Terminal Agents
2026-09-30 15:40:39Minki Kang, Ryo Hachiuma, Shaokun Zhang, Subhashree Radhakrishnan, Yonggan Fu, Jindong Jiang, Mingjie Liu, Ehsan Hosseini-Asl, Yi Dong, Yu-Chiang Frank Wang, Byung-Kwan LeeAbstract
Terminal agents act through stochastic model generations, yet the ability to generate a useful action does not ensure its reliable execution. A poor command (e.g., wrong package install) can change the environment in ways that hinder subsequent progress, even when the model could generate a better alternative. We investigate whether allocating test-time compute at the model-harness boundary can improve action reliability and trajectory success, and what makes this allocation effective. To study these questions, we introduce Mid-Harness, which samples and verifies candidate actions before forwarding one for execution, while keeping the generator and harness unchanged. With a TMAX-9B generator, more action sampling yields little benefit under weak verification, whereas a capable verifier can exploit useful alternatives from the same generator. On TerminalBench-Lite, a GPT-5.6 Sol verifier raises Pass@1 from 50.00% for the base agent to 68.03% with 8 sampled actions. When the same TMAX-9B model serves as the verifier, pairwise verification performs best among the evaluated verification mechanisms. Distilling responses from the stronger verifier into TMAX-9B further improves Pass@1, while leaving the action generator unchanged. With TMAX-9B on TerminalBench-Lite, combining action and trajectory scaling reaches higher success at lower estimated token cost than generating more trajectories alone. Mid-Harness also improves performance across additional models, benchmarks, and harnesses. These findings identify action scaling as a promising target for test-time compute scaling in terminal agents.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39982
PDF
-
AdaGEPA: Adaptive Feedback Allocation for Reflective Prompt Optimization
2026-09-30 15:13:40Junyang Chen, Zecheng Wang, Jingbang ChenAbstract
Prompt optimization improves the performance of language-model systems on downstream tasks by refining their prompts. Classical methods evaluate prompts on task examples and use the resulting feedback to guide prompt revisions through reflection. However, when feedback selection does not account for the prompt's weaknesses, these revisions may improve performance on selected examples without yielding broader task improvements. To address this issue, we propose AdaGEPA, an adaptive feedback-allocation method that uses the prompt's performance and task structure to select examples for the next prompt revision. Our method replaces at most one example in each feedback minibatch to target an identified weakness while preserving the remaining feedback context. Across our main experiments on six downstream benchmarks, AdaGEPA achieves higher mean validation scores than non-adaptive feedback selection under matched rollout budgets. AdaGEPA also finds high-performing prompts earlier across several tasks. In the initial Schema-Guided Dialogue (SGD) study, its half-budget prompts outperform the non-adaptive baseline's full-budget prompts in joint goal accuracy on new dialogues from services seen and unseen during search. Overall, our findings highlight the potential of adaptive feedback allocation to improve both the effectiveness and rollout-budget efficiency of reflective prompt optimization.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39927
PDF
-
FIGS: Evaluating Multi-Turn Sycophancy Without Penalizing Empathy
2026-09-30 14:46:13Sidharth Pulipaka, Ruta Binkyte, Ivaxi Sheth, Sahar AbdelnabiAbstract
Large language models frequently fail to balance staying truthful with being supportive. They often exhibit sycophancy in responses to users, agreeing with false claims, offering unwarranted flattery, and giving advice skewed toward users' expressed views. In reality, sycophancy rarely happens in a single exchange; it may emerge organically as users repeatedly insist or subtly steer the dialogue over time. Current evaluations, however, rely on rigid, single-turn tests or fixed scripts that fail to capture these natural dynamics. Furthermore, these benchmarks often mistake showing basic empathy for yielding, penalizing models for acknowledging a user's feeling. This view may drive future models to over-correct into cold, dismissive rigidity. To address this gap, we introduce FIGS (Factual Integrity and Grounded Support), a dual-axis evaluation framework built around extended, realistic dialogue. We use an adaptive 10-turn conversational simulator that dynamically challenges the target model, reflecting how users repeat requests, push back, or steer a conversation toward a preferred answer. To accurately evaluate these trajectories, we apply a taxonomy that strictly separates Sycophancy (whether the model holds firm to the truth and keeps its praise proportional) from Calibrated Validation (showing empathetic understanding of the user's feelings without overdoing it). We release our complete testing environment, including 500 diverse multi-turn scenarios and an automated judge. Our evaluation of leading models reveals a consistent trade-off: over the course of a sustained interaction, current systems either slowly drift to sycophancy or over-correct into robotic detachment. This demonstrates that balancing honesty with appropriate support throughout a natural conversation remains a critical, unsolved challenge.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39863
PDF
-
ArchitectureIQ: On the Measure of Training Intuition
2026-09-30 13:28:28Zirui Ren, Shaoyang Guo, Chencheng Tang, Jinxin Wang, Chengyu Xiong, Shanbin Yu, Peihang Li, Yidi Wu, Bangzhe Huang, Qingyu Qu, Leqian Yang, Ziming LiuAbstract
Top researchers have good intuition, but do language models have as good intuition about model training as top AI researchers? To measure model intuition of LLMs and humans, we introduce the ArchitectureIQ benchmark. Each question presents a synthetic dataset and several training recipes, and the test-taker is asked to predict the recipe yielding the best test metric. Overall, we find that LLMs' model intuition is good but has four limitations: (1) The intuition is imperfect, or even sub-human in some cases. Frontier models achieve around 76% accuracy (random choice 33%) vs best human researcher (66.0%), yet remain far from perfect. For architecture-only questions, best human achieves 65% while GPT-6 Astra only has 38%. (2) The intuition is empirical, not structured, supported by the fact that more CoT compute does not lead to substantial improvement. Unlike math, we still lack a "Science of AI" language that enables structured reasoning on AI. (3) The intuition is not maximally condensed, and can be further compressed into a knoledge base. Our constructed knowledge base with only 20 items yields large gains for weak models: GPT-4o equipped with the accumulated knowledge almost matches the performance of Claude Opus 5. (4) The intuition is insensitive to dataset properties, but the best model should in general depend on data properties. This suggests that data is the real "dark matter" in AI -- LLMs (so do human researchers) understand too little about data, even less than model architectures.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39714
PDF
-
GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives
2026-09-30 12:20:54Qize Yu, Lianrui Fan, Boyu Chen, Jiaqi Liang, Xini Ding, Yue Chen, Zetian Song, Yuran Wang, Yi Zou, Kaixuan Wang, Tianxing Chen, Wenxuan Song, Bohan Zhou, Mingleyang Li, Siqiao Huang, Yuqi Ye, Caigao Jiang, Wei Wei, Ruihai Wu, Hang Zhang, Yixiao Ge, Shuchang Zhou, Shilong Liu, Xianming Liu, Ping Luo, Shiyu HuangAbstract
Precise grounding matters. It specifies which object is the target and where that object is, even in clutter and for tiny objects, and it has to be fast enough for closed-loop control. Yet vision-language-action (VLA) and world-action models (WAMs) take perception from general-purpose vision-language and video-generation backbones, which still fail in these settings. We introduce GroundingPI, a 4B grounding foundation model that generates points and boxes as quantized coordinates in a shared vocabulary. Training combines multimodal and spatial pretraining, supervised fine-tuning, and reinforcement learning with GRPO, using supervision from public datasets and dedicated data engines. Against 44 baselines across 34 grounding benchmarks spanning 11 perceptual capabilities, GroundingPI establishes a new state of the art, averaging 73.68%, above the larger GPT-6 Astra (71.54%). As a downstream visual backbone, GroundingPI improves performance on robotic manipulation and autonomous driving. On RoboTwin 2.0, it outperforms every mainstream backbone we evaluate in all four out-of-distribution settings, by up to 24.8% relative to the strongest backbone. On RoboCasa-GR1, GroundingPI trained with 50% of the demonstrations outperforms those baselines trained with 75%. On nuScenes, used as the visual backbone, GroundingPI attains an average open-loop L2 error of 0.296 m. We systematically analyze GroundingPI's pretraining in scale and data composition. Downstream autonomous driving and robotic manipulation improve as the pretraining is scaled. Analyzing the data recipe across these 11 perceptual capabilities shows dense grounding's substantial benefits for both, and OCR's potential as a catalyst for perceptual learning. These results support grounding as a perceptual foundation, and dedicated perceptual pretraining as a promising direction for foundation models of physical intelligence.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39601
PDF
-
GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed
2026-09-30 12:20:51Qize Yu, Lianrui Fan, Bowen Ping, Xini Ding, Zetian Song, Junbo Niu, Kaixuan Wang, Tianxing Chen, Yue Chen, Minghua He, Yuran Wang, Jie Huang, Haojun Zhang, Min Chen, Hao Li, Wenxuan Song, Ruihai Wu, Xianming Liu, Shilong Liu, Shuchang Zhou, Ping Luo, Shiyu HuangAbstract
Autoregressive (AR) grounding models serialize spatial predictions, introducing sequential latency and imposing a causal order on output tokens. We view grounding as visual evidence extraction: objects, locations, and spatial relations are jointly constrained by the image and query, yet their dependencies do not imply an intrinsic left-to-right generation order. This distinction makes bidirectional diffusion a natural fit, allowing spatial hypotheses to emerge in parallel and be jointly refined through iterative denoising. We introduce GroundAnything, a 4B-parameter grounding foundation model that reconciles fast parallel decoding with precise localization through blockwise denoising. Training combines grounding pretraining from public datasets and dedicated data engines, direct AR-to-diffusion conversion with joint AR and diffusion objectives, supervised fine-tuning, and GRPO-based reinforcement post-training. Across 30 grounding benchmarks, our autoregressive variant, GroundAnything-VLM, establishes a new overall state of the art among similarly sized models at 72.42%, remaining competitive with GPT-6 Astra (71.35%). With entropy-guided decoding, GroundAnything also surpasses the prior state of the art at this scale, averaging 61.75% versus 53.32% for the fast MTP-based LocateAnything model. We further explore decoding strategies, showing that an optional self-speculative mode achieves a $4.51\times$ speedup over the AR counterpart with a 0.74 percentage-point drop in COCO F1mIoU. Infrastructure experiments show that progressive inference optimizations translate parallel decoding into practical speedups. These support efficient visual grounding in latency-sensitive real-world systems.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39600
PDF
-
From Given to Gathered Evidence: Agentic Learning for Longitudinal Medical Reasoning
2026-09-30 12:01:45Minye Shao, Chaohui Yu, Yixuan Wu, Fan Wang, Ling Shao, Yang LongAbstract
Foundation models can serve as clinical agents through tool-use harnesses. However, conventional medical benchmarks assess reasoning over preselected evidence rather than the ability to seek it across clinical records and longitudinal imaging. We propose CASE: a series of role-specific Clinical Agents for Seeking Evidence, together with a tool-use harness and an agentic post-training framework for compact vision-language policy models. We further introduce a longitudinal multimodal benchmark built on UK Biobank, comprising 50,401 clinical questions derived from real-world ICD-10-coded diagnoses of 4,739 participants. Each question links to a patient-specific environment containing clinical context and multi-sequence MRI from baseline and follow-up visits, where agents autonomously select which visits, organs, modalities, slices, and specialist tools to inspect and compare. Supervised fine-tuning transfers evidence-seeking workflows from 14,734 frontier-model interaction trajectories, followed by agentic reinforcement learning on the learner's own environment interactions. Privileged on-policy self-distillation and rubric-based LLM feedback refine evidence-to-conclusion reasoning without prescribing tool sequences. Experiments show that CASE moves beyond question-answer imitation toward transferable investigation policies, strengthening evidence-grounded longitudinal reasoning. Under matched evaluation conditions, our Qwen3-VL-8B based agent achieves over 16% and 10% relative improvements in answer accuracy over GPT-5.4 and Claude Opus 4.8. Code will be available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39566
PDF
-
RankEvolve: A Reliable Multi-Agent Auto-Research Harness for Evolving Ranking Models
2026-09-30 11:50:46Zheng Chen, Linfeng Liu, Hong Li, Hong YanAbstract
Auto-research agents, LLM systems that propose, implement, train, and evaluate model changes across iterations, promise to automate applied ML's experimental loop. Over long horizons, execution accuracy is a binding constraint: a change can silently leak held-out data, omit normalization, disconnect a gradient, or leave a train/eval flag unwired, invalidating expensive runs and compounding error across iterations. We present RankEvolve, an auto-research framework for evolving generative ranking models. An Executable Operating Protocol (EOP) declares phases, gates, branches, and loops, and the runtime enforces the compiled state machine. A meta-meta-harness composes complete black-box coding-agent products, including Claude Code and Codex, as execution-graph nodes that review and repair one another's work. In a budget-matched evaluation, heterogeneous composition raises all-oracle execution accuracy from the best single-product baseline of 45.8 percent to 62.5 percent (paired +16.7 points, 95 percent CI [6.6, 26.7]) while achieving a 10.4 percent silent critical-defect rate. An implemented knowledge layer carries findings, including negative results, across iterations. In a twelve-iteration deployment on the open-source HSTU recommender, RankEvolve reported NDCG@10 of 0.2192 on MovieLens-20M LARGE (+4.48 percent over the published anchor) and 0.1948 on BASE (+2.80 percent). ExecML-HSTU, seeded by incidents from that deployment, provides the oracle benchmark for the execution-accuracy evaluation. A pre-specified LitGPT transfer split replicates the heterogeneous-composition effect beyond recommendation (+12.5 points, 95 percent CI [3.0, 22.0]), and a paired ablation isolates per-step from full-protocol instruction injection. These results characterize when runtime-controlled composition of coding-agent products improves execution accuracy.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39551
PDF
-
Comparative study of adapting pre-trained models for driving behavior video captioning
2026-09-30 11:41:16Sayak Mallick, Philipp Geiger, Augustin KelavaAbstract
This report examines and compares some of the many fine tuning and prompting methods existing, applying them within the domain of autonomous driving. The idea is to compare these methods by adapting a Large Language Model (LLM) on a video dataset. LLM's have become extremely good at achieving a good understanding of different forms of data and this study aims to induce a low dimensional understanding of driving situations into our primary test model SpaceTimeGPT. Experiments on BDD-X (Berkeley DeepDrive eXplanation) dataset demonstrate good performance of the full fine tuning framework on some automatic metrics, and in some metrics, it even surpasses the baseline. We also try Low-Rank Adaptation (LoRA) and prompt engineering on VideoLLaVA model and discuss its limitations.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39542
PDF
-
Referential Uncertainty in Human--AI Collaboration
2026-09-30 11:17:46Christian Poelitz, Finale Doshi-Velez, Si\^an LindleyAbstract
Effective human-AI collaboration requires partners to establish references through interaction, which becomes fragile when descriptions are ambiguous, similar referents compete, or partners see different things. We study referential uncertainty - uncertainty over which candidate object a description refers to - in a collaborative puzzle task where a human Helper instructs an AI Worker to place pieces. The Worker must identify and communicate its uncertainty, and the Helper must recognize and act on it. We show that a separately elicited belief distribution over candidate pieces is better calibrated (ECE 0.15) and better discriminates correct from incorrect placements (AUROC 0.65) than raw action-token probabilities, which are severely overconfident (0.97 mean confidence, ECE 0.44). Across three frontier vision-language models (GPT-4.1, GPT-5, GPT-5.5), this elicited uncertainty rises predictably with instruction vagueness, but not with competing referents in context, even when those increase errors. The models seldom externalize it, asking for clarification on only 3.5-16.7% of turns. In a controlled human study (N=210), participants given only the Worker's default message accept 78% of wrong placements and cannot tell right from wrong (AUC 0.50). Precise descriptions and, especially, well-targeted hedges cut wrong-move acceptance to 36% while largely preserving correct-move acceptance, compensating for missing shared awareness such as not seeing the Worker's action. But this benefit depends on targeting: a deployable hedge derived from the model's own belief entropy inherits that signal's weakness and can do more harm than good. Externalized uncertainty helps a human partner only when it is accurately targeted.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39518
PDF
-
LIBERO-Agent: Evaluating General-Purpose Agents for Direct Embodied Manipulation
2026-09-30 11:10:10Zijie Diao, Yitong Chen, Sicheng Xie, Tianyi Lu, Wujian Peng, Guojin Zhong, Houze Xu, Ziyi Ye, Zuxuan Wu, Yu-Gang JiangAbstract
General-purpose agents can plan, use tools, and revise their behavior from feedback, but it remains unclear whether these capabilities transfer from digital environments to embodied manipulation. To investigate this question, we introduce LIBERO-Agent, an agent-native benchmark for evaluating these agents in robot manipulation tasks. Rather than asking agents to submit task-level Python control programs or operate through high-level robot skills, LIBERO-Agent provides an interactive robotic environment where agents can select which observations to inspect, process them with their own tools, and issue native action commands. LIBERO-Agent integrates 200 tasks into a common interaction framework and provides a 30-task primary suite that separates perception, short-horizon execution, and long-horizon composition. Results reveal a pronounced reliability gap: while agents perform well on perception and easy short-horizon tasks, their performance degrades substantially on hard short-horizon and long-horizon tasks. Richer observations improve short-horizon manipulation, while demonstration benefits depend on the agent and format. Among these agents, GPT-6 Astra achieves the strongest overall performance. Further analysis shows its major advantage lies in mechanism interaction, especially when sustained physical contact is needed, while its remaining failures stem from cross-stage interference and geometric errors.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39507
PDF
-
NarrativeSteward: Coordinating Delegation, Guidance, and Verification in Agent-Assisted Interactive Narrative Authoring
2026-09-30 09:04:28Wenjin Wang, Jiazhen Lei, Yuxin Sha, Nuwa Xi, Meng Zhao, Xingxi Yin, Qi Liu, Yuliang Shen, Zixun SunAbstract
Autonomous AI agents can turn authors' goals into interactive narratives by independently organizing and carrying out generation and revision. As agents generate and revise extensive content, authors struggle to grasp its overall structure, local details, and relationships, complicating continued guidance. We present NarrativeSteward, an authoring environment that organizes outlines, worldbuilding, and narrative graphs as linked artifacts for agent implementation and author guidance. Agent dialogue and project-wide structural review help authors understand the evolving work and guide local and cross-layer revisions, while change records and execution verification help authors assess the resulting work. Technical tests validated the system's change records, recovery mechanisms, and execution diagnostics. In a 12-participant within-subject study, NarrativeSteward supported easier formulation of revision requests and inspection of changes, and greater perceived understanding of changes and story structure, than general-purpose agents. Qualitative findings show how reviewing the work and feedback helps authors develop requirements and guide subsequent delegation. We open-source NarrativeSteward at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39333
PDF
-
4MT-VLM: How Coarse Is a VLMs Cognitive Map?
Abstract
An agent that moves must recognise a place from a viewpoint it has never seen. We introduce 4MT-VLM, a dataset of procedurally generated landscapes, each rendered across five stimulus modes that remove appearance cues while holding layout fixed: shape and colour, shape only, colour only, bare terrain peaks with no objects, and a valley viewpoint that puts the peaks on the horizon. The last condition is commonly used in clinics to probe hippocampal function in human patients. We test this benchmark across sixteen different open and closed-source models and report 4AFC performance, a measure which is also used to grade human participants. We observe that models identify a place from the studied viewpoint but lose it once the camera moves, dropping below the 25% chance level at 135° where a human observer scores 85%. Frontier models (Gemini 3.8 Flash, GPT-5.6) answer only 39% and 31% of rotated trials correctly, recovering to 85% and 55% only when distractors are moved more than 30 meters apart. Our benchmark demonstrates that while current VLMs possess rudimentary cognitive maps, their spatial resolution remains fundamentally too coarse to maintain a stable, 3D understanding of the world once the viewpoint changes.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39238
PDF
-
RAIM: Robust Aggregation of Inexpensive Models for Hallucination Detection
2026-09-30 08:03:31Elia Onofri, Roberto Di PietroAbstract
Automatic evaluation of faithfulness increasingly relies on a large language model acting as a judge, yet the most reliable judges are proprietary frontier models, costly and ill-suited to high-throughput monitoring. We investigate whether a panel of cheap open-weight judges (4--9B) can be aggregated to stand in for a frontier one, what the substitution sacrifices, and when it is worth making. We propose RAIM, an aggregation scheme robust to the members' correlated errors, coupling a cross-fitted stacked logistic regression with an admissibility test that, read from the members' own outputs, identifies when aggregating them improves on their best member and stays within reach of the frontier judge. We instantiate RAIM with ten judges from disjoint families across eight faithfulness benchmarks. Against Claude Sonnet, the panel retains a median 93% of its Cohen's $\kappa$ and gives up only 2.9 points of balanced accuracy on average; read as paired differences, it clearly improves on one benchmark and clearly worsens on three (only two by a non-negligible margin), leaving four unresolved. At a sixty-fourth of the frontier's inference price, the operative expense is a one-time in-domain calibration on 50--100 labelled records. The panel is also competitive with purpose-trained detectors on their home benchmarks (within 1.3 accuracy points of GPT-4o and 1.9 of the LLM-AggreFact leader), and beats the strongest one we reran by 6 points on our grounded sets. Whether aggregation pays depends on the members themselves: where several capable members err on different items, the panel improves on its best judge and approaches the frontier; where one dominates, the stacker recovers the leader, and only there does the frontier remain materially ahead. Both conditions are read off the calibration set at no further cost, so a cheap panel can stand in for a frontier one wherever this audit admits it.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39229
PDF
-
Argument Structure Prediction in Online Conversations: A Comparative Study of Modeling Paradigms and Task Architectures
2026-09-30 08:01:11Siddharth Bhargava, Sara Tonelli, Patricia Mart\'in-Rodilla, Javier ParaparAbstract
Argument structure prediction (ASP) constructs complete argument structures from discourse by identifying argumentative units and their relations. While recent work has explored diverse approaches---including unified neural models, multi-step pipelines, and prompt-based large language models (LLMs)---their relative trade-offs remain under-explored, particularly in dialogical settings. We present a systematic evaluation of ASP under strict schema constraints, comparing supervised fine-tuning and prompt-based LLMs across single- and multi-step task architectures, generating complete argument structures from dialogical input end-to-end. We benchmark them on three diverse dialogical corpora adapted from Inference Anchoring Theory into bipolar argument structures. Under a shared evaluation framework, we assess predictive performance, cross-domain generalization, schema compliance, and computational efficiency. Our results show that ASP remains a challenging task, with identifying argumentative relations emerging as the primary bottleneck, largely due to the implicit and context-dependent nature of dialogical argumentation. To facilitate future research, we release our data processing pipeline and end-to-end modeling framework for computational ASP on dialogical corpora.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39225
PDF
-
Uruqi: Learning Spatial Cognition from Visual Experience
2026-09-30 07:47:02Shichao Li, Meiqi Wang, Fei Su, Zhicheng ZhaoAbstract
Spatial intelligence requires maintaining a coherent understanding of the world as the embodied agent moves. Like humans, the agent must use its own motion to interpret changes across observations and update object locations and spatial relations accordingly. Despite spatial post-training having substantially broadened the spatial intelligence of vision-language models (VLMs), they still struggle with two atomic spatial capabilities: tracking self-motion and mapping the surrounding world during motion. To address this gap, we provide dense multi-turn supervision over interleaved atomic capabilities within each training episode, mimicking the visual experience of a continuously moving agent that reasons as it observes. To scale this up, we synthesize 11,738 motif-driven camera trajectories over a broad range of 3D scenes, supporting self-motion tracking, persistent object mapping, and rich spatial operations within each visual experience. By training models to reason over these atomic questions, our URUQI$_{\mathrm{Syn}}$-8B improves accuracy from 15.84% to 47.73% on our Uruqi benchmark comprising 52k questions across 2.7k episodes. URUQI-SI-Mix-8B further reaches 50.41%, comparable to the 50.08% achieved by GPT-6 Astra. Trained solely on our synthesized data, URUQI$_{\mathrm{Syn}}$-8B achieves an average relative accuracy improvement of 17.13% over its InternVL3-8B backbone across three external spatial benchmarks. These results highlight continuous visual experience as a scalable source of supervision for developing spatial cognition in VLMs.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39195
PDF
-
DAGent: Evaluate-then-Grow Planning for Deep Research Agents
2026-09-30 07:19:01Hanwen Liu, Yuanfu Sun, Qiaoyu TanAbstract
Deep research tasks require agents to navigate large knowledge spaces, synthesize evidence across many sources, and adapt their plans as findings emerge. Directed acyclic graph (DAG)-based multi-agent systems suit this setting because they support parallel execution and isolate each sub-task within a focused dependency context. Yet existing DAG-based agents instantiate a task-level plan before execution and repair the graph only after failures or missing evidence are observed. This Plan-then-Patch strategy is brittle for deep research: the system commits most strongly when its evidence is weakest, and later revisions waste computation on branches that should not have been planned. We propose DAGent, a DAG-based multi-agent framework with Evaluate-then-Grow incremental planning: an Orchestrator grows the task graph one batch at a time, conditioning each expansion on confidence and uncertainty signals from completed nodes. A hierarchical context layer propagates compact QueryDocs by default while preserving full execution traces for on-demand recall. The recorded DAG topology admits structural RL signals that outcome-only recipes cannot define; DAGRPO, a GRPO adaptation, injects topology-conditioned credit on Executor rollouts and a structural compliance regularization on Orchestrator plans. Across BrowseComp-Plus, GAIA, and xbench-DeepSearch, DAGent surpasses the strongest open-source baseline by 5.3 / 5.8 / 2.0 points at the Qwen3-235B-A22B scale, and the lead replicates across four open-source backbones and extends to GPT-5 at 327K context. At the Qwen3-8B scale, DAGRPO improves over a same-budget outcome-only GRPO baseline by 3.0 average Pass@1 points. A same-architecture comparison shows that evidence-conditioned planning reaches higher accuracy at lower per-task token, tool-call, and step footprints than its Plan-then-Patch counterpart. Code: this https URL
Abstract (translated)
URL
https://arxiv.org/abs/2609.39154
PDF
-
BELIEFRAG: Making Adaptive RAG State-Aware under Evolving Evidence
Abstract
Adaptive RAG uses signals such as confidence, relevance, support, and retrieval quality to decide when to search or correct evidence. In multi-step retrieval, however, these local signals must be combined into a persistent view of what the current evidence supports, what remains missing, and which action should follow. Existing methods often use such signals as separate triggers, making it difficult to preserve a coherent evidence state across a trajectory; we call this problem evidence-state fragmentation. We introduce BELIEFRAG, a closed-loop controller that updates an explicit state over sufficiency, reliability, conflict, uncertainty, evidence gaps, and acquisition cost, then chooses among retrieval, query rewriting, verification, answering, stopping, and abstention. Across six QA benchmarks with gpt-oss-120b, BELIEFRAG reaches mean token F1 0.572 with 3.89k tokens per question, outperforming fixed iterative retrieval (0.555 F1) while using 39% fewer tokens. The same quality-cost pattern transfers to Qwen3-32B, where BELIEFRAG reaches 0.552 F1 versus 0.523 for iterative retrieval while using 35% fewer tokens. Analysis shows that the main gains come from corrective re-retrieval rather than pruning alone, while several belief dimensions are redundant and calibrated answerability plays the strongest operational role. Calibration improves threshold stability across related evidence sources, although source shift can still invalidate the same decision signal.
Abstract (translated)
URL
https://arxiv.org/abs/2609.39139
PDF
