Hang Zhang, Qihao Ye, Litong Liu, Yunbei Zhang, Yugyung Lee, Yue Zhou, Bowen Liu, Emily First 等 10 位作者
Neural theorem proving increasingly uses retrieved premises and project context to support proof generation. Evaluating these resources requires measuring the advantage of target-selected declarations over comparable alternatives and how this advantage depends on training. We introduce LeanContext, a benchmark of 129,826 Lean whole-proof generation problems, to investigate these questions. We group selected complete preceding mathematical declarations as Mathematical Context (MC) and the remaining selected local source commands as Environment Context (EC). Four aligned input views support crossed training and inference evaluations with shared targets and original Lean verification environments. For both diffusion- and autoregressive-based models, selected declarations outperform comparable alternatives by a larger margin when training includes MC. Final-input audits identify omitted declarations referenced by the target. Prioritizing these dependencies improves Pass@32 by 1.40% and 1.70% for LLaDA and Llama, respectively, on 2,000 targets. The selection experiment operates without access to target reference proofs and holds checkpoints, EC, and inference budgets constant. The code and dataset are available at https://github.com/haaaardcvZ/LeanContext.
Shuoyang Jasper Zheng, Keigo Yoshida, Nico García-Peguinho, Jiatong Liu, Dan Hearn, Anna Xambó Sedó, Nick Bryan-Kinns
Neural audio autoencoders, a deep learning method for sound synthesis, are increasingly popular in AI-enhanced NIMEs. It is timely to explore as a community how this technological opportunity has opened a domain of design in NIME. This paper focuses on one compelling technique of using autoencoders for sound synthesis: navigating their latent space as a generative sound space. We introduce Latent Terrain, a Max/MSP tool package designed to tailor latent spaces into corpus-based sound spaces for NIMEs. We describe the rationale and development process of Latent Terrain to offer insights into the use of autoencoders in a material-oriented crafting space of musical interface design. We deliver an annotated portfolio resulting from a collaborative artistic exploration of Latent Terrain with four NIME makers, to showcase the design possibilities opened by autoencoders. We reflect on our practice-based account to discuss the challenges and opportunities of enabling neural audio autoencoders as design materials for AI-enhanced NIMEs.
Yuetong Liu, Zhehao Zhou, Yitong Zhang, Haorui Chen, Jia Li
Large language model (LLM) agents typically append raw tool outputs to their context. Over multi-step interactions, long and noisy observations incur substantial context overhead and dilute the agent's attention to task-relevant information. Existing token pruning and summarization methods remain constrained by discrete text representations, limiting the achievable compression ratios. In this work, we challenge the prevailing assumption that agents must consume tool observations as explicit text tokens. To this end, we propose Coca-Cola, a continuous latent context compression model for LLM agents. Coca-Cola compresses tool observations into latent tokens and embeds them into the input embedding space of the agent. We further introduce Action-Preserving Training, which aligns the agent's next-action distributions under full and compressed observations. This trains the compression model to learn action-sufficient latent tokens that retain decision-critical information. Experiments with web-search agents on four QA benchmarks and two model families show that Coca-Cola reduces the agent's token consumption by 70.01%-89.24% and end-to-end runtime by 24.45%-55.97%, while consistently improving task performance. These results suggest that Coca-Cola offers a promising approach for context compression in agents. Our code is available at https://github.com/THU-Agent/Coca-cola.
Scaling computer-use agents from individual actions to complex software work requires experience that connects decisions to complete outcomes. We present NorthCUA, an integrated system for turbocharging computer-use agents through experience at scale. Its Experience Flywheel supplies diverse executable tasks to the NorthCUA Harness, which sustains long interactions, records decisions, and checks task outcomes. The Flywheel uses these records to curate verified trajectories. For unsolved tasks, recursive self-improvement evolves reusable skills and preserves verified milestones so later attempts build on earlier progress. Staged Learning combines this experience with grounding data from the Flywheel to develop GUI competence through mid-training and long-horizon supervised fine-tuning. The policy then learns from fresh attempts in the same Harness through joint desktop-web reinforcement learning, closing the loop between experience production and policy improvement. The resulting Nano, Mini, and Pro family achieves state-of-the-art results among open-source models across GUI grounding, desktop execution, and Web interaction under the reported evaluation settings. For more responsive Web workflows, NorthCUA-Ultrafast pairs direct action selection with open-ended generation on pre-RL Nano, offering a measured quality-latency tradeoff. Together, these advances establish NorthCUA as a high-performing model family and a reusable foundation for agents that carry complex software work from intent to completion.
This paper revisits TEE-assisted BFT under a universal partial-TEE model, where an arbitrary subset of replicas execute inside TEEs while the remaining replicas operate without hardware trust guarantees. We show that heterogeneous trust changes the structure of quorum formation and fault tolerance. In particular, we derive a tight resilience bound f < max {n/3, m/2}, where n is the total number of replicas and m is the number of TEE-enabled replicas. The result reveals a sharp threshold phenomenon: TEEs improve fault tolerance only once they exceed two-thirds of the deployment. Guided by this characterization, we introduce two protocol principles: (1) a dual-quorum construction that safely combines TEE-only and mixed quorums, and (2) a TEE-leader fast path that leverages hardware-enforced non-equivocation to reduce both consensus and view-change latency. We realize these ideas in Raftel, which is, to our knowledge, the first HotStuff-style BFT protocol designed explicitly for arbitrary partial-TEE deployments, and in chained-Raftel, a pipelined variant that further accelerates mixed-trust execution. We implement both protocols atop Intel SGX and evaluate them in LAN and WAN environments. Our results show that Raftel achieves up to 625 TPS with sub-670 ms latency in WAN settings, outperforming HotStuff by up to 308 TPS in throughput while approaching the performance of fully TEE-assisted protocols.
Guard models are increasingly used to safeguard LLM-based agents, primarily by identifying actions that agents are forbidden to perform. However, identifying forbidden actions alone is insufficient to ensure agent safety. In this paper, we argue that agent safety also depends on identifying required yet unperformed safety-critical actions, which we call obligations. Our preliminary study on a popular benchmark for evaluating safety shows that 56.92% of GLM-5.3 trajectories contain unfulfilled obligations, compared with only 30.00% containing forbidden actions. This finding reveals unfulfilled obligations as a major and previously overlooked source of safety risk. However, to our knowledge, no existing benchmark evaluates whether guard models can identify these obligations. To close this gap, we introduce ObligationBench, the first benchmark for evaluating the capability of obligation identification, comprising 240 expert-validated trajectories covering issue resolution, feature development, and terminal operations. Our evaluation of 14 representative models reveals substantial limitations: the highest recall and exact-match rate are only 48.97% and 10.00%, respectively. To address these limitations, we develop ObligationGuard using 40,000 synthetic training examples. ObligationGuard achieves 57.52% recall and an exact-match rate of 21.67%, surpassing all evaluated models on both metrics. We call on the community to incorporate obligation identification into the design and evaluation of future guard models to improve agent safety.
We present Reap, an open framework for learning and analyzing formal proof search in Lean4. Reap implements policy-value-guided Monte Carlo Tree Search directly in Lean, sharing proof-state transitions, AND-OR search, and replay verification between inference and training rollouts. Structured action logs expose how neural predictions and tactic execution interact during search. Using this framework, we train Reaper-1, a 1.7B-parameter policy-value model, through continued pre-training, supervised tactic learning, and iterative search-based reinforcement learning. Using a separate premise selector, the final system solves 79.9% of miniF2F-test at pass@32, establishing state-of-the-art performance among open provers of comparable size while remaining competitive with substantially larger models. Beyond improved solve rates, our results suggest that RL strengthens the prover's ability to organize proofs around useful intermediate results, favoring substantial advances over repeated local adjustments. This more directed proof construction is accompanied by increasingly elaborate tactic programs: fewer explicit search decisions are needed, while more work is performed within each decision. The resulting trade-off between search effort and tactic execution cost provides insight into how learning changes the prover's problem-solving strategy.
Realizing Embodied AI for Science requires robots to translate experimental goals into reliable physical procedures, reducing hazardous and repetitive human work while generating evidence for scientific discovery. Although language-based planning and learned manipulation provide complementary capabilities, their integration requires robots to adapt subsequent operations to actual execution outcomes. For example, an unsuccessful placement changes what must be done next, even when the experimental goal remains unchanged. Existing benchmarks assess planning, manipulation, and laboratory workflows through task completion and reasoning or execution metrics; however, aggregate performance leaves the connection between outcome-dependent choices and physical execution unresolved. To address this evaluation gap, we introduce Labference, a benchmark for evaluating how observed physical outcomes guide subsequent choices and how those choices translate into robotic laboratory actions. Its evaluation protocol is designed to test outcome-dependent behavioral adaptation through controlled evidence variations, while independently measuring operation selection, physical execution, and task completion. Matched Oracle-assisted continuations from shared checkpoints then assess which forms of support improve subsequent choices and task completion. Initial weighing evaluations reveal target-selection errors in the tested VLA policies despite successful object transfers, whereas GPT-driven control achieves correct selection but encounters execution failures and substantial computation costs. Together, these findings highlight the need to connect task-appropriate reasoning with reliable, timely action. By making this connection measurable and diagnosable, Labference supports targeted improvements toward autonomous experimentation for Embodied AI for Science.
Coding agents can now rewrite repository-scale codebases while preserving tested behavior, creating a new challenge for plagiarism detection, license auditing, and software provenance. We ask whether such rewrites can still be traced to their source. Existing clone benchmarks emphasize fragments and bounded edits, while repository-transformation benchmarks evaluate functional correctness rather than provenance. Moreover, ranking the most similar source does not establish derivation when independent implementations can be similarly close. We define repository-level agentic code laundering and introduce Launder-Bench, to our knowledge the first repository-level benchmark for provenance under agentic rewriting, constructed from 282 real-world repositories. A source-hidden reconstruction protocol decomposes source behavior, reconstructs modules and repository structure, and retains rewrites only after execution-based verification. We evaluate the same provenance problem in two complementary settings: Launder-Bench-Pair contains 1,128 pairwise instances asking whether two repositories share provenance, while Launder-Bench-Attr contains 229 source-attribution instances asking which candidate is the source, with abstention when evidence is insufficient. Measured same-domain independent repositories establish how much similarity can arise without shared provenance and provide detector-specific reference bands. Existing methods remain far from saturating Launder-Bench: the strongest static method reaches only 0.66 AUROC on pairwise detection, while repository encoders rank the true source first on up to 22% of attribution queries but earn calibrated attribution credit on only 1-2%. These results reveal a systematic gap between relative code similarity and source-specific provenance evidence, providing a reproducible target for provenance detection under agentic rewriting.
Efficient object detection on resource-constrained edge devices is essential for safety-critical applications such as fire-smoke monitoring. However, current lightweight frameworks, such as YOLO26, still rely on the parameter-dense multi-branch Feature Pyramid Network (FPN), preventing full exploitation of their lightweight advantage. Replacing FPN with YOLOF's single-input-single-output (SiSo) neck is a promising solution but causes substantial accuracy loss. To alleviate this, we propose YOLOF-MLD, a CPU-efficient multi-level distillation (MLD) framework that combines a YOLO26n backbone, a SiSo neck, and MLD. It aligns features at the backbone via sliced-semantics matching, neck with receptive-field matching, and response levels. On D-Fire and Annotated Fire-Smoke (AFS) datasets, YOLOF-MLD reduces the parameter count of YOLO26n by 35.7\% and achieves a 62.3\% speedup on Raspberry Pi 5, attaining 71.6% and 59.9% mAP@50, respectively, compared with 75.5% and 60.0% for YOLO26n. These results demonstrate a competitive accuracy--latency trade-off for edge deployment. Our code is available at https://github.com/XiaolongWU111/YOLOF---MLD/blob/main/README.md.
Benders decomposition and column generation solve large optimization problems by passing cuts, columns, and dual values between a master problem and subproblems, and are hard to implement correctly. Large language models now automate parts of algorithm design, which points toward agents that build decomposition solvers for new problems. As a first step toward such agents, we study how well current coding agents design and implement these two algorithms, and how they fail. We introduce DecompGym, 81 instances from nine problem families (five Benders, four column generation), in which an agent writes a solver for a prescribed method. Each solver must meet three criteria: execution within a 60-second time limit; numerical feasibility together with a method audit, in which an LLM judge reads the solver's code; and an objective within 1% of the stored optimal objective. We evaluate 12 GPT, Claude, and Gemini models running in coding-agent tools. Over up to ten attempts, we compare independent regeneration with reflection, which first diagnoses the failure. On the first attempt, two models solve every instance and seven solve more than 90%. Most Benders failures (48 of 62) are missing or suboptimal solutions. The most common column-generation failure is a method-audit rejection. Output checks alone would accept 93 of the 108 attempts that fail the method audit. The most common rejection, an optimalif, occurs 96 times, 92 of them in one model.Reflection turns 66% of retries into successes, compared with 39% for regeneration, and it repairs method-audit failures, for which the judge names the defecective gaps. Agents for decomposition therefore need checks on how a solver reaches its answer and feedback that locates the error. The benchmark is available at https://github.com/SIKAI-C/DecompGym.
Siyuan Li Peng Cheng Laboratory and Harbin Institute of Technology, Shenzhen, Yunjia Wu, Yiyong Xiao Fudan University, Pingyang Huang, Peize Li, Ruitong Liu, Yan Wen, Te Sun
Temporal knowledge graph (TKG) forecasting requires predicting future facts by jointly modeling structural dependencies within each snapshot and temporal evolution across snapshots. However, most existing methods are stateless: they recompute entity representations at each timestamp from a limited query window, leading to episodic amnesia and rapid decay of long-term dependencies. To address this limitation, we propose Entity State Tuning (EST), an encoder-agnostic framework that endows TKG forecasters with persistent and continuously evolving entity states. EST maintains a global state buffer and progressively aligns structural evidence with sequential signals via a closed-loop design. Specifically, a topology-aware state perceiver first injects entity-state priors into structural encoding. Then, a unified temporal context module aggregates the state-enhanced events with a pluggable sequence backbone. Subsequently, a dual-track evolution mechanism writes the updated context back to the global entity state memory, balancing plasticity against stability. Experiments on multiple benchmarks show that EST consistently improves diverse backbones and achieves state-of-the-art performance, highlighting the importance of state persistence for long-horizon TKG forecasting.
尚无 AI 解读，展示原文摘要
cs.CL
Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) · 2026 · 正式发表