让下一篇重要论文，不再错过。

For LLMs, skills are textual recipes encoding mature problem-solving experience. Despite widespread deployment, their utility is limited by the model's ability to comprehend and follow skill instructions. To address this limitation, we propose ParametricSkills, a framework that parameterizes textual skills at test time, enabling context-free management of the full skill lifecycle. Specifically, we first construct a large-scale and high-quality skill collection, and then synthesize single-turn and multi-turn skill-exploitation trajectories. Using this data, we then train a hypernetwork that parameterizes both the content and the exploitation methodologies of skills by converting textual skills into LoRA adapters. To facilitate the real-world application of our framework, we further align the hypernetwork on complex, long-horizon trajectories collected with OpenCode, enabling the models to operate reliably in realistic interactive workflows with ParametricSkills. Across evaluations, ParametricSkills effectively parameterizes skills, yielding substantial gains in task completion. More notably, as parametric skills accumulate as adapters rather than crowding the context, they open a preliminary but promising path toward continual learning at test time.

LLM-based generative recommendation reads long, text-rich interaction histories but generates only short semantic identifiers (SIDs). This asymmetry makes supervised fine-tuning inefficient: encoding the history is expensive, yet each history usually provides supervision for only one target. We present Merge and Branch (M&B) to reduce input cost and learn more from each encoded history. Merge compresses each historical item's SID and text into a small number of learned tokens. Branch uses the compressed history to predict multiple target SIDs. To share history computation while keeping these targets independent, M&B organizes each example as a shallow, wide tree. We develop tree FlashAttention to execute this layout efficiently, processing the shared history once and propagating gradients from every branch for end-to-end training. A tree-structured key/value cache extends history sharing to constrained decoding. Experiments on public benchmarks show that M&B outperforms strong recommendation baselines while substantially improving training and decoding efficiency. Online A/B tests on a large-scale e-commerce platform suggest that M&B effectively combines item text with multi-target supervision, yielding statistically significant gains in per-user clicks, orders, and GMV.

Deterministic benchmark scores show that an agent received credit, but not whether that credit was earned, reported honestly, or would hold on a second run. We introduce a Trust Layer for Agent Evaluation, an additive post-hoc framework that reports, beside each recorded score, whether it should be believed. It verifies four properties: whether the result is supported by the benchmark's own grading logic, whether a passing answer was earned through traceable computation, whether the agent's completion claim matches what occurred, and whether the result is stable under repeated execution. The first three use only saved artifacts; the fourth re-runs the agent. Model judgments only label evidence under majority voting; all verdicts follow deterministic rules and never modify the recorded score. Applied to five agent configurations on 108 tasks from Agents' Last Exam, every model shows passing runs with no traceable computation (at rates varying tenfold), confirmed false completion claims, and unstable results: 18-46% of tasks do not stay in one score band over five runs. Only 22.6% of recorded passes clear all four checks (95% CI 15.0-32.6, n=84). Measuring what an agent can do and verifying that it did it are different problems, and current benchmarks address only the first.

Turning a moving-camera perspective video into a 360° video means generating content outside the input frame under several constraints: observations revealed over time must be preserved, unseen regions plausibly completed, and the resulting scene kept consistent as the camera viewpoint changes. Existing methods assign these different requirements to a single conditional video model, requiring them to be jointly learned from panoramic video data. We instead present a modular, 3D-driven framework that distributes these three responsibilities across specialized components, all connected through a shared scene representation. Input observations are reconstructed explicitly in 3D, while a pretrained panoramic image model completes unobserved content from a selected panoramic render. This generated content is lifted into the shared 3D scene, where it remains spatially anchored across camera viewpoints. Without requiring any additional panoramic-video training, our framework achieves state-of-the-art input-view fidelity, perceptual quality, and geometric consistency on a static-scene subset of the Argus evaluation set. These results support our decomposed, 3D-driven formulation as a promising alternative to direct video generation in this setting.

Cryptocurrency transaction analysis often relies on black-box machine learning models that achieve high predictive performance but offer limited interpretability of transaction behaviors. This paper investigates the capabilities of large language models (LLMs) in analyzing real-world Bitcoin transaction graphs through a three-level framework: foundational metrics, characteristic overview, and contextual interpretation. To address token constraints, we introduce LLM4TG, a compact, human-readable graph representation, and CETraS, a connectivity-enhanced graph sampling algorithm. Experiments across five LLMs show that node-level information extraction achieves over 98.50% accuracy on most metrics, while meaningful graph characteristics are identified in up to 95.00% of responses. For address classification with limited labeled examples, top-1 accuracy reaches 50.49% using raw graphs, while top-3 accuracy reaches 72.43% using engineered features. These findings demonstrate the potential of LLMs for transaction graph analysis while revealing limitations in global graph reasoning, precise address classification, and explanation reliability.

Over the past few years, several quantum software stacks (QSS) have been developed in response to rapid hardware advances in quantum computing. A QSS includes a quantum programming language, an optimizing compiler that translates a quantum algorithm written in a high-level language into quantum gate instructions, a quantum simulator that emulates these instructions on a classical device, and a software controller that sends analog signals to a very expensive quantum hardware based on quantum circuits. In comparison to traditional compilers and architecture simulators, QSSes are difficult to tests due to the probabilistic nature of results, the lack of clear hardware specifications, and quantum programming complexity. This work devises a novel differential testing approach for QSSes, named QDIFF with three major innovations: (1) We generate input programs to be tested via semantics-preserving, source to source transformation to explore program variants. (2) We speed up differential testing by filtering out quantum circuits that are not worthwhile to execute on quantum hardware by analyzing static characteristics such as a circuit depth, 2 qubit gate operations, gate error rates, and T1 relaxation time. (3) We design an extensible equivalence checking mechanism via distribution comparison functions such as Kolmogorov Smirnov test and cross entropy

Large language models have made rapid progress in agentic reasoning and tool use, raising expectations for agents that can take over realistic productivity work from humans. However, current agents still struggle to finish such tasks, where success requires following long-horizon professional workflows, understanding heterogeneous file contexts, and producing usable deliverables under fine-grained requirements. A key bottleneck is the lack of scalable post-training data that reflects real work intent, authentic file environments, and deliverable-level evaluation. To close this gap, we introduce Agent-R$^2$, a $\textbf{r}$eal-source data synthesis framework that scales training data for $\textbf{r}$ealistic productivity work. Instead of relying on manually specified workflows or LLM-generated files, Agent-R$^2$ grounds synthesis throughout the pipeline in real occupational seeds, corpus-derived workflow evidence, and authentic web-sourced files. From 10k web-corpus candidates, Agent-R$^2$ distills 4k task instances that pair real multi-format attachments with fine-grained, deliverable-level rubrics. On this data, we train Agent-R$^2$-4B via supervised fine-tuning and rubric-based reinforcement learning, improving both end-to-end productivity performance and file-grounded reasoning. Trained on only 4k instances, Agent-R$^2$-4B surpasses Qwen3-235B-A22B-Thinking-2507 on both GDPval (51.0$\%$ vs. 48.0$\%$) and OfficeQA (14.8$\%$ vs. 13.8$\%$). Further analysis shows that Agent-R$^2$ mitigates shortcut behavior, where agents produce superficially plausible deliverables before sufficiently grounding, constructing, or verifying them, while promoting sustained exploration.

Dynamic flexible assembly flow shop scheduling with multi-product delivery, often with kitting supply and machine-level flexibility, can be found in many applications. It is difficult to design rules manually for it. Genetic programming (GP) has been used to automatically generate dispatching rules, enabling responsive scheduling while meeting high responsiveness demands. However, GP methods are dependent on fixed terminal sets and have poor interpretability. In this work, we develop an evolving dispatching rules framework (EvoDR) that leverages the semantic understanding and generation capabilities of large language models to achieve cross-domain integration of rule design and scheduling knowledge. In our proposed framework, multi-stage assembly supply decisions are modeled as priority sorting of directed edges based on heterogeneous graphs. A dual-expert co-evolution mechanism is proposed, which allows LLM-A to generate code and LLM-S conduct scheduling analysis and reflection. Guided by improvements in hybrid evaluation, adaptive rules that fit dynamic features are continuously evolved. Experimental results show that EvoDR can achieve lower average tardiness than other automatically designed state-of-the-art approaches. It also outperforms expert-designed competitors across 480 instances over 24 distinct scenarios with varying resource configurations and disturbance levels, and EvoDR demonstrates robust adaptability across diverse operating conditions.

Large language models are classically trained in stages: pretraining on raw text followed by post-training for instruction following and reasoning. However, this separation creates a fundamental limitation: many desirable behaviors such as safety, factuality, overall generation quality, and reasoning ability are only added at a late stage, even though the patterns learned earlier strongly shape a model's capabilities. To tackle this issue, we introduce a new way to pretrain and mid-train models that incorporates these behaviors earlier. We utilize an existing strong, post-trained model to both rewrite data and to judge policy model rollouts, thus introducing thinking and reinforcement earlier in training. We further improve reasoning at all stages by introducing the concept of self-improving post-thinking: improving the model's reasoning and self-judging ability by reasoning after generation. In our experiments, we show these approaches can give strong gains across multiple axes.

Self-Driving Chemical Laboratories (SDCLs) are moving chemical experimentation toward embodied automation, where machines must perceive laboratory scenes, reason about experimental states, and act under physical, procedural, and safety constraints. Although Multimodal Large Language Models (MLLMs) are increasingly considered as embodied brains for such systems, existing laboratory benchmarks mainly focus on safety, anomaly detection, or general scientific understanding, leaving their execution-oriented embodied capabilities underexplored. To fill this gap, we introduce BEAKER, a Benchmark for evaluating Embodied Actionability, Knowledge, Experimentation, and Reasoning in SDCLs. BEAKER contains 1,000 expert-curated question-answer samples across 11 image and video task types, covering key judgments required for chemical laboratory execution, including affordance grounding, trajectory planning, equipment, state, and safety recognition, apparatus setup, operation and step reasoning, and temporal procedure reasoning. All questions and annotations are manually created and reviewed by chemistry and embodied Artificial Intelligence (AI) experts, without AI-assisted generation or annotation. We evaluate 28 general-purpose, science-oriented, chemistry-specialized, and embodied MLLMs in an offline, single-turn setting. Results show that current models can follow coarse experimental procedures in videos, but still struggle to ground actions into operable regions and feasible trajectories, to recognize fine-grained laboratory apparatus, states, and safety norms, and to order steps in mid-length procedures. BEAKER thus reveals a clear gap between coarse procedural understanding and the precise, action-grounded perception required for laboratory execution, providing a diagnostic benchmark for future MLLMs in SDCLs.

AI agents are playing an increasingly important role in physics research. However, existing physics benchmarks only evaluate fixed questions, isolated sub-tasks, or paper reproduction with given solutions, and rarely check research quality alongside correctness. We introduce FrontierPhysics, an end-to-end benchmark with 56 tasks made by 42 physicists based on their own research, covering nine domains across theoretical, computational, and experimental physics. Each task requires agents to review the literature, develop a research plan, conduct research and experiments, and produce structured outputs and scientific manuscripts. Combining deterministic verifiers with expert-authored rubrics, FrontierPhysics evaluates whether agents' research is correct and publishable. Across 22 models in 26 configurations, no configuration passes more than 46 of its 168 rollouts. Manual review of agent trajectories shows that frontier agents' main weaknesses are insufficiently rigorous reasoning and limited expert intuition, rather than gaps in physics knowledge or execution errors.

AI research agents automate the cycle of proposing, implementing, and evaluating experiments, opening a path toward recursive self-improvement. Yet their ability to propose experiments outpaces their capacity to execute them in real environments, making outcome prediction a key capability for sustained self-improvement under limited experimental budgets. We investigate language models as Research World Models (RWMs), which predict the outcomes of candidate interventions across research environments. Our evaluation draws on over 2,600 experimental records from nine research environments spanning pretraining, post-training, and inference, representing more than 171,000 H100 GPU-hours of experimentation. Research knowledge acquired from real experimental experience improves RWM predictions of unseen interventions within the same environment (Spearman +0.27), and can be reused across environments. For example, using only pretraining experience from OLMo3, Marin, and Nanochat, an RWM reduces selection regret in the Qwen3 environment by 78% compared with zero-experience setting. These benefits extend to multi-round Autoresearch under a fixed selection budget: RWMs with in-env and cross-env research knowledge increase the best gain achieved by 15.8% and 11.6%, respectively. Ablations across 13 language models used as RWMs show that adding research knowledge can improve intervention ranking more than changing models or increasing reasoning effort alone. These findings support language models as RWMs and motivate accumulating experimental data for future RWM training.

Financial backtests of large language models (LLMs) are meaningful only if the model did not already know the outcomes it is asked to predict. Practitioners guard against this by trusting a vendor's stated knowledge cutoff, by asking the model what it knows, or by instructing it to answer as of a past date. To verify whether any of these safeguards can be trusted, we introduce FinDendro, a benchmark of 2,226 dated U.S. facts from 1999 to 2026, covering monthly macroeconomic series, month-end stock prices and Federal Reserve decisions. Each fact yields four prompts: a value question, Recall and Predict questions that pair each candidate value with a claim of knowing or guessing without a declared date, and a Predict question with a historical declared date. Controls for persistence and dominant outcomes keep chance accuracy near 0.27. Results show that for most frontier models, self-reports understate knowledge. GPT-5 answers 76% of controlled facts correctly but claims to know 38%, and answers 61% of the facts it says it does not know. Knowledge claims can persist beyond the measured knowledge boundary, by up to seven quarters, and a declared date removes most of the claims of prior knowledge while leaving accuracy unchanged. Backtests inside a model's training window therefore cannot be certified leak-free by self-report, instruction or stated cutoff. These findings motivate behavioral measurement of temporal knowledge boundaries, and FinDendro provides a benchmark and procedure for selecting backtest periods based on demonstrated factual knowledge.

World models learn to predict the temporal evolution of visual observations given a control signal, potentially enabling agents to reason about environments through forward simulation. Because of the focus on forward simulation, current world models generate predictions based on factual observations. For many emerging applications, such as comprehensive evaluations of physical AI behavior under varying conditions, the ability of world models to answer counterfactual queries -- such as "what would happen if this object was removed?" -- is of increasing importance. We formalize counterfactual world models that additionally take interventions as explicit inputs, predicting temporal sequences under hypothetical modifications to observed scene properties. Traditional world models operate directly on entangled pixel-space representations where object properties and relationships cannot be selectively modified. This modeling choice prevents targeted interventions on specific scene properties. We introduce CWMDT, a framework to overcome those limitations, turning standard video diffusion models into effective counterfactual world models. First, CWMDT constructs digital twins of observed scenes to explicitly encode objects and their relationships, represented as structured text. Second, CWMDT applies large language models to reason over these representations and predict how a counterfactual intervention propagates through time to alter the observed scene. Third, CWMDT conditions a video diffusion model with the modified representation to generate counterfactual visual sequences. Evaluations on two benchmarks show that the CWMDT approach achieves state-of-the-art performance, suggesting that alternative representations of videos, such as the digital twins considered here, offer powerful control signals for video forward simulation-based world models.

Frontier models such as GPT-6 Astra now directly control robots and can generalize better than vision-language-action models (VLAs). We argue that this ability could arise from two distinct, composable skills: semantic reasoning, to understand what a task asks for in a scene, and motion control, to move the robot to do it. We show that small open VLMs are already capable of the first and mainly lack the second. We introduce Zero-real Agent Policy (ZAP), a recipe that teaches this missing motion skill, entirely in simulation on simple primitive tasks and from the model's own successes, with no frontier teacher or distillation and no real-world robot data. ZAP reaches 96.7% success on real-world tasks, matching GPT-6 Astra in our harness while running 3x faster and 17x cheaper, and 74.0% on a bimanual adaptation of RoboLab-120, on par with GPT-6 Astra's 72.2%. ZAP also generalizes to new objects, goals, and strategies, and in simulation to a robot embodiment on which it never trained.

Efficient preconditioning plays an important role in large-scale first-order optimization. In this paper, we propose an adaptive accelerated optimistic dual averaging ($\text{A}^2$ODA) method for unconstrained convex optimization that adopts low-cost preconditioning based on $p$-norm geometry for $p\in[2,\infty]$. The method is line-search-free and employs a new stepsize selection technique based on an aggregated affine lower model. Its update structure is simple and unified across $p\in[2,\infty]$, with the case $p>2$ requiring no more elaborate updates than the Euclidean case $p=2$. Under a generalized local smoothness condition, we derive a convergence bound that separates the effects of constant and gap-dependent curvature, with the latter potentially yielding faster-decaying terms. In the special case of local smoothness, this bound recovers a nearly optimal convergence rate.

In LLM-based recommendation, memory acts as a user-level agent in a memory-augmented backbone. Yet users and items remain asymmetric: only the user is given a structured profile; it enters a bandwidth-limited reranker prompt; and it is frozen after warmup. We recast LLM-based recommendation as an agentic-communication process in four stages: BUILD a self-description, TRANSMIT it along the collaborative graph, UNDERSTAND the other side, and SYNC it from the next interaction. This formulation reveals three unresolved communication challenges: the item side never produces a comparable self-description (one-sided BUILD); profile-level signals enter only the bandwidth-limited reranker prompt rather than retrieval, where they would not compete for context tokens (prompt-trapped UNDERSTAND); and the user's description is frozen after warmup with no principled update rule (absent SYNC). We instantiate the missing stages as HARP (Homophily-aware Agentic Recommendation via dual Profiles), a training-free, three-operator addition to a frozen LLM backbone: a rule-extracted symmetric item profile (zero extra LLM calls), a channel-decomposable profile-level homophily H(phi_u, phi_i) consumed at retrieval time rather than inside the reranker prompt, and a single-pass reflective update realising in-context profile evolution without any gradient step. On four public benchmarks (Yelp, Books, MovieTV, Goodreads), HARP lifts Hit@1 over the strongest published baseline by 6.9%-24.0%, with family-wise Holm-corrected p < 1e-9 on every dataset. Ablations and a per-case attribution study show that the operators contribute complementary gains, with the retrieval-side gate as the main source; we also report the trade-off introduced by the reflective pass. We formulate agentic communication and provide a training-free instantiation. Code is available at https://github.com/Jove-Zhen/HARP.

Skill memories encode past experiences as textual prompts to improve reinforcement learning (RL) for self-evolving agents. Complementing RL, On-Policy Self-Distillation (OPSD) provides dense, fine-grained supervision to internalize knowledge from skill-guided trajectories and accelerate policy learning. However, self-generated memories often contain noisy supervision and functionally entangled, redundant, or conflicting knowledge, making direct distillation difficult. Under limited training budgets, these issues can hinder knowledge internalization and even degrade performance. To address this problem, we propose MDSKILL, a multi-view distillation approach for skill memory internalization. MDSKILL organizes skill memories into three complementary functional views (state grounding, action planning, and execution control) and selects high-quality skills using dynamic metrics to construct view-specific and integrated reasoning branches. OPSD then transfers and consolidates complementary experiential knowledge across these branches into a shared policy, enabling efficient skill internalization and continuous agent self-evolution. We evaluate MDSKILL on three agent benchmarks against RL, skill-augmented, and self-distillation baselines that do not rely on stronger external teacher models. Without additional computational overhead, MDSKILL improves upon SOTA performance by up to 7.1% and 8.1% on ALFWorld and WebShop. Extensive ablations and evolution analyses further validate its effectiveness in knowledge internalization and policy improvement. The code is available at https://anonymous.4open.science/r/MDSKILL-CF53.

Motivated by recursive self-improvement (RSI), we study how LLM agents can learn through external memory without updating model parameters. Extracting useful experience from a single attempt is difficult: flawed reasoning can produce success, while poor execution can obscure a sound strategy. Preserving its value over time is also difficult: our accumulation baseline without consolidation initially improves but eventually performs below a no-memory baseline. Our key idea is to compare attempts at the same task to identify reusable decisions and refine when they apply. We introduce Hippo, a memory framework inspired by selective hippocampal encoding and replay that learns from these comparisons when shortcomings are diagnosed. It compares the resulting experience with existing memory to merge overlaps and resolve conflicts through explicit applicability conditions. This allows prior guidance to be revised without storing every lesson as a separate entry. Experiments on SWE-bench Verified, WebArena, and Mind2Web show performance gains in all three settings and substantially fewer retained experiences than the accumulation baseline. The gains extend across the evaluated base models of varying capabilities, indicating that memory-based adaptation remains useful even for highly capable models.

Infant pose and shape estimation is essential for applications in childcare, developmental monitoring, and medical diagnosis. However, existing methods and datasets are largely designed for adults, and direct transfer to infants fails due to substantial differences in body proportions, articulation limits, and frequent self-occlusion. To address this gap, we introduce InfantNet, the largest real-image infant dataset to date, comprising 108,902 RGB images of infants aged 6-18 months. Each image is annotated with 2D keypoints, and a curated subset of 11,642 images additionally includes 3D pose and shape annotations with full SMIL parameters, verified through multi-view rendering and reprojection-error filtering. We use an iterative annotation pipeline to ensure high fidelity across both 2D and 3D labels. InfantNet establishes a large-scale, comprehensive benchmark for infant 2D keypoint detection and 3D pose-and-shape recovery. Baseline experiments demonstrate that state-of-the-art adult pose estimators do not generalize well to infants, whereas fine-tuning on InfantNet yields a consistent improvement. The gains are even more pronounced for 3D pose and shape estimation, where SMIL-based estimation demonstrates robust performance with accuracy comparable to results reported for SMPL-based models. By releasing the InfantNet dataset and benchmark, we provide a vital resource for advancing infant pose analysis and related healthcare applications.

We introduce Continuous-Action World Models (CoAWM), a framework for learning cross-embodiment spatiotemporal dynamics priors that support robot action generation across embodiments and execution frequencies. At its core is a shared latent function space that captures the spatial and temporal structure of robot actions. We learn this space through two stages. First, function pretraining brings trajectories from different embodiments and sampling frequencies into a common representation, learning shared dynamics priors through action reconstruction and consistency across temporal sampling views. By accommodating heterogeneous robot trajectories, this stage broadens the range of data sources available for pretraining and facilitates scaling to larger and more diverse datasets. Second, we connect the pretrained function space to a world-model backbone, enabling observations, language instructions, and the current robot state to condition the prediction of a latent action function. A single decoder takes any query time and returns the corresponding end-effector action. By separating shared dynamics priors from task-conditioned action selection and temporal sampling, our approach provides a unified foundation for transferring robot experience across embodiments and control frequencies.

Mathematical supervision for language models is typically flattened into isolated problem-solution pairs, discarding the curricular organization and evidential structure of the original sources. We introduce TOME-Math (Textbook-Oriented Mathematical Evidence), a large-scale corpus that converts 2,118 textbooks spanning the core areas of university-level mathematics subjects into approximately 380K structured proof-supervision records while preserving chapter order, proposition-proof links, dependency references, and source provenance. Leveraging these structural signals, we explore curriculum learning across multiple mathematical subjects and characterize scaling patterns with supervision volume and source breadth. At full scale, fine-tuning on TOME-Math raises the unweighted mean score across our held-out evaluations and established public benchmarks by 16.7 percentage points over the base model, while exceeding all three public instruction-data baselines evaluated in aggregate. Our findings highlight structured textbooks as both a scalable source of mathematical supervision and an empirical framework for studying the impact of data organization on mathematical reasoning. We will release the corpus to facilitate further research on structured mathematical supervision.

Scaling critic capacity is a promising direction for improving off-policy reinforcement learning (RL). However, recent work shows that larger critics are prone to overfitting and instability in replay-based bootstrapped training. In this paper, we propose using Low-Rank Adaptation (LoRA) as a structural regularizer for critic learning from scratch. Our approach freezes randomly initialized base matrices and optimizes only the corresponding low-rank adapters, thereby constraining the learned weight changes to low-rank residuals. We evaluate our method on state-based continuous-control tasks, using SAC with SimbaV2 and BRC on DMC-Hard, and FastTD3 with its SimbaV2 variant on IsaacLab. Empirically, LoRA reduces critic loss during training and improves overall policy performance, achieving the best or competitive results on most tasks. Extensive experiments demonstrate that our low-rank updates provide a simple and effective form of structural regularization for critic learning in off-policy RL.

High-quality articulated assets are essential for embodied perception, simulation, and manipulation, yet constructing them at scale remains difficult. Manual annotation and procedural modeling require substantial human effort. Recent agentic systems automate asset construction through program synthesis, but repeat costly reasoning and refinement for every new asset. Our key observation is that assets within a category can share construction logic despite variations in structure and appearance. Agentic synthesis and procedural generation are therefore complementary: agents can construct reusable programs, while procedural execution reuses the resulting logic across many samples. We therefore present RFW-A, which combines the two in a new articulated-asset generation paradigm. Starting from existing articulated programs, RFW-A organizes reusable components and their dependencies into category-level generators. Agents test and refine these generators across sampled configurations and motion states. Once constructed and validated, the generators produce diverse assets through procedural execution with no further model calls or token charges. The estimated API cost of generator authoring averages $10.62, amortizing to $0.05 per asset over 200 assets, versus $1.24 for the sampled per-asset agentic synthesis baseline. Using RFW-A, we built RFW-A-400K, containing over 400K sim-ready articulated assets across more than 500 categories. Experiments show high structural and kinematic validity and stable execution across physics simulators. Beyond asset validity, training with RFW-A-400K improves part-level articulation estimation on seen and unseen categories, while policies trained in simulation with retrieved and program-edited assets transfer zero-shot to real articulated objects.

Large Language Models (LLMs) are considered to be a more efficient and cost-effective alternative to human judgment for Machine Translation (MT) evaluation. With MT evaluation spanning a large number of language pairs, domains and levels of annotation granularity, LLMs must be thoroughly evaluated across these dimensions before being reliably used as alternatives to human evaluation. In this paper, we evaluate the performance of LLMs for two prominent MT quality evaluation schemes: Multidimensional Quality Metrics (MQM) and Error Span Annotation (ESA) by comparing their agreement with human annotators. We present results on a long-context test set of 70 language pairs and the publicly available WMT23 and WMT25 data, investigating both score and error span annotation agreement across a variety of language pairs and domains. Our results show that while LLM agreement with human annotators exceeds agreement between human annotators for some evaluation tasks, both vary substantially across annotation schemes, language pairs and domains and remain unreliable for most settings. Furthermore, we identify challenges facing both human and LLM annotators: humans are particularly challenged by fine-grained MQM annotations and low-resource language pairs, while LLMs struggle with minor errors, wrong language variants and error span annotation. Our results highlight the potential for improvement for both human and LLM annotation performance, possibly through human-LLM collaborative annotation pipelines that address the reliability issues identified in this work.

Function calling significantly extends the application boundary of large language models, where high-quality and diverse training data is critical for unlocking this capability. However, real function-calling data is quite challenging to collect and annotate, while synthetic data generated by existing pipelines tends to lack coverage and accuracy. In this paper, we present ToolACE, an automatic agentic pipeline designed to generate accurate, complex, and diverse tool-learning data. ToolACE leverages a novel self-evolution synthesis process to curate a comprehensive API pool of 26,507 diverse APIs. Dialogs are further generated through the interplay among multiple agents, guided by a formalized thinking process. To ensure data accuracy, we implement a dual-layer verification system combining rule-based and model-based checks. We demonstrate that models trained on our synthesized data, even with only 8B parameters, achieve state-of-the-art performance on the Berkeley Function-Calling Leaderboard, rivaling the latest GPT-4 models. Our model and a subset of the data are publicly available at https://huggingface.co/Team-ACE.

In long context scenarios, large language models (LLMs) face three main challenges: higher computational cost, performance reduction, and position bias. Research indicates that LLM performance hinges on the density and position of key information in the input prompt. Inspired by these findings, we propose LongLLMLingua for prompt compression towards improving LLMs' perception of the key information to simultaneously address the three challenges. Our extensive evaluation across various long context scenarios demonstrates that LongLLMLingua not only enhances performance but also significantly reduces costs and latency. For instance, in the NaturalQuestions benchmark, LongLLMLingua boosts performance by up to 21.4% with around 4x fewer tokens in GPT-3.5-Turbo, leading to substantial cost savings. It achieves a 94.0% cost reduction in the LooGLE benchmark. Moreover, when compressing prompts of about 10k tokens at ratios of 2x-6x, LongLLMLingua can accelerate end-to-end latency by 1.4x-2.6x. Our code is available at https://aka.ms/LongLLMLingua.

Despite the advancements of open-source large language models (LLMs), e.g., LLaMA, they remain significantly limited in tool-use capabilities, i.e., using external tools (APIs) to fulfill human instructions. The reason is that current instruction tuning largely focuses on basic language tasks but ignores the tool-use domain. This is in contrast to the excellent tool-use capabilities of state-of-the-art (SOTA) closed-source LLMs, e.g., ChatGPT. To bridge this gap, we introduce ToolLLM, a general tool-use framework encompassing data construction, model training, and evaluation. We first present ToolBench, an instruction-tuning dataset for tool use, which is constructed automatically using ChatGPT. Specifically, the construction can be divided into three stages: (i) API collection: we collect 16,464 real-world RESTful APIs spanning 49 categories from RapidAPI Hub; (ii) instruction generation: we prompt ChatGPT to generate diverse instructions involving these APIs, covering both single-tool and multi-tool scenarios; (iii) solution path annotation: we use ChatGPT to search for a valid solution path (chain of API calls) for each instruction. To enhance the reasoning capabilities of LLMs, we develop a novel depth-first search-based decision tree algorithm. It enables LLMs to evaluate multiple reasoning traces and expand the search space. Moreover, to evaluate the tool-use capabilities of LLMs, we develop an automatic evaluator: ToolEval. Based on ToolBench, we fine-tune LLaMA to obtain an LLM ToolLLaMA, and equip it with a neural API retriever to recommend appropriate APIs for each instruction. Experiments show that ToolLLaMA demonstrates a remarkable ability to execute complex instructions and generalize to unseen APIs, and exhibits comparable performance to ChatGPT. Our ToolLLaMA also demonstrates strong zero-shot generalization ability in an out-of-distribution tool-use dataset: APIBench.