对比式强化学习（CRL）通过将状态--动作对与随后到达的状态建立关联来学习目标条件价值，其前提是不存在任何未观测变量同时影响记录的动作及其结果。当离线示范来自一个掌握了学习者无法获取的信息的示范者时，这一前提便不再成立，评论家会吸收虚假的动作--结果关联。我们利用混杂马尔可夫决策过程对这一情形进行形式化，并引入反事实转移，以区分数据能够识别的内容与无法识别的内容；在所有相容转移中进行最小化，可得到一个对真实占用率构成下界的最坏情况占用率。因果对比式强化学习（CauCRL）从正向和负向离线示范中学习一种悲观的反事实转移，并在反事实展开上训练评论家；在存在混杂的目标到达任务中，其成功率显著高于标准 CRL。
The Model Context Protocol provides a shared interaction substrate for the agentic web, allowing heterogeneous services to expose capabilities through a standardized interface. Learning to use this ecosystem entails a hierarchy of capabilities: tool-format skills, server-level transfer, and cross-server composition. We present MCPDisCo, an ecosystem-scale reinforcement learning framework for MCP-native agents. Using a Thought-Markov Decision Process (Thought-MDP), the framework explicitly represents internal reasoning alongside routing, execution, and termination and trains on real-server, long-horizon interactions spanning 70 MCP servers and 527 tools across six domains. We study the reasoning--tool seesaw, a joint-learning failure mode in which improvement in one objective can accompany deterioration in the other. To mitigate potential interference, MCPDisCo combines token-masked reward attribution with bidirectional gradient surgery: the former separates direct supervision for reasoning and action spans, while the latter removes conflicting components from both objective gradients in shared parameters. MCPDisCo achieves higher success with fewer chain-of-thought tokens. On the out-of-distribution MCP-Atlas and the GAIA benchmark, MCPDisCo maintains monotone gains, confirming that the learned behavior generalizes beyond the training server distribution and broader agentic tasks. Model weight and implementation are available.
Activation steering has recently been used as a memory for skills: past experiences of an agent are stored as steering vectors and retrieved later. Facts and preferences about the user are assumed to need text instead. We test that assumption with a training-free activation memory beside a frozen LLM. The change an experience causes in late-layer states is stored as a shift, filed under a whitened key for the conversation's gist, and added back in a later session only when the gist matches. We argue that such a shift is a fixed average of an effect that attention, in context, recomputes for every question, so only question-independent content can survive. In a pre-registered study on Qwen3.5-2B and 9B, with judges from four model families, the memory is selective (about 98% of unrelated replies are unchanged), but facts are stored as the word to say, in the last layers only, and are almost never used (at most 15% clean use, against 96-98% in context). Leanings transfer only in a narrow strength window. Most strikingly, the reference subtracted when writing a preference decides whether its concept or its direction survives: unless the opposite is subtracted, a stored dislike is recalled as a like.
Formulaic alpha mining seeks to discover executable expressions that capture predictive market signals in a vast, noisy search space. Recent LLM-based approaches use prompting, iterative search, or seed-factor evolution, leaving direct factor construction through policy learning less explored. To address this, we introduce AlphaRLM, a reinforcement-learning framework that trains LLMs to construct executable symbolic factors from financial literature without input factor expressions. AlphaRLM incorporates Literature-Guided Operator Assembly (LEGO), which conditions factor construction on literature-derived ideas and executable operators, providing semantic priors for exploration. It further aligns the generation policy with market feedback through executable-validity and grouped-return-separation rewards that capture non-monotonic structures across factor quantiles. Extensive experiments on CSI 300, CSI 1000, and S&P 500 show that AlphaRLM outperforms strong factor-mining and LLM-based baselines in annualized excess return. Notably, the 8B model achieves strong cross-market performance, highlighting the promise of literature-conditioned, market-aligned factor generation.
Can an autonomous AI agent reverse engineer a binary as well as a skilled human analyst? Do they adopt the same strategy, and can agents scale to real-world programs? We compare human and LLM-based software reverse engineering (SRE): nine models tackle two compact static-analysis challenges previously solved by human analysts, scored with a common understanding rubric, and we record their complete interactions with IDA Pro. This lets us compare not only what humans and agents recover, but also how they investigate a binary. The best agents (GPT-5.5 Pro, Claude Opus 4.8, and Kimi K3) achieve statistically human-competitive understanding on both tasks, while weaker models remain far behind. Ablations show that the gap depends on code representation and workflow demands, not on model capability alone: open-weight models improve substantially with richer code representations, whereas frontier models show no significant change. Similar success rates hide different workflows. Humans repeatedly revisit functions and follow local program structure; strong agents jump across the binary, retrieve information broadly, and rarely return to inspected code, a behavior we call greedy fetching. This strategy is not fixed: given semantically richer code, agents fetch less without losing understanding. Because greedy fetching aggregates large amounts of code in the agent's context, we measure context-window saturation across 4,000 real-world binaries: large windows accommodate many programs, but larger applications and lower-level representations quickly exhaust them. Autonomous reverse engineering is becoming competitive, with a large speed advantage, but scaling it will require better navigation, selective retrieval, and persistent analysis state, not merely larger context windows.
长时域智能体必须完成一系列工作流，其中后续活动依赖于前序输出以及共享状态的变化。评估面临的一项关键挑战，是在确保任务可执行且结果可验证的同时，扩展任务要求及其依赖关系。我们提出了 SkillHorizon，这是一个通过技能组合在七种商业环境中构建长时域任务的基准。其生成器利用特定领域规则，将可复用的技能流程转化为具体的商业活动。这些活动通过输入—输出依赖、共享资源和状态更新相互连接，使工作负载与协调需求同步增长。成功的参考执行结果用于验证任务可行性并衡量交互工作负载，而程序化评估器则评估局部完成情况和整体任务成功率。我们在 1,050 个任务上评估了五个语言模型，参考交互工作负载范围从 64K 到 1M 个词元。在这些规模下，平均局部目标完成率从 90.4% 降至 46.8%，因为已完成的工作未能跟上不断增长的要求。在更大的工作负载下，智能体更常提交未成功的最终答案，同时仍有大量工作未完成且保留着工具调用额度。轨迹分析还发现了一些案例：脚本能够成功执行，但其输出未能通过后续的商业检查。SkillHorizon 提供了一个可配置的测试平台，用于研究不断增长的工作流中的持续进展、依赖协调和完整任务交付。我们的代码已发布于 https://github.com/KnowComp/SkillHorizon。
With the rapid evolution of large language models (LLMs), there is an increasing demand for effective personalization techniques that can adapt model behavior to individual user preferences. Personalized LLMs must infer a user's query-dependent preferences from limited, heterogeneous, and evolving interaction histories. Existing approaches rely primarily on in-context personalization, parametric adaptation, or personalized reward modeling, but respectively incur recurring costs in context management, model optimization and deployment, or multi-candidate generation and scoring. These limitations become particularly restrictive when user observations are sparse and the preference evidence relevant to each query varies across contexts and over time. To address these challenges, we propose \modelname, a fine-grained and instance-tailored steering framework that dynamically generates sample-level interference vectors from user data and injects them into the model's forward pass for personalized adaptation. Our approach introduces: match contrast that produces user-relevant signals, fine-grained steering component that captures nuanced signals by hooking activations from attention and MLP layers, and query-conditioned aggregation module that synthesizes these signals into contextually relevant enhancements. Extensive experiments across diverse scenarios, including short-to-long text generation, and web function calling, validate the effectiveness and compatibility of \modelname. Results show that our method consistently brings significant performance gains, and show extreme superiority in fast-shifting and data-sparse environments. Additionally, \modelname rids us of training cycles and incurs light overhead of storage or inference latency. These make \modelname a strong and real-world deployable method. Implementation is available \footnote{https://anonymous.4open.science/r/FleetPilot}.
We present a generator of Thai klon pad verse, one stanza from a short topic, and evaluate it against human verse on held-out topics. The generator is a masked language model initialised from WangchanBERTa over a fixed canvas with two layers: a sound layer of pronounced syllables, where rhyme, tone and syllable count are enforced as hard masks, and a word layer of whole words, so that no word is cut into written pieces. Decoding commits one cell at a time and samples at temperature 0.4; one sample is drawn per topic, with no selection, reranking or repair. On 78 held-out topics never used in development, judged by three independent runs of a frozen LLM judge in the same pools as the human stanzas from which the topics were written, 70.5% of the generated stanzas pass form, lexical cleanness of all four lines and topic alignment together, against 76.9% for the human stanzas (difference -6.4 points, 95% CI [-19.2, +6.4]). Cleanness is level with human (85.9% vs. 87.2%), and the stanzas pass form and cleanness together slightly more often than the human ones (+3.8 points, [-6.4, +14.1]); the difference lies in alignment (84.6% vs. 91.0%), on topics that were written from the human stanzas. The low temperature matters: at higher temperatures the same model passes clearly fewer of the same topics, because lexical errors lie in the model's lower-probability choices; sharpening the sampling distribution removes most of them, at a cost in vocabulary (7.5% rare words against 16.5% in the human stanzas) and diversity. The pre-registered target, all-three within 5 points of human, is missed by 1.4 points. KlonSuphap-LM, the only public klon pad generator we know of, passes none of the 78 held-out topics when judged in separate pools together with the method's and the human stanzas (form 19.2%, cleanness 3.8%).
Recurrent algorithmic reasoners are trained with per-step hints on small graphs and run for many more steps on large ones. On breadth-first search in SALSA-CLRS, we find that a reasoner can reproduce every training hint and still fail on deep graphs: after some exact steps, it declares nodes reached before the search gets there. We call this a premature flood and trace two routes to it. In our GRU processor with a degree input, the flood is carried in the hidden state and set off when that input is in the tail of the training distribution. Setting the input higher at inference brings the flood sooner, and none of the six flooding models counts steps. Without the input, summing the neighbors' states instead of averaging them also breaks the processor. In the benchmark's own GIN and PGN, trained with its code, some seeds instead flood from a fixed step, on sparse and dense graphs alike. Clamping the degree input or adding 200 dense training graphs removes every failure of the GRU processor. Detaching the state between steps avoids the flood in that processor (0 of 23 seeds, against 6 under full backpropagation) but not with sums or a residual update, and two published remedies leave floods. None of the benchmark's test sets is both deep and dense, but its deepest one shows the second route: on its Delaunay graphs at 1,600 nodes, one GIN seed floods from step 18. We therefore recommend auditing the range of every input and aggregate, which catches the first route, and testing on deep graphs, which catches both.
Generative AI makes options cheap and shifts the bottleneck of creative work from generating options to screening them. We argue that AI's value depends on whether its options arrive before or after the team's screen. In our model, early options compete for a place in the portfolio; late options arrive after that competition and are kept unscreened. In a preregistered experiment, we randomly assigned 148 three-person teams to five schedules of access to the same language model during a product-design task with timed search and selection stages. Moving the same 15 minutes of access from search to selection lowered expert-rated creativity by 0.62 standard deviations, although teams used the model at similar rates. Platform logs locate the difference at the screen. Early access enlarged the pool of ideas fivefold, and teams pruned it. Late-access teams reached the screen with the same small pool as teams without AI and then searched again, devoting 29% of their selection-stage queries to generating new options, compared with 11% for teams with access throughout. Late access added breadth, not quality: these teams submitted 73% more proposals than did teams without AI. Language-model judges detected the benefit of access but not the cost of late timing.
强化学习中的平均回报设定具有挑战性，因为已知其 Bellman 算子只有在底层马尔可夫链具有混合性时才在跨度半范数下收缩。现有工作通常假设收缩性质成立，例如通过环境的混合性来保证；而我们考虑的设定是：给定一个与奖励无关的 $m$ 步混合策略，通过向该混合策略正则化，控制所学策略的收缩性质和探索性质。我们对可采用多种正则化方式实例化的 $Q$-学习算法提供了有限时间分析，并给出了相对于正则化不动点的理论保证。我们的框架揭示了依赖于正则化器的新型对象，它们控制着收缩性、探索性和算子偏差。在单步设定下，我们的收敛速率与 $Q$-迭代和同步 $Q$-学习的已知最优速率一致；对于 $m$ 步混合，我们推导出了新的速率。我们还研究了 $Q$-规划的变体，并表明规划相当于时间局部化的学习，其中规划时域 $H$ 在边界值引入的偏差与样本复杂度之间进行权衡。最后，我们通过数值研究验证了分析结果：(i) 已知一个具有强混合性的策略能够改善收敛和探索，且其优势随瓶颈严重程度的增加而增大；(ii) 我们的理论能够预测 $α$-散度族中各正则化器在收敛和偏差方面的排序。
Choosing a few robust algorithms or schedules can require evaluating every candidate under many adverse scenarios, even though an agent only needs to identify the winners. A predictor may suggest those winners, but cannot rule out a surprise in an unevaluated scenario. We introduce Certified Adaptive Selection (CAS), which returns the exact top-$k$ set as soon as remaining uncertainty cannot change it. CAS keeps guaranteed lower and upper bounds on Expected Shortfall (ES), the average loss in the worst scenarios. Its stopping test compares the worst admissible selected candidate with the best admissible excluded one; for information-disjoint Cartesian intervals this one-cut test is necessary and sufficient. Prediction has a separate role: selecting the next outcome to evaluate, never authorizing the decision. For empirical ES with exact unit-cost queries, we compute in polynomial time the fewest evaluations that could have proved the realized decision in hindsight. Our prediction-guided agent CAS-PG is exact for every predictor, including adversarial ones. On two public algorithm-selection matrices, CAS-PG certifies the exact top three after 630 of 3150 and 981 of 2200 replayed queries, or $1.14$ and $1.18$ times that hindsight minimum. Without historical outcomes, its transcript-only variant requires $1.28$ and $1.37$ times the minimum, outperforming the tested non-predictive certified policies. Separately, a prediction-free CAS policy avoids $93.0\%$ of exact recourse solves in synthetic energy scheduling. One certificate thus supports several acquisition policies and separately proved risk/cost extensions: prediction changes the work, not the justified choice.
表格基础模型通过上下文学习（ICL）适应新数据集，并且和大多数 Transformer 一样，将计算深度与独立参数化层的数量绑定在一起。我们探讨这样一个问题：此类模型的 ICL 阶段是否需要每一层都有一组不同的参数，还是只需要增加深度。我们引入 Tab-Loop，以4个共享参数的模块重复执行 $k$ 次，替代 TabICLv2 中12个参数不共享的 ICL 模块，其他部分完全不变：编码器、解码器、合成先验和训练方案均与一个在相同计算量下训练的非共享参数基线保持一致。在预训练计算量相同（18.2 EFLOP）、推理计算量相同（应用模块12次）的条件下，Tab-Loop 的参数量减少62%，但性能与基线相当（10.4M 对 27.6M）：留出集对数损失为0.4583 对 0.4588，在200个 TALENT 数据集上的平均准确率为84.10% 对 83.92%，在38个 TabArena 数据集上的 Elo 得分为1409 对 1396。一个参数量与 Tab-Loop 相同的4模块非共享参数对照模型，在相同计算量下接触到多32%的训练表格数据，但表现落后于二者（83.46%；Elo 得分1343）。由于权重共享，深度成为推理时可以选择的项。在训练覆盖的循环次数范围内，准确率会随着循环次数增加而提升，并在超出该范围后保持稳定；通过以更深的循环次数继续训练，可扩展有效范围，并在达到基线推理计算量4--5倍时，取得基线的阶段2得分。在这一规模下，表格基础模型的 ICL 阶段更需要深度，而不是更多彼此不同的参数。
语言模型通常通过不完美的评估器根据人类偏好进行对齐，而这些评估器的错误可能在优化过程中被利用。我们训练策略来搜索这样的比较：在这些比较中，评估器可能会反转底层偏好，从而为进一步监督提供候选样本。由于策略训练期间无法获得人类偏好，我们使用偏好模型集成之间的分歧，作为预测误差的替代指标。我们将这一搜索过程表述为两个对立策略之间的零和Stackelberg博弈：Leader试图最小化分歧，而Follower则最大化分歧。对于一类一般的有限上下文Stackelberg博弈，我们证明了均衡的存在性，并证明在正则化条件下均衡具有唯一性和闭式刻画。我们提出Stackelberg Mirror Descent-Ascent（SMIDA），这是一种双时间尺度算法，在表格化设置下具有线性的末次迭代收敛速度；同时，我们还开发了相应的神经策略梯度版本。使用合成oracle进行的实验表明，分歧可以作为预测误差的有效替代指标。在留出数据上，训练得到的策略优于参考策略，并提高了超过分歧阈值且反转oracle偏好的比较所占比例。
多环境 POMDP（ME-POMDPs）通过编码一组有限的可能环境来推广 POMDP，从而刻画模型不确定性。ME-POMDP 的有限时域鲁棒奖励最大化问题要求寻找一个策略，使其在最差环境中的奖励最大化；近期研究表明，该问题是 $\mathsf{PSPACE}$-完全的。我们通过确定哪些子类仍保持这种 $\mathsf{PSPACE}$-完全性，拓展了 ME-POMDP 的复杂性版图。具体而言，我们选取单环境情形下易处理的 POMDP 子类，并研究这些类别扩展到多环境设置后的计算复杂性。首先，我们利用某种对混合运算封闭的条件，证明了多环境盲 MDP 和确定性 POMDP 的 $\mathsf{NP}$-完全性。随后，我们证明了一系列不满足该闭包条件的子类是 $\mathsf{PSPACE}$-完全的，从而为多观测 POMDP 提供了新的复杂性结果。最后，我们证明，在时变马尔可夫策略或有限状态控制器下，众多对于标准 POMDP 属于 $\mathsf{NP}$-完全的 ME-POMDP 问题是实数存在理论（$\exists \mathbb{R}$）-完全的。这些结果揭示了 POMDP 与 ME-POMDP 复杂性之间一种微妙的关系，而这种关系在很大程度上取决于所涉及的子类。
Recent bidirectional encoders have been modernized with innovations from causal large language models (LLMs). However, the pretraining recipe itself remains inherited from small-scale conventions, leaving encoders underoptimized, with no single model performing competitively across both classification and retrieval tasks. In this work, we investigate how far encoder performance can be pushed by revisiting core design choices in training, revealing three key findings: (1) Encoders saturate on downstream tasks beyond a few trillion tokens, indicating that additional compute past this point does not help. (2) Across model sizes, lower masking ratios benefit classification while higher ones benefit retrieval, contradicting prior work suggesting larger ratios for larger models. (3) Distillation with pruning improves classification at low masking ratios but degrades retrieval at high ones. Guided by these findings, we present UltraBERT, a single encoder that assigns separate LM heads to low and high masking ratios within a batch and restricts the distillation objective to the low ratio. To our knowledge, UltraBERT is the first encoder to achieve state of the art on both classification and retrieval tasks, reaching 91.0 on GLUE and 47.7 on BEIR. Together, these results show how far encoder performance can be pushed and provide a principled basis for compute allocation, masking ratio design, and knowledge transfer in bidirectional models.
基于患者自述症状的自动疾病分类，需要在轻量级判别模型的速度与自回归大语言模型（LLM）的上下文理解能力及延迟成本之间取得平衡。决策基础模型提供了另一种选择：无需生成自由文本，而是将表示直接映射到候选类别。我们将基于 Qwen3.5 骨干网络构建的 90 亿参数决策模型 Clef-Flash，针对 Symptom2Disease 基准中的 24 类疾病分类任务进行适配。采用结合差分学习率的 4 位量化低秩适配（QLoRA），并将数据分层划分为 960 个训练样本和 240 个测试样本后，该模型取得了 95.42% 的诊断准确率和 0.9551 的宏平均 F1 分数。这一结果优于已报告的零样本基线：远程 API 为 61.67%，本地未适配初始化模型为 65.00%；同时也超过了 ClinicalBERT 基线（90.90%）。通过单次前向传播评估候选类别的 logits，无需进行自回归解码，模型在单块 GPU 上对每条记录的推理耗时低于 80 毫秒，据报告，相较生成式 LLM 基线实现了 31 倍加速。为考察症状层面的证据，我们提出了留一症状法（Leave-One-Symptom-Out，LOSO）反事实扰动框架。在一个银屑病示例中，移除定义疾病的症状描述后，预测类别的置信度从 100.00% 降至 0.12%，并将预测结果转为脓疱病，置信度为 99.59%；而移除次要的全身性症状时，置信度变化小于 0.01%。这些发现表明，基于决策的架构能够将大型基础模型的语言能力与低延迟推理及症状层面的可解释性相结合，从而服务于临床分诊。
工业异常检测基准发展迅速，但总体上仍然是静态的，因此无法充分反映这样一种部署场景：随着时间推移，检测系统必须从一种工业组件扩展到另一种工业组件。在本文设定中，持续工业异常检测被定义为一个任务增量问题：每个组件对应一个任务，而学习目标仍然是半监督的。其目标超越了传统异常检测的范畴，因为模型在依次接触新组件的过程中，应当保持在所有组件上的性能。这一设定尤为重要，因为许多现有工业流程要么假设拥有完全监督的缺陷标签，要么仅研究相互孤立的数据集，因此半监督异常检测器在一系列异构组件上的表现仍是一个悬而未决的问题。为弥补这一空白，我们将基准组织为一系列相互关联的组件级任务，并定义了多种场景，包括随机、由易到难和由难到易。除单数据集持续学习场景外，我们还提出了跨多个数据集的挑战性场景。进一步地，我们从两个互补的角度评估持续学习性能：图像级异常检测和像素级异常定位。通过将工业异常检测重新定义为持续学习基准，而非静态的训练/测试问题，我们为研究不断演化的工业环境中的知识保持、适应性和泛化能力，建立了一条切实可行的路径。
Beliefs in a Partially Observable Markov Decision Process (POMDP) are posterior distributions over the hidden state given the agent's history of actions and observations. They summarize everything the history says about the state, so planning in a POMDP is planning in a Markov decision process (MDP) over beliefs. Belief MDPs have continuous state spaces even for finite POMDPs, and exact planning in them is intractable in general. Planners must therefore approximate beliefs and estimate values. Monte Carlo planners do both inside a search tree, with an approximate belief at every node and a value estimate at every leaf. Approximations usually consist of sets of sampled states, with leaf values coming from rollouts, hand-designed bounds, or learned networks. Policies and values computed from these approximations are limited by their quality, and prior methods with learned networks rely on hand-chosen summaries of the sampled states. A belief representation should be computable from the parent node's representation through the POMDP's dynamics, let the search sample from it, and be learned so that it provides the information that policies and value functions need. We present Belief Embedding Tree Search (BETS), a Monte Carlo planner that meets these requirements by representing each node's belief with a generative distribution embedding (GDE) while updating it through the POMDP dynamics. BETS trains its belief model alongside its policy and value networks through experience, with the realized hidden state of the POMDP as the belief model's training target. We evaluate BETS on six configurations of RockSample, LightDark, and PocMan against BetaZero and three planners without learned components. BETS is competitive or leading on all six, while every baseline falls far behind on at least one. Where beliefs are highest-dimensional, it outperforms them by a large margin, even without search.
Self-evolving agents are usually scored by endpoint Gain: held-out success after an update minus success before. That difference can credit a frozen rescore as evolution. We introduce Trace Evolution Bench (TEB), a protocol that separates that rescore from a context or weight update, together with three frozen 80/20/40 packs. An update is accepted only when validation does not fall, held-out identifiers never enter the update, and a text update is not ranked with a weight update. The primary scoreboard is coding, graded by hidden unit tests in a closed workspace. On that pack (40 held-out tasks, one family), DeepSeek v4 Pro under no_evolve never changes its policy yet moves 7.5, 5.0, and 12.5 percentage points across three seeds. Of nine text-update cells inspired by GEPA, ACE, and ExpeL, seven have positive excess above that seed's own band, and one gepa_llm seed moves down by the width of its band (-12.5, regression 0.48). GLM and MiniMax do not reproduce that seven-of-nine pattern. No method stays above its own band on every seed of all three models: GLM has two paired cells above their own bands, and MiniMax expel_llm seed 44 is a third (regression 0.30). The methods were fixed on a validation screen before held-out was read. Office, graded by rule-bounded delivery, and manufacturing table QA are separate packs: office DeepSeek does not exceed its own bands (widest 17.5 points), and manufacturing has no paired positive excess beyond its seed band.
大型语言模型可以利用语言上下文比较 N-best 候选，并在需要时改写转写文本，从而在无需重新训练识别器的情况下，通常改善 WER。然而，仅依赖提示词的纠错方式较为脆弱：当 top-1 结果正确时可能过度编辑，在参考答案不在候选列表中时可能产生幻觉，而且往往需要规模非常大的模型。我们提出了一种 Oracle 引导的微调方法，通过三个经过筛选的数据池，将 N-best 列表转化为监督信号：纠错样本（参考答案在列表中但不在 top-1 中）用于教授选择，正则化样本（top-1 正确）用于抑制不必要的改写，以及困难负样本（参考答案不在列表中）用于抑制幻觉。在这些数据上进行微调，对于学习混合式“选择或改写”策略而言，比随机子集微调提供了更有效的监督；在专有数据集上，从约 3B 参数规模的模型开始即可观察到最明显的收益，但在 CommonVoice 上的结果则不太确定。这种差异表明，公开的朗读语音文本可能与 LLM 的预训练数据存在重叠，从而掩盖了真实的后编辑收益。
Accurate and efficient analysis of dental radiographs has become increasingly important for disease diagnosis and clinical decision making. In particular, dental radiographs segmentation plays a critical role as it is challenging due to several reasons, such as low contrast, noise in the images, variations in the image quality, etc. Previous studies have primarily addressed these challenges using convolutional neural networks (CNN)-based models. However, direct comparison of these segmentation models is still challenging due to the variations among the experimental settings. Therefore, we mainly compared four segmentation models, including Pyramid Scene Parsing Network (PSPNet), DeepLabV3, UNet, and UNet++, through a fair, controlled experimentation setting under varying conditions such as image sizes and batch sizes. For this, we used 1269 dental radiography images, and prior to training, we implemented a series of preprocessing techniques. To assess the models, we used several metrics, including Dice coefficient (DICE), Intersection over Union (IoU), accuracy, precision, training and validation losses. According to our results, UNet++ demonstrated the highest IoU score of 0.6719, and a DICE value of 0.7869, showing superior results in terms of segmentation, specifically when configured with an image size of 300x300 and a batch size of 16. In addition, we also explored the effect of image resolutions and batch sizes on model performance and accuracy under different architectures. Therefore, this study provides insights into the strengths of different segmentation models and the potential of deep learning-based semantic segmentation methods for improving automated dental radiograph analysis.
支持边缘人工智能（AI）推理作为网络服务颇具挑战，因为当前移动系统能够传输张量，却无法暴露或理解赋予这些张量实际运行意义的推理上下文。已有研究通过动态分区、协同推理和分割流水线来弥合这一差距；然而，这些方案仍然受应用约束、对网络无感知或缺乏上下文。本文提出 Inference on Demand（IoD），这是一种按需推理抽象，通过类似 CAMARA 的应用程序接口（API）暴露分割推理，并在 5G 用户面路径中以带内方式传输感知分割的张量上下文。我们在基于可编程用户面功能（UPF）的测试平台上，使用可分区的视觉 AI 工作负载对 IoD 进行了评估，结果表明，自适应分割选择可以作为面向网络的服务运行，同时避免静态配置僵化的成本特征。结合我们的研究结果，当服务意图、运行时分区和张量传输对移动数据平面可见时，分割推理便不再只是卸载。这为网络原生 AI 服务开辟了道路，使可编程电信基础设施能够为面向用户的应用协调计算、上下文和连接。
Adapting to a new world and reasoning correctly about which outcomes are possible under rules one has not met before is a basic capacity underlying general intelligence. We study it in small discrete worlds: sets of grid configurations consistent with a shared set of rules, concretely Sudoku-like puzzles built from a common base and a handful of extra rules. We train a single depth-recurrent model on many combinations of rules and test it at four levels of novelty: a new puzzle, a new combination of known rules, a pairing of rules never seen together, and a rule instance not seen at all during training. The model receives the active rules either as a symbolic description or as two solved examples, and a model that gets no rule information is used as a baseline. At recombining rules it was trained on, the description matches or exceeds examples ($99\%$ against $95\%$ on new combinations). On rule instances whose components never co-occurred in training, the description falls below the no-information baseline while examples stay above it. The framework allows us to test many such generalization regimes in more systematic way than benchmarks such as ARC-AGI and provides new ground for evaluating depth-recurrent reasoning models, which are typically tested on saturated Sudoku benchmarks.
As text-to-speech (TTS) systems continue to advance, provenance tracking and security verification of synthesized speech have become increasingly important. Audio watermarking provides active provenance by embedding imperceptible, machine-recoverable information into speech. However, existing post-hoc methods typically modify the waveform after synthesis, leaving the watermark vulnerable to compound perturbations. To address this limitation, we propose GAMW, a watermark injection framework integrated into the TTS generation process. Before waveform synthesis, GAMW embeds a content-adaptive multi-bit watermark residual into the generated representation, and its decoding objective combines focal modulation with dynamic per-bit weighting. We train and evaluate GAMW separately on two TTS backbones. With a 16-bit payload, GAMW improves decoding accuracy over the strongest baseline by 9.2 and 23.5 percentage points under individual and compound perturbations, respectively, while maintaining competitive speech quality. Reliable clean-speech decoding is also retained at higher payloads. Code is available at https://github.com/ZConJugaTe/GAMW.
将推理与工具调用交织进行的视觉智能体，为多模态任务提供了一个颇具前景的方向。然而，有效工具调用的学习通常依赖大规模精心整理的问答对或专家示范，而获取这些数据的成本高昂。在本研究中，我们提出 Marvo，这是一种简单而有效的自我演化方法，能够从未标注图像中学习工具调用。Marvo 由两个智能体组成：（1）挑战者，从未标注图像中生成问答对；（2）求解器，通过强化学习解答问题，从而学习工具调用。为了促进有效的工具调用，优质的训练问题应当足够困难，能够促使求解器使用工具，同时又应处于求解器的能力范围内。为此，我们向挑战者引入边际奖励，其定义为求解器在有无工具访问权限时的性能差异。借助边际奖励，我们构建了一种自适应课程学习机制，鼓励求解器学习有效的工具调用。在涵盖智能体推理、视觉搜索、数学推理和通用现实世界推理的各项基准测试中，Marvo 将平均准确率最高提升了 7.1\%，证明了边际奖励是无需人工监督、扩展问题复杂度的更优方式。
论证结构体现了单个论证如何在话语中相互作用。然而，利用这些结构推断有意义的知识仍然具有挑战性，因为这需要对论证影响如何在结构中传播进行形式化表示。我们提出 Neuro-symbolic Argumentation Reasoning System（NARS），用于直接从在线讨论中检测说服力。其神经组件采用检索增强的迭代式 LLM 推理来提取双极论证结构，而符号组件则应用渐进式论证语义，在结构中传播论证影响。我们使用 Webis-CMV-20 语料库，构建了 CMV2BAS 数据集。该数据集包含经过 LLM 推导的双极论证结构标注以及 delta 事件标签（观点改变或未改变）的长篇、多线程 ChangeMyView 讨论。我们研究了渐进式论证语义能否有效检测说服力、将其与论证质量和基于图的中心性相结合是否能够提升性能，以及在现代 LLM 已经具备相关能力的情况下，符号推理是否仍然必要。在最佳配置下，NARS 达到 F1=0.57 和 AUC=0.56，优于零样本 Qwen3.6:27B 和 GPT-oss:20B。我们发现，以支持为主导的讨论以及较长的论证链是限制攻击影响传播的主要失败模式。本研究展示了在嘈杂的真实世界话语上实施形式化渐进式论证语义、开展可扩展的宏观层面说服力分析的可行性，同时指出了神经符号论证推理未来研究面临的关键挑战和发展方向。我们公开发布 CMV2BAS 数据集和 NARS 框架，供公众使用。
Quantization plays a crucial role in efficiently deploying deep learning models on resource-constrained devices. Post-training quantization does not require access to the original dataset or retraining the full model. Current methods that achieve high performance close to baseline results typically require INT8 fixed-point integers. However, achieving higher model compression with lower bit-widths often leads to significant performance degradation. In this paper, we propose VLCQ, a post-training quantization method that relaxes the fixed-point encoding constraint, enabling more flexible exploration of quantization techniques through variable-length encoding. This approach achieves competitive or improved performance while reducing bit-widths, without requiring access to training data or model fine-tuning. Extensive experiments on image classification, segmentation, and object detection demonstrate that VLCQ improves model compression and accuracy compared with state-of-the-art post-training quantization approaches. For per-channel quantization, VLCQ surpasses FP32 accuracy and Piece-Wise Linear Quantization (PWLQ) on most evaluated models, achieving up to 6x compression compared with FP32 and up to 1.7x compared with PWLQ. When prioritizing compression with limited accuracy loss, VLCQ achieves up to 12.25x compression compared with FP32 with less than 4% performance degradation. For per-tensor quantization, VLCQ is competitive with Data-Free Quantization (DFQ), while providing greater flexibility to achieve lower bit rates across different tasks and models.
AI coding agents operate in iterative loops: they modify code, run tools, inspect the results, and decide what to do next. We call the design of this process loop engineering, which includes context management, model selection, compaction, and termination. Existing efficiency work mainly compresses tool observations or routes selected turns to cheaper models, but the practical opportunity for these optimizations remains unclear. We analyze 600 publicly released coding-agent trajectories spanning two agent scaffolds and two language models. Median observations contain only 118-286 tokens, while the large build logs commonly used to motivate compression occur near the ninety-ninth percentile. Moreover, 98.6% of input tokens are served from a prefix cache, so flat input pricing overestimates loop cost by roughly 7x on average under near-ideal caching. We derive and empirically validate a closed-form break-even condition for model delegation. Delegation increases cost in half of the evaluated settings, near the predicted boundary. More broadly, turn count explains 84% of the variance in log trajectory cost, compared with 4% for mean observation size; the most expensive trajectories have the most turns despite having the smallest observations. These associations suggest that observation-focused methods target a secondary cost factor, and identify unnecessary turns as the larger and largely unexamined opportunity, though we do not establish that turns can be removed while holding task success fixed. We release TurnCost, a tool for re-scoring trajectory corpora under alternative caching, pricing, compression, and routing policies.
Chain-of-thought prompting enables reasoning in large language models but produces explanations that are difficult to verify, expensive to generate, and not always faithful. We introduce a Structural Chain-of-Thought framework where models generate executable computational graphs instead of free-form reasoning. The final answer is obtained by deterministically executing the graph. The model is trained using graph-based reinforcement learning that rewards structural validity, answer correctness, and node overlap with reference graphs. Experiments across five benchmarks and four model families show that our method consistently improves accuracy over supervised graph training and can reduce generation length by up to 5-6x compared with standard GRPO and chain-of-thought supervised training. Our faithfulness experiments reveal that models trained with our approach tend to rely on deriving their answers from the generated graphs rather than producing post-hoc rationalizations as usually in chain-of-thought reasoning.
What do you think will be the hot trend in 2027 with on-device AI? Comments URL: https://news.ycombinator.com/item?id=50081396 Points: 2 # Comments: 0
The AI Hub provides 91 curated lesson plans teaching middle and high school students about AI - how it works, what the risks and benefits of AI are, and how it is impacting society. These activities come from six different nonprofit organizations that are building educational resources for classrooms. Our nonprofit worked with 16 different teachers to comb through more than 200 different resources, with the teachers vetting these resources to find the best resources for their classrooms. The end result is 91 different lesson plans and activities that cover a wide range of different topics. Schools across Pittsburgh will be using these resources over the course of this year to help teach their students about AI, and they will be scaled up nationally over the coming years. If you're a human that has thoughts or feedback, please drop a comment! What's top of mind: "What else should we teach students?" I'd appreciate any ideas you all have as we continue to build and refine this library of resources. Comments URL: https://news.ycombinator.com/item?id=50081297 Points: 1 # Comments: 0