Computer Science (arXiv)

Large Language Models (LLMs) offer new potential for translation tasks but often experience performance degradation when handling low-resource languages. To address this limitation, we propose an approach for fine-tuning LLMs on a low-resource language, Kupang Malay. Our approach involves designing a set of instructions by leveraging explicit lexical and semantic features from a bilingual dictionary, and introducing Continual Instruction Tuning (CIT), a training paradigm that enables iterative instruction-based training. Experimental results demonstrate that our model, named Lius, yields notable improvements over standard instruction-tuned models by outperforming 4-6 points, and surpassing both Neural Machine Translation (NMT) and Multilingual LLM models by 10-13 points on several evaluation metrics. These findings highlight the potential of our approach to mitigate the reliance on large-scale parallel data in low-resource language translation.

Recent progress in video generation has shown impressive visual synthesis capabilities. However, open-domain customized video generation remains limited by the lack of large-scale, annotated datasets capturing diverse identity-specific attributes. To address this, we introduce PexelsCustom-1M, the first publicly available million-scale dataset for identity-preserving video generation, containing one million curated <identity, text, video> triplets across 8,000+ categories. Leveraging this, we propose CustoMDiT, a parameter-efficient framework that adapts a pretrained multimodal Diffusion Transformer into a customized video generator with only 8% additional learnable parameters. Our method surpasses prior state-of-the-art. However, benchmarks such as DreamBooth cover only 100 classes, which is insufficient for real-world applications. To overcome this, we construct OpenCustom, a new benchmark with 1,000+ categories, created via cross-dataset knowledge fusion from ImageNet and MS-COCO. Extensive experiments confirm the advantages of both our dataset and model. We will open-source the entire ecosystem--including dataset, pipeline, benchmark, and implementations--to support further research.

While 3D Gaussian Splatting (3DGS) achieves impressive real-time rendering, it frequently struggles to synthesize high-frequency textures, a limitation heavily exacerbated in memory-constrained and rate-distortion-optimized (RDO) pipelines. To address this, we propose a versatile 2D perceptual wrapper that enhances the rendered outputs of existing 3DGS representations in a content- and view-dependent manner. Our method leverages a lightweight synthesis network conditioned on pseudo-random Gaussian noise to synthesize perceptually plausible textures. Supervised by Wasserstein Distortion, the network learns to match local feature statistics rather than strictly enforcing pixel-wise reconstruction fidelity, effectively mitigating the blurriness inherent in standard frameworks. We demonstrate the broad applicability of our plug-and-play approach across vanilla, memory-constrained, and RDO 3DGS methods. Comprehensive subjective and objective experiments confirm that our method significantly improves over existing baselines, yielding superior perceptual quality at sharply reduced file or model sizes.

We establish conditions for embedding a corpus of $N$ documents as $d$-dimensional vectors such that every $k$-subset $S \subseteq [N]$ is realizable as a result of top-$k$ retrieval by some query vector. Recent work shows that $d = O(k)$ suffices for such embeddings to exist in $\mathbb{R}^d$, independently of $N$. We theoretically prove that this corpus-independent bound is specific to infinite precision. With $B$ bits per coordinate, perfect top-$k$ retrieval requires $Bd = Ω(k \ln N)$; thus, at any fixed precision, the dimension must grow at least logarithmically with $N$. Specializing to a $\ell_2$-normalized $B$-bit uniform scalar quantization model, we also identify a threshold on the precision $B^{*} = O(\ln \ln N)$ below which no dimension suffices, together with two further regimes that bound the feasible $(B, d)$ pairs. Our result implies that in practical vector databases and dense retrieval systems where quantization is standard, the embedding dimension and possibly the precision must grow with the corpus size.

The need for detecting and sorting batteries is drastically increasing for many applications. This study proves the potential of transfer learning in predicting whether the image contains a battery or not, the location and identifying three types of batteries, namely: prismatic, pouch, and cylindrical Lithium-Ion Batteries (LIB). Particularly, it focuses on the transfer learning method in two applications: Training a large-scale dataset to detect electronic devices using a pre-trained YOLOv5m, then using these latter trained weights to detect and classify the batteries. The precision of battery detection achieves 94%, which outperforms the pretrained YOLOv5m weights with 5%, in 22 ms inference time.

We establish the correct parameter governing the convergence time of the 3-Majority and 2-Choices dynamics on the complete graph in the synchronous model. Recent work [Shimizu and Shiraga, PODC'25] provides matching upper and lower bounds on the number of rounds to consensus, but only in a weak sense: the bounds are shown to coincide for some initial opinion configuration. In contrast, we obtain tight bounds in a strong sense, with upper and lower bounds matching up to logarithmic factors for every initial configuration. Let $α$ (0) be the initial opinion-frequency vector, and denote by ___$α$ (0) ___ $\infty$ its maximum entry. We show that 3-Majority reaches consensus in $Θ$(min{___$α$ (0) ___ -1 $\infty$ , $\sqrt$ n}) rounds w.h.p., while 2-Choices reaches consensus in $Θ$(___$α$ (0) ___ -1 $\infty$ ) rounds w.h.p. Our results demonstrate that the convergence time of both dynamics is governed not by global parameters such as the number of opinions k or the squared ${\ell}$ 2 norm of the initial opinion distribution, but rather by the ''local'' parameter ___$α$ (0) ___ $\infty$ , the maximum initial opinion density.

Anomaly detection plays a critical role in identifying unusual patterns across domains such as fraud detection, network intrusion, and system fault diagnosis. Recently, Christoffel function-based methods, rooted in polynomial optimization, have emerged as promising alternatives to deep learning due to their strong mathematical foundations and computational frugality. However, their practical applicability is hindered by the need to invert a matrix whose size grows exponentially with the data dimension, rendering the method intractable even for moderate-dimensional datasets. This paper addresses the dimensionality limitations of Christoffel function-based anomaly detection while preserving its key theoretical properties, i.e., the on-off support dichotomy behavior and the accurate support shape capture. We introduce UCF, a univariate Christoffel function which is based on the squared distance between the query point and the support points. Extensive experiments on the ADBench benchmark demonstrate that UCF consistently outperforms 14 state-of-the-art baselines in terms of Average Precision. By resolving the scalability bottleneck of the Christoffel Function, this work expands the toolkit of anomaly detection methods with a robust, theoretically grounded, and universally applicable approach.

Optimistic Gradient Descent Ascent (OGDA) and Optimistic Multiplicative-Weights Update (OMWU) are two very popular algorithms to solve convex/concave saddle-point problems, where OMWU is the non-Euclidean, entropic version of OGDA. It is known since the '80s that the last iterate of OGDA asymptotically converges to a saddle point in smooth problems. On the other hand, it is unknown if OMWU has the same property. In this paper, I show that OMWU converges asymptotically for smooth convex-concave saddle-point problems, with a small enough constant learning rate. The result does not require uniqueness, strict complementarity, an error bound, or initialization near a solution. The main new ingredient is a boundary argument showing that every cluster point satisfies the inactive-coordinate KKT inequalities. The boundary argument was discovered with assistance from ChatGPT and is documented in the appendix.

In this paper, we propose a segment-wise soft robotic antenna (SRA) system, where each soft robotic arm referred to as a tentacle, comprises multiple independently controllable segments with bending, elongation-retraction, and sweeping motions. By adjusting segment motion parameters, the positions of surface-mounted antennas are reconfigured, distinguishing it from conventional reconfigurable antenna (RA) systems. Based on this model, we propose two antenna deployment schemes: the segmented end-antenna configuration (SEAC), where fixed antennas are mounted at the segment ends and reconfigured via segment motions; and the hybrid end-and-intermediate antenna configuration (HEIAC), where RAs are further integrated as intra-segment antennas. In HEIAC, soft-robot segment deformation provides large-scale spatial reconfiguration, while RAs enable fine-grained adjustment. For SEAC, we formulate a sum-rate maximization problem accounting for inter-segment connectivity and the nonlinear mapping from segment deformation parameters to antenna coordinates, and develop a penalty dual decomposition-projected gradient ascent (PDD-PGA) algorithm. For HEIAC, we jointly optimize segment deformation, intra-segment antenna positions, and antenna activation using a block coordinate descent (BCD)-PDD-PGA algorithm with greedy backward antenna selection. Simulation results demonstrate that the proposed schemes substantially outperform fixed-position antenna arrays and conventional RA baselines. In particular, SEAC and HEIAC achieve 37.9% and 32.1% sum-rate gains over conventional 3D reconfigurable arrays, respectively, while SEAC provides up to a 49.3% gain in compact array deployments.

Spatial reasoning remains a challenge for Multimodal Large Language Models (MLLMs), as it requires reliable multi-hop inference over both intermediate states and state transitions. Current studies often leave intermediate states unverified and treat state transitions as implicit processes, which limits reliability in multi-hop spatial reasoning. To address this, we propose State-aware Visualization-of-Thought (SVoT), a reinforcement learning framework that generates interleaved, verifiable intermediate states and visualizations. SVoT integrates transition reasoning chains into the generation processes, enabling the model to verify action preconditions and effects through interleaved textual and visual reasoning. We train SVoT via Group Relative Policy Optimization (GRPO), instantiating verification through reward design and evaluating the efficacy of different fine-grained rewards. As existing benchmarks reduce state transitions to single-variable updates, substantially simplifying the problems, we establish five domains by extending classical environments and introducing two novel domains, Pacman and Gather, that require multi-object interactions and numerical reasoning. These domains support systematic evaluation of multi-hop spatial reasoning with quantitative verification of generated intermediate states and transition reasoning. SVoT with transition-aware supervision achieves state-of-the-art performance across the introduced domains, yielding up to a 65% absolute accuracy gain on out-of-distribution test sets.

The European AI Act is the first comprehensive regulation of artificial intelligence (AI), setting out extensive obligations, particularly for so-called high-risk and general-purpose AI systems. A key distinguishing feature of AI systems under the AI Act is the capability to infer. Since the AI Act does not clearly define what inference is, there is a gray area for certain data-driven systems. A specific example is credit scoring systems, which are listed by Annex III of the AI Act. At the same time, however, these are often implemented using statistical models for which it is unclear whether they have the capability to infer and thus fall under the AI definition of the AI Act at all. Motivated by statistical learning theory, this work develops a framework for grading different levels of the capability to infer. Based on the AI Act and the Commission Guidelines on the definition of an artificial intelligence system, we analyze which levels constitute sufficient capability to infer within the meaning of the AI Act and where further regulatory clarity is needed. We illustrate the framework by creating two realistic credit scoring workflows and show whether and where inference occurs in them. Our analysis illustrates that not only individual models but the entire data processing workflow must be considered. It also shows that the involvement of human experts during development can have significant influence on the capability to infer. Code can be found at https://github.com/fraunhofer-iais/inference-framework-creditscorecards.

Blind grasping with a dexterous hand is a crucial manipulation capability. Nevertheless, learning such tactile-only policies for real robots remains challenging due to the tactile sim-to-real gap and the limited expressiveness of sparse tactile signals. To bridge this gap, we propose a framework for tactile-only blind grasping that is deployable on a physical multi-fingered robotic hand. Our approach combines three key components. First, we introduce a Real2Sim tactile calibration pipeline that constructs a contact-calibrated digital-twin simulator capable of reproducing real tactile signals. Second, we improve the expressiveness of sparse tactile observations using a layout-aware tactile encoder, which incorporates sensor-geometry priors through self-supervised pretraining. Third, to improve generalization to unseen objects, we train object-specific reinforcement-learning experts in the calibrated simulator and aggregate their successful grasp trajectories into a tactile-conditioned Diffusion Policy. We evaluate our method on a physical LEAP Hand equipped with distributed tactile sensing across 10 seen and 10 unseen objects. The deployed policy achieves a 27\% real-world grasp success rate across all 20 objects, without real-world grasping demonstrations or visual input. Simulation ablations show that layout-aware tactile pretraining improves grasping performance, while sensing-level evaluations confirm that Real2Sim calibration increases the consistency of tactile contact events between simulation and hardware. Together, these results suggest that contact-event calibration, geometry-aware tactile representation learning, and diffusion-based policy aggregation provide an effective path toward tactile-only blind grasping on real dexterous robotic hands. Project page:Dex-Blind-Grasp.github.io.

Distilling a large speech foundation model (SFM) into an efficient student model has been successfully applied to low-resource environments. Although distillation reduces inference latency, it requires an additional student model training. However, the training efficiency of SFM distillation remains underexplored. In this work, we explore training acceleration of SFM distillation to speed up model deployment. We examine the potential of stacking, in which the model depth is progressively increased through training until the target model depth is reached. While existing stacking methods improve training speed, they suffer from performance degradation. To handle this limitation, we propose interleaved stacking, a novel stacking method that consistently preserves layer position throughout the stacking process. This property is particularly critical in SFMs, in which each layer encodes distinct layer-specific knowledge. We validate the effectiveness of the proposed method on SUPERB.

Recently, constructions of linear complementary pairs (LCPs) of codes and linear complementary dual (LCD) codes on function fields have attracted considerable attention due to the wide range of applications of these codes. Such constructions rely on non-special divisors of degrees $g$ and $g-1$. In this work, we investigate Kummer extensions defined by $y^m = f(x)$ with $f(x)\in\mathbb{F}_q(x)$ and establish an arithmetic characterization of non-special divisors whose support can contain non-totally ramified places. Based on this characterization, we explicitly construct non-special divisors of degree $g-1$ on the GK curve. Moreover, utilizing pure gaps, we explicitly provide several families of effective non-special divisors of degree $g$ on Kummer extensions with the same multiplicities. We then develop a general framework for constructing LCPs of algebraic geometry (AG) codes on Kummer extensions. By virtue of canonical divisors, we show that the security parameters of LCPs of AG codes can be determined within this framework, which also enables the construction of LCD AG codes. Finally, we illustrate our results with representative examples, including LCPs of codes on the GK curve and LCD codes on quotients of the Hermitian curve.

Large language models (LLMs) have demonstrated strong reasoning and instruction-following capabilities, making them potentially powerful tools for time-series analysis. However, time series lie outside their native textual modality, raising a fundamental question: how should time series be represented so that LLMs can reason about them effectively? Existing work typically serializes raw numerical sequences or fine-tunes pre-trained LLMs on time-series data. These approaches place the burden of extracting temporal structure directly on the LLM, creating a modality mismatch that often degrades performance on long sequences and introduces substantial computational overhead. In this work, we introduce Time-Series-to-Structured-Program representation (T2SP), a deterministic, training-free method that represents a time series as a structured symbolic program. T2SP decomposes time series into trends, periods, and salient events, expressing them in a program-friendly format aligned with the textual and code-like modalities on which LLMs are natively trained. By shifting temporal-structure extraction from the model to the representation itself, T2SP enables off-the-shelf LLMs to leverage their existing reasoning capabilities for time-series understanding. We evaluate T2SP on three reasoning tasks -- editing, captioning, and question answering -- where it consistently improves performance, reduces reasoning time, and lowers failure rates compared with raw-string representations. Our results demonstrate that T2SP provides an effective interface between time series and LLMs.

Large language models (LLMs) have achieved remarkable progress in language understanding, reasoning, and generation, sparking growing interest in their creative potential. Realizing this potential requires systematic and scalable methods for evaluating creativity across diverse tasks. However, most existing creativity metrics are tightly coupled to specific tasks, embedding domain assumptions into the evaluation process, and limiting scalability and generality. To address this gap, we introduce an automated, domain-agnostic framework for quantifying LLM creativity across open-ended tasks. Our approach separates the measurement apparatus from the creative task itself, enabling scalable, task-agnostic assessment. Divergent creativity is measured using semantic entropy, a reference-free and robust metric for novelty and diversity, validated against human annotations, LLM-based novelty judgments and baseline diversity measures. Convergent creativity is assessed via a novel retrieval-based multi-agent judge framework that delivers context-sensitive evaluation of task fulfilment with over 60% improved efficiency. We validate our framework in three qualitatively distinct domains: problem-solving (MacGyver), research ideation (HypoGen), and creative writing (BookMIA), using a broad suite of LLMs. Empirical results show that our framework reliably captures key facets of creativity, including novelty, diversity, and task fulfilment, and reveal how model properties, such as size, temperature, recency, and reasoning, impact creative performance. Our work establishes a reproducible and generalizable standard for automated LLM creativity evaluation, paving the way for scalable benchmarking and accelerating progress in creative AI.

Dynamic data pruning techniques aim to reduce computational cost while minimizing information loss by periodically selecting representative subsets of input data during model training. However, existing methods often struggle to maintain strong worst-group accuracy, particularly at high pruning rates, across balanced and imbalanced datasets. To address this challenge, we propose RCAP, a Robust, Class-Aware, Probabilistic dynamic dataset pruning algorithm for classification tasks. RCAP applies a closed-form solution to estimate the fraction of samples to be included in the training subset for each individual class. This fraction is adaptively adjusted in every epoch using class-wise aggregated loss. Thereafter, it employs an adaptive sampling strategy that prioritizes samples having high loss for populating the class-wise subsets. We evaluate RCAP on six diverse datasets ranging from class-balanced to highly imbalanced using five distinct models across three training paradigms: training from scratch, transfer learning, and fine-tuning. Our approach consistently outperforms state-of-the-art dataset pruning methods, achieving superior worst-group accuracy at all pruning rates. Remarkably, with only $10\%$ data, RCAP delivers $>1\%$ improvement in performance on class-imbalanced datasets compared to full data training while providing an average $8.69\times$ speedup. The code can be accessed at https://github.com/atif-hassan/RCAP-dynamic-dataset-pruning

We consider the problem of privately releasing a $k$-dimensional vector under updates: Starting with a zero vector, at times $t_1, t_2,\dots$ the vector is updated by adding $x^{(1)}, x^{(2)},\dots$, respectively. For positive integers $T$, $k$ we model the updates as a data set $\{(t_i, x^{(i)})\}_i$, where $t_i \in [T]$ and $x^{(i)} \in B_k$ (the $k$-dimensional unit ball). Two such data sets are said to be neighboring if their symmetric difference has size at most $1$. The continual release consists of the sum $A^{(t)} = \sum_{i \; : \; t_i \leq t} x^{(i)}$ for each time step $t=1,\dots,T$. Classical continual release techniques allow us to release an approximation of $A^{(1)},\dots,A^{(T)}$ with additive noise of magnitude $\text{polylog}(T)$, computed in time $O(kT)$, even in the on-line, adaptive case where data is continually revealed for the current time step. Motivated by private sketching techniques, we consider the setting where only a \emph{subset} of entries in $A^{(t)}$ need to be released at time step $t$. Our new result is that it is possible to sample any desired entry in a given noise vector in \emph{constant time} while reproducing exactly the distribution of the binary tree mechanism with Gaussian noise. The improvement on the known time bound of $O(\log T)$ comes from a new data structure that allows us to sample a new noise value with the correct correlations in constant time using Brownian bridges. We present two data management applications, of independent interest, that use our technique in conjunction with differentially private CountSketches: 1) A dynamic data structure for orthogonal range counting queries with a better privacy/accuracy/space trade-off than previous data structures, and 2) Join size estimation, where in addition we show improved high-probability bounds.

With the widespread adoption of Large Language Models (LLMs) in software engineering (SE) tasks such as code understanding, debugging, and vulnerability detection, their powerful semantic reasoning ability has also introduced new security and privacy risks. LLMs can analyze, reconstruct, or even reverse-engineer source code logic, potentially leading to the leakage of intellectual property. To address this issue, we propose Acoda, a genetic algorithm-based adversarial code obfuscation framework that defends against LLM-based code analysis. Acoda leverages two key mechanisms of LLMs, namely safety alignment and token-based information processing, to design 8 semantics-preserving obfuscation methods. It iteratively optimizes obfuscation strategies through a genetic algorithm to generate adversarial samples that maximize defensive effectiveness. In addition, we propose a quantitative evaluation framework based on LLM responses, which combines an auxiliary LLM and four evaluation metrics to assess how target LLMs analyze obfuscated code comprehensively. Experimental results show that Acoda can effectively induce LLMs to refuse or misinterpret code analysis. On 7 state-of-the-art LLMs, including GPT-4o, DeepSeek, Qwen, Llama, and Gemma, Acoda achieves an attack success rate (ASR) of up to 70%, with strong cross-model transferability and minimal runtime overhead, while ensuring that the semantics of the original code remain unchanged. Overall, this study provides a new perspective for code protection and LLM security defense in the era of LLMs.

Multi-turn image editing is essential for iterative design, yet current models often struggle with identity drift and error accumulation over successive steps. While existing research leverages video priors for consistency, their reliance on bidirectional attention is fundamentally misaligned with the causal, sequential nature of interactive editing. In this paper, we propose AnchorEdit, the first autoregressive (AR) diffusion-based framework designed specifically for high-resolution, long-term multi-turn editing. AnchorEdit bridges the gap between video priors and causal inference through a three-stage training curriculum: identity-preserving sing-turn pretraining, causal AR forcing fine-tuning with a novel self-rollout strategy to mitigate exposure bias, and consistency distillation for efficient 4-step generation. During inference, we introduce a memory mechanism to anchor the initial subject identity and ensure stable extrapolation across extended editing trajectories. To evaluate performance, we provide a new high-resolution multi-turn editing benchmark designed to stress-test long-horizon stability. Extensive experiments demonstrate that AnchorEdit achieves state-of-the-art results, maintaining exceptional subject fidelity and instruction following even over 10+ interaction rounds.

In 'On Denoting' Russell proposed the most influential theory of definite descriptions, expressions of the form 'the F'. Characteristic for Russell's approach is that definite descriptions are not treated as what they appear to be on the surface, i.e. as singular terms. Instead they are eliminated by a contextual definition. Russell formalises definite descriptions in the context of complete sentences of the form 'The F is G'. This requires scope markers to distinguish, e.g., internal from external negation. It was recognised by Burge, and Kalish and Montague, however, that the essential features of Russell's approach may be formalised while respecting the syntactic category to which definite descriptions appear to belong. An alternative, favoured by Neale, follows Russell in that complete sentences 'The F is G' are formalised by a binary quantifier. The undeniable importance of the theory of definite descriptions for logic, mathematics and philosophy demands that it be formalised to meet the standards of modern proof theory. This is the topic of the present paper. We systematise, compare and extend existing approaches. After presenting its essential features, we formalise Russell's theory of definite descriptions in sequent calculus. Three approaches will be considered. The first uses a binary quantifier, whereas the remaining two employ the term-forming iota operator. The first of these employs only the iota operator, the other employs in addition the lambda operator which does duty as a scope marker. All systems satisfy the standards for modern proof theory, in particular cut elimination. The appendix reformulates these systems in natural deduction, which is more convenient for practical purposes.

Multimodal entity linking (MEL) is the task that consists of matching textual and visual mentions of entities in unstructured data to their corresponding entities in a knowledge base (KB). To be effective in large-scale practical settings, MEL systems must meet three objectives: high linking accuracy, computational efficiency, and storage efficiency, i.e., a compact yet efficient index of the KB. In this paper, we highlight that state-of-the-art systems fail to simultaneously satisfy these 3 requirements. To meet this three-fold objective, we propose FAST-MEL, a lightweight encoder-based MEL solution that relies on a novel and compact fixed-size vectorized representation of both the textual and visual information of each entity or mention. It matches the accuracy of the best systems but performs three orders of magnitude faster. It also consumes one order of magnitude less storage than the fastest systems.

Visual causal reasoning is essential for understanding and intervening in the physical world, requiring identification of causal variables from visual inputs and reasoning over intervention effects. Despite recent progress, large vision--language models (VLMs) remain brittle at such tasks, especially for interventional and counterfactual queries over multi-image inputs. Most existing explorations inject causal knowledge via textual prompts, leaving causal mechanisms external to model execution and limiting reliable control during inference. To address this problem, we propose BridgeVLM, which internalizes visual causal reasoning by inducing a causal graph from multi-image inputs and converting it into structured Causal Tokens executed by RAMP layers injected into the LLM decoder for causal message passing. We further introduce a unified training interface M3S for fine-grained causal supervision from different granularities (local/global level). BridgeVLM achieves 54.4% accuracy on intervention tasks on CausalVLBench (vs. 33.2% with prompt-level supervision), improves results on Causal3D from 43.6% to 49.0%, and substantially improves causal structure learning on CausalVLBench ($F_1$: 33.4% $\rightarrow$ 75.1%).

Large language models are now widely used for everyday learning, but the underlying interactions are typically unstructured chats rather than following a curriculum. Unlike formal online learning systems, these interactions carry no prior record of the student, so any estimate of what the student already knows must be inferred from the dialogue itself. We show that this gap is not closed by scaling models alone. Frontier and education-tuned LLMs perform poorly when asked to tutor a student over an extended session, because doing so requires three things at once. The tutor must sequence a curriculum, conduct Socratic dialogue, and infer the student's knowledge state from that dialogue. We propose separating these responsibilities. Given a student query, our system constructs a prerequisite knowledge graph in which subtopics are nodes and dependencies are edges, and frames tutoring as deciding which node to teach next and how many dialogue turns to spend on it before moving on. A lightweight PPO policy handles this sequencing decision, while an LLM conducts the Socratic exchange at the chosen node and returns a signal of student progress. Across held-out STEM and non-STEM topics, our PPO-paired tutor outperforms heuristic baselines, frontier general-purpose models, and a model specialised for Socratic dialogue: on both the rate at which students reach full curriculum mastery and the number of turns required. Explicit curriculum structure delivers gains that scaling the underlying model does not.

Vision-language-action (VLA) models provide strong visual, language, and action priors for robot manipulation, but visual observations alone often miss the local contact state required for contact-rich tasks. We present TacCoRL, a scalable framework that injects Tactile feedback into VLA policies and improves them through sim-real Co-training and simulation-based reinforcement learning (RL), without requiring large-scale tactile pretraining or extensive real-world contact exploration. The key idea is not only adding touch as an input, but learning how contact readings should modulate action responses in near-failure states that are rare in demonstrations and risky to collect on hardware. We use a real-aligned simulator as a closed-loop training environment for contact interaction. Mixed simulated and real trajectories first warm-start tactile-conditioned actions in the pretrained policy. Reinforcement learning with verifiable task rewards then optimizes the policy using simulated contact rollouts. It reinforces tactile-conditioned actions that lead to task completion, while a supervised objective on real trajectories keeps the refined policy anchored to deployment visual, tactile, and action distributions. The resulting policy transfers directly to the real robot without privileged simulation state or online real-world RL. Across four bimanual contact-rich tasks, the final visuo-tactile policy achieves an average success rate of 72.5%, compared to baseline of 50.0%. Result videos and more details are available at https://tac-corl.github.io/