Recent advancements in large language models (LLMs) have shown very impressive capabilities in code generation across many programming languages. However, even state-of-the-art LLMs generate programs that contains syntactic errors and fail to complete the given tasks, especially for low-resource programming languages (LRPLs). In addition, high training cost makes finetuning LLMs unaffordable with constrained computational resources, further undermining the effectiveness of LLMs for code generation. In this work, we propose SLMFix, a novel code generation pipeline that leverages a small language model (SLM) finetuned using reinforcement learning (RL) techniques to fix syntactic errors in LLM-generated programs to improve the quality of LLM-generated programs for domain-specific languages (DSLs). In specific, we applied RL on the SLM for the program repair task using a reward calculated using both a static validator and a static semantic similarity metric. Our experimental results demonstrate the effectiveness and generalizability of our approach across multiple DSLs, achieving more than 95% pass rate on the static validator. Notably, SLMFix brings substantial improvement to the base model and outperforms supervised finetuning approach even for 7B models on a LRPL, showing the potential of our approach as an alternative to traditional finetuning approaches.

Large Language Models (LLMs) have demonstrated remarkable capabilities in challenging, knowledge-intensive reasoning tasks. However, extending LLMs to perceive and reason over a new modality (e.g., vision), often requires costly development of large-scale vision language models (VLMs) with LLMs as backbones. Smaller VLMs are more efficient and adaptable but often lack the broad knowledge and reasoning capabilities of frontier LLMs. In this work, we propose BeMyEyes, a modular, multi-agent framework for extending LLMs to multimodal reasoning by orchestrating collaboration between efficient, adaptable VLMs as perceivers and powerful LLMs as reasoners through conversations. We then introduce a data synthesis and supervised fine-tuning pipeline to train the perceiver agent to effectively collaborate with the reasoner agent. By combining the complementary strengths of perception and reasoning agents, BeMyEyes avoids the need for training large-scale multimodal models, preserves the generalization and reasoning capabilities of LLMs, and allows flexible extension to new domains and modalities. Experiments show that our framework unlocks the multimodal reasoning capabilities for LLMs, enabling a lightweight and fully open-source solution, i.e. equipping text-only DeepSeek-R1 with Qwen2.5-VL-7B perceiver, to outperform large-scale proprietary VLMs such as GPT-4o on a wide range of knowledge-intensive multimodal tasks. These results demonstrate the effectiveness, modularity, and scalability of our multi-agent approach for building future multimodal reasoning systems.

Large-scale video generative models have recently demonstrated strong visual capabilities, enabling the prediction of future frames that adhere to the logical and physical cues in the current observation. In this work, we investigate whether such capabilities can be harnessed for controllable image-to-video generation by interpreting visual signals embedded within the frames as instructions, a paradigm we term In-Video Instruction. In contrast to prompt-based control, which provides textual descriptions that are inherently global and coarse, In-Video Instruction encodes user guidance directly into the visual domain through elements such as overlaid text, arrows, or trajectories. This enables explicit, spatial-aware, and unambiguous correspondences between visual subjects and their intended actions by assigning distinct instructions to different objects. Extensive experiments on three state-of-the-art generators, including Veo 3.1, Kling 2.5, and Wan 2.2, show that video models can reliably interpret and execute such visually embedded instructions, particularly in complex multi-object scenarios.

Advances in object tracking and acoustic beamforming are driving new capabilities in surveillance, human-computer interaction, and robotics. This work presents an embedded system that integrates deep learning-based tracking with beamforming to achieve precise sound source localization and directional audio capture in dynamic environments. The approach combines single-camera depth estimation and stereo vision to enable accurate 3D localization of moving objects. A planar concentric circular microphone array constructed with MEMS microphones provides a compact, energy-efficient platform supporting 2D beam steering across azimuth and elevation. Real-time tracking outputs continuously adapt the array's focus, synchronizing the acoustic response with the target's position. By uniting learned spatial awareness with dynamic steering, the system maintains robust performance in the presence of multiple or moving sources. Experimental evaluation demonstrates significant gains in signal-to-interference ratio, making the design well-suited for teleconferencing, smart home devices, and assistive technologies.

Accurate lung cancer tumor staging is crucial for prognosis and treatment planning. However, it remains challenging for end-to-end deep learning approaches, as such approaches often overlook spatial and anatomical information that are central to the tumor-node-metastasis system. The tumor stage depends on multiple quantitative criteria, including the tumor size and its proximity to the nearest anatomical structures, and small variations can alter the staging outcome. We propose a medically grounded hybrid pipeline that performs staging by explicitly measuring the tumor's size and distance properties rather than treating it as a pure image classification task. Our method employs specialized encoder-decoder networks to precisely segment the lung and adjacent anatomy, including the lobes, tumor, mediastinum, and diaphragm. Subsequently, we extract the necessary tumor properties, i.e. measure the largest tumor dimension and calculate the distance between the tumor and neighboring anatomical structures by a quantitative analysis of the segmentation masks. Finally, we apply rule-based tumor staging aligned with the medical guidelines. This novel framework has been evaluated on the Lung-PET-CT-Dx dataset, demonstrating superior performance compared to traditional deep learning models, achieving an overall classification accuracy of 91.36%. We report the per-stage F1-scores of 0.93 (T1), 0.89 (T2), 0.96 (T3), and 0.90 (T4), a critical evaluation aspect often omitted in prior literature. To our knowledge, this is the first study that embeds explicit clinical context into tumor stage classification. Unlike standard convolutional neural networks that operate in an uninterpretable "black box" manner, our method offers both state-of-the-art performance and transparent decision support.

Pixel diffusion aims to generate images directly in pixel space in an end-to-end fashion. This approach avoids the limitations of VAE in the two-stage latent diffusion, offering higher model capacity. Existing pixel diffusion models suffer from slow training and inference, as they usually model both high-frequency signals and low-frequency semantics within a single diffusion transformer (DiT). To pursue a more efficient pixel diffusion paradigm, we propose the frequency-DeCoupled pixel diffusion framework. With the intuition to decouple the generation of high and low frequency components, we leverage a lightweight pixel decoder to generate high-frequency details conditioned on semantic guidance from the DiT. This thus frees the DiT to specialize in modeling low-frequency semantics. In addition, we introduce a frequency-aware flow-matching loss that emphasizes visually salient frequencies while suppressing insignificant ones. Extensive experiments show that DeCo achieves superior performance among pixel diffusion models, attaining FID of 1.62 (256x256) and 2.22 (512x512) on ImageNet, closing the gap with latent diffusion methods. Furthermore, our pretrained text-to-image model achieves a leading overall score of 0.86 on GenEval in system-level comparison. Codes are publicly available at https://github.com/Zehong-Ma/DeCo.

Physics simulators are essential in science and engineering, enabling the analysis, control, and design of complex systems. In experimental sciences, they are increasingly used to automate experimental design, often via combinatorial search and optimization. However, as the setups grow more complex, the computational cost of traditional, CPU-based simulators becomes a major limitation. Here, we show how neural surrogate models can significantly reduce reliance on such slow simulators while preserving accuracy. Taking the design of interferometric gravitational wave detectors as a representative example, we train a neural network to surrogate the gravitational wave physics simulator Finesse, which was developed by the LIGO community. Despite that small changes in physical parameters can change the output by orders of magnitudes, the model rapidly predicts the quality and feasibility of candidate designs, allowing an efficient exploration of large design spaces. Our algorithm loops between training the surrogate, inverse designing new experiments, and verifying their properties with the slow simulator for further training. Assisted by auto-differentiation and GPU parallelism, our method proposes high-quality experiments much faster than direct optimization. Solutions that our algorithm finds within hours outperform designs that take five days for the optimizer to reach. Though shown in the context of gravitational wave detectors, our framework is broadly applicable to other domains where simulator bottlenecks hinder optimization and discovery.

The challenge of designing effective reward functions in reinforcement learning (RL) represents a significant bottleneck, often requiring extensive human expertise and being time-consuming. Previous work and recent advancements in large language models (LLMs) have demonstrated their potential for automating the generation of reward functions. However, existing methodologies often require preliminary evaluation metrics, human-engineered feedback for the refinement process, or the use of environmental source code as context. To address these limitations, this paper introduces LEARN-Opt (LLM-based Evaluator and Analyzer for Reward functioN Optimization). This LLM-based, fully autonomous, and model-agnostic framework eliminates the need for preliminary metrics and environmental source code as context to generate, execute, and evaluate reward function candidates from textual descriptions of systems and task objectives. LEARN-Opt's main contribution lies in its ability to autonomously derive performance metrics directly from the system description and the task objective, enabling unsupervised evaluation and selection of reward functions. Our experiments indicate that LEARN-Opt achieves performance comparable to or better to that of state-of-the-art methods, such as EUREKA, while requiring less prior knowledge. We find that automated reward design is a high-variance problem, where the average-case candidate fails, requiring a multi-run approach to find the best candidates. Finally, we show that LEARN-Opt can unlock the potential of low-cost LLMs to find high-performing candidates that are comparable to, or even better than, those of larger models. This demonstrated performance affirms its potential to generate high-quality reward functions without requiring any preliminary human-defined metrics, thereby reducing engineering overhead and enhancing generalizability.

The discovery of high-performance photosensitizers has long been hindered by the time-consuming and resource-intensive nature of traditional trial-and-error approaches. Here, we present \textbf{A}I-\textbf{A}ccelerated \textbf{P}hoto\textbf{S}ensitizer \textbf{I}nnovation (AAPSI), a closed-loop workflow that integrates expert knowledge, scaffold-based molecule generation, and Bayesian optimization to accelerate the design of novel photosensitizers. The scaffold-driven generation in AAPSI ensures structural novelty and synthetic feasibility, while the iterative AI-experiment loop accelerates the discovery of novel photosensitizers. AAPSI leverages a curated database of 102,534 photosensitizer-solvent pairs and generate 6,148 synthetically accessible candidates. These candidates are screened via graph transformers trained to predict singlet oxygen quantum yield ($ϕ_Δ$) and absorption maxima ($λ_{max}$), following experimental validation. This work generates several novel candidates for photodynamic therapy (PDT), among which the hypocrellin-based candidate HB4Ph exhibits exceptional performance at the Pareto frontier of high quantum yield of singlet oxygen and long absorption maxima among current photosensitizers ($ϕ_Δ$=0.85, $λ_{max}$=650nm).

State-of-the-art symbolic music generation models have recently achieved remarkable output quality, yet explicit control over compositional features, such as tonal tension, remains challenging. We propose a novel approach that integrates a computational tonal tension model, based on tonal interval vector analysis, into a Transformer framework. Our method employs a two-level beam search strategy during inference. At the token level, generated candidates are re-ranked using model probability and diversity metrics to maintain overall quality. At the bar level, a tension-based re-ranking is applied to ensure that the generated music aligns with a desired tension curve. Objective evaluations indicate that our approach effectively modulates tonal tension, and subjective listening tests confirm that the system produces outputs that align with the target tension. These results demonstrate that explicit tension conditioning through a dual-level beam search provides a powerful and intuitive tool to guide AI-generated music. Furthermore, our experiments demonstrate that our method can generate multiple distinct musical interpretations under the same tension condition.