ManipDreamer: Boosting Robotic Manipulation World Model with Action Tree and Visual Guidance

While recent advancements in robotic manipulation video synthesis have shown promise, significant challenges persist in ensuring effective instruction-following and achieving high visual quality. Recent methods, like RoboDreamer, utilize linguistic decomposition to divide instructions into separate lower-level primitives, conditioning the world model on these primitives to achieve compositional instruction-following. However, these separate primitives do not consider the relationships that exist between them. Furthermore, recent methods neglect valuable visual guidance, including depth and semantic guidance, both crucial for enhancing visual quality. This paper introduces ManipDreamer, an advanced world model based on the action tree and visual guidance. To better learn the relationships between instruction primitives, we represent the instruction as the action tree and assign embeddings to tree nodes, each instruction can acquire its embeddings by navigating through the action tree. The instruction embeddings can be used to guide the world model. To enhance visual quality, we combine depth and semantic guidance by introducing a visual guidance adapter compatible with the world model. This visual adapter enhances both the temporal and physical consistency of video generation. Based on the action tree and visual guidance, ManipDreamer significantly boosts the instruction-following ability and visual quality. Comprehensive evaluations on robotic manipulation benchmarks reveal that ManipDreamer achieves large improvements in video quality metrics in both seen and unseen tasks, with PSNR improved from 19.55 to 21.05, SSIM improved from 0.7474 to 0.7982 and reduced Flow Error from 3.506 to 3.201 in unseen tasks, compared to the recent RoboDreamer model. Additionally, our method increases the success rate of robotic manipulation tasks by 2.5% in 6 RLbench tasks on average.

Killing Two Birds with One Stone: Unifying Retrieval and Ranking with a Single Generative Recommendation Model

In recommendation systems, the traditional multi-stage paradigm, which includes retrieval and ranking, often suffers from information loss between stages and diminishes performance. Recent advances in generative models, inspired by natural language processing, suggest the potential for unifying these stages to mitigate such loss. This paper presents the Unified Generative Recommendation Framework (UniGRF), a novel approach that integrates retrieval and ranking into a single generative model. By treating both stages as sequence generation tasks, UniGRF enables sufficient information sharing without additional computational costs, while remaining model-agnostic. To enhance inter-stage collaboration, UniGRF introduces a ranking-driven enhancer module that leverages the precision of the ranking stage to refine retrieval processes, creating an enhancement loop. Besides, a gradient-guided adaptive weighter is incorporated to dynamically balance the optimization of retrieval and ranking, ensuring synchronized performance improvements. Extensive experiments demonstrate that UniGRF significantly outperforms existing models on benchmark datasets, confirming its effectiveness in facilitating information transfer. Ablation studies and further experiments reveal that UniGRF not only promotes efficient collaboration between stages but also achieves synchronized optimization. UniGRF provides an effective, scalable, and compatible framework for generative recommendation systems.

A General Infrastructure and Workflow for Quadrotor Deep Reinforcement Learning and Reality Deployment

Deploying robot learning methods to a quadrotor in unstructured outdoor environments is an exciting task. Quadrotors operating in real-world environments by learning-based methods encounter several challenges: a large amount of simulator generated data required for training, strict demands for real-time processing onboard, and the sim-to-real gap caused by dynamic and noisy conditions. Current works have made a great breakthrough in applying learning-based methods to end-to-end control of quadrotors, but rarely mention the infrastructure system training from scratch and deploying to reality, which makes it difficult to reproduce methods and applications. To bridge this gap, we propose a platform that enables the seamless transfer of end-to-end deep reinforcement learning (DRL) policies. We integrate the training environment, flight dynamics control, DRL algorithms, the MAVROS middleware stack, and hardware into a comprehensive workflow and architecture that enables quadrotors' policies to be trained from scratch to real-world deployment in several minutes. Our platform provides rich types of environments including hovering, dynamic obstacle avoidance, trajectory tracking, balloon hitting, and planning in unknown environments, as a physical experiment benchmark. Through extensive empirical validation, we demonstrate the efficiency of proposed sim-to-real platform, and robust outdoor flight performance under real-world perturbations. Details can be found from our website https://emnavi.tech/AirGym/.

Learning and Generating Diverse Residential Load Patterns Using GAN with Weakly-Supervised Training and Weight Selection

The scarcity of high-quality residential load data can pose obstacles for decarbonizing the residential sector as well as effective grid planning and operation. The above challenges have motivated research into generating synthetic load data, but existing methods faced limitations in terms of scalability, diversity, and similarity. This paper proposes a Generative Adversarial Network-based Synthetic Residential Load Pattern (RLP-GAN) generation model, a novel weakly-supervised GAN framework, leveraging an over-complete autoencoder to capture dependencies within complex and diverse load patterns and learn household-level data distribution at scale. We incorporate a model weight selection method to address the mode collapse problem and generate load patterns with high diversity. We develop a holistic evaluation method to validate the effectiveness of RLP-GAN using real-world data of 417 households. The results demonstrate that RLP-GAN outperforms state-of-the-art models in capturing temporal dependencies and generating load patterns with higher similarity to real data. Furthermore, we have publicly released the RLP-GAN generated synthetic dataset, which comprises one million synthetic residential load pattern profiles.

EG-Gaussian: Epipolar Geometry and Graph Network Enhanced 3D Gaussian Splatting

In this paper, we explore an open research problem concerning the reconstruction of 3D scenes from images. Recent methods have adopt 3D Gaussian Splatting (3DGS) to produce 3D scenes due to its efficient training process. However, these methodologies may generate incomplete 3D scenes or blurred multiviews. This is because of (1) inaccurate 3DGS point initialization and (2) the tendency of 3DGS to flatten 3D Gaussians with the sparse-view input. To address these issues, we propose a novel framework EG-Gaussian, which utilizes epipolar geometry and graph networks for 3D scene reconstruction. Initially, we integrate epipolar geometry into the 3DGS initialization phase to enhance initial 3DGS point construction. Then, we specifically design a graph learning module to refine 3DGS spatial features, in which we incorporate both spatial coordinates and angular relationships among neighboring points. Experiments on indoor and outdoor benchmark datasets demonstrate that our approach significantly improves reconstruction accuracy compared to 3DGS-based methods.

Climbing the Ladder of Reasoning: What LLMs Can-and Still Can't-Solve after SFT?

Recent supervised fine-tuning (SFT) approaches have significantly improved language models' performance on mathematical reasoning tasks, even when models are trained at a small scale. However, the specific capabilities enhanced through such fine-tuning remain poorly understood. In this paper, we conduct a detailed analysis of model performance on the AIME24 dataset to understand how reasoning capabilities evolve. We discover a ladder-like structure in problem difficulty, categorize questions into four tiers (Easy, Medium, Hard, and Extremely Hard (Exh)), and identify the specific requirements for advancing between tiers. We find that progression from Easy to Medium tier requires adopting an R1 reasoning style with minimal SFT (500-1K instances), while Hard-level questions suffer from frequent model's errors at each step of the reasoning chain, with accuracy plateauing at around 65% despite logarithmic scaling. Exh-level questions present a fundamentally different challenge; they require unconventional problem-solving skills that current models uniformly struggle with. Additional findings reveal that carefully curated small-scale datasets offer limited advantage-scaling dataset size proves far more effective. Our analysis provides a clearer roadmap for advancing language model capabilities in mathematical reasoning.

Bregman Linearized Augmented Lagrangian Method for Nonconvex Constrained Stochastic Zeroth-order Optimization

In this paper, we study nonconvex constrained stochastic zeroth-order optimization problems, for which we have access to exact information of constraints and noisy function values of the objective. We propose a Bregman linearized augmented Lagrangian method that utilizes stochastic zeroth-order gradient estimators combined with a variance reduction technique. We analyze its oracle complexity, in terms of the total number of stochastic function value evaluations required to achieve an \(\epsilon\)-KKT point in \(\ell_p\)-norm metrics with \(p \ge 2\), where \(p\) is a parameter associated with the selected Bregman distance. In particular, starting from a near-feasible initial point and using Rademacher smoothing, the oracle complexity is in order \(O(p d^{2/p} \epsilon^{-3})\) for \(p \in [2, 2 \ln d]\), and \(O(\ln d \cdot \epsilon^{-3})\) for \(p > 2 \ln d\), where \(d\) denotes the problem dimension. Those results show that the complexity of the proposed method can achieve a dimensional dependency lower than \(O(d)\) without requiring additional assumptions, provided that a Bregman distance is chosen properly. This offers a significant improvement in the high-dimensional setting over existing work, and matches the lowest complexity order with respect to the tolerance \(\epsilon\) reported in the literature. Numerical experiments on constrained Lasso and black-box adversarial attack problems highlight the promising performances of the proposed method.

EmbodiedOcc++: Boosting Embodied 3D Occupancy Prediction with Plane Regularization and Uncertainty Sampler

Online 3D occupancy prediction provides a comprehensive spatial understanding of embodied environments. While the innovative EmbodiedOcc framework utilizes 3D semantic Gaussians for progressive indoor occupancy prediction, it overlooks the geometric characteristics of indoor environments, which are primarily characterized by planar structures. This paper introduces EmbodiedOcc++, enhancing the original framework with two key innovations: a Geometry-guided Refinement Module (GRM) that constrains Gaussian updates through plane regularization, along with a Semantic-aware Uncertainty Sampler (SUS) that enables more effective updates in overlapping regions between consecutive frames. GRM regularizes the position update to align with surface normals. It determines the adaptive regularization weight using curvature-based and depth-based constraints, allowing semantic Gaussians to align accurately with planar surfaces while adapting in complex regions. To effectively improve geometric consistency from different views, SUS adaptively selects proper Gaussians to update. Comprehensive experiments on the EmbodiedOcc-ScanNet benchmark demonstrate that EmbodiedOcc++ achieves state-of-the-art performance across different settings. Our method demonstrates improved edge accuracy and retains more geometric details while ensuring computational efficiency, which is essential for online embodied perception. The code will be released at: https://github.com/PKUHaoWang/EmbodiedOcc2.

Type-Constrained Code Generation with Language Models

Large language models (LLMs) have achieved notable success in code generation. However, they still frequently produce uncompilable output because their next-token inference procedure does not model formal aspects of code. Although constrained decoding is a promising approach to alleviate this issue, it has only been applied to handle either domain-specific languages or syntactic language features. This leaves typing errors, which are beyond the domain of syntax and generally hard to adequately constrain. To address this challenge, we introduce a type-constrained decoding approach that leverages type systems to guide code generation. We develop novel prefix automata for this purpose and introduce a sound approach to enforce well-typedness based on type inference and a search over inhabitable types. We formalize our approach on a simply-typed language and extend it to TypeScript to demonstrate practicality. Our evaluation on HumanEval shows that our approach reduces compilation errors by more than half and increases functional correctness in code synthesis, translation, and repair tasks across LLMs of various sizes and model families, including SOTA open-weight models with more than 30B parameters.

Sculpting Subspaces: Constrained Full Fine-Tuning in LLMs for Continual Learning

Continual learning in large language models (LLMs) is prone to catastrophic forgetting, where adapting to new tasks significantly degrades performance on previously learned ones. Existing methods typically rely on low-rank, parameter-efficient updates that limit the model's expressivity and introduce additional parameters per task, leading to scalability issues. To address these limitations, we propose a novel continual full fine-tuning approach leveraging adaptive singular value decomposition (SVD). Our method dynamically identifies task-specific low-rank parameter subspaces and constrains updates to be orthogonal to critical directions associated with prior tasks, thus effectively minimizing interference without additional parameter overhead or storing previous task gradients. We evaluate our approach extensively on standard continual learning benchmarks using both encoder-decoder (T5-Large) and decoder-only (LLaMA-2 7B) models, spanning diverse tasks including classification, generation, and reasoning. Empirically, our method achieves state-of-the-art results, up to 7% higher average accuracy than recent baselines like O-LoRA, and notably maintains the model's general linguistic capabilities, instruction-following accuracy, and safety throughout the continual learning process by reducing forgetting to near-negligible levels. Our adaptive SVD framework effectively balances model plasticity and knowledge retention, providing a practical, theoretically grounded, and computationally scalable solution for continual learning scenarios in large language models.