LeVo: High-Quality Song Generation with Multi-Preference Alignment

Recent advances in large language models (LLMs) and audio language models have significantly improved music generation, particularly in lyrics-to-song generation. However, existing approaches still struggle with the complex composition of songs and the scarcity of high-quality data, leading to limitations in sound quality, musicality, instruction following, and vocal-instrument harmony. To address these challenges, we introduce LeVo, an LM-based framework consisting of LeLM and a music codec. LeLM is capable of parallelly modeling two types of tokens: mixed tokens, which represent the combined audio of vocals and accompaniment to achieve vocal-instrument harmony, and dual-track tokens, which separately encode vocals and accompaniment for high-quality song generation. It employs two decoder-only transformers and a modular extension training strategy to prevent interference between different token types. To further enhance musicality and instruction following, we introduce a multi-preference alignment method based on Direct Preference Optimization (DPO). This method handles diverse human preferences through a semi-automatic data construction process and DPO post-training. Experimental results demonstrate that LeVo consistently outperforms existing methods on both objective and subjective metrics. Ablation studies further justify the effectiveness of our designs. Audio examples are available at https://levo-demo.github.io/.

SongBloom: Coherent Song Generation via Interleaved Autoregressive Sketching and Diffusion Refinement

Generating music with coherent structure, harmonious instrumental and vocal elements remains a significant challenge in song generation. Existing language models and diffusion-based methods often struggle to balance global coherence with local fidelity, resulting in outputs that lack musicality or suffer from incoherent progression and mismatched lyrics. This paper introduces $\textbf{SongBloom}$, a novel framework for full-length song generation that leverages an interleaved paradigm of autoregressive sketching and diffusion-based refinement. SongBloom employs an autoregressive diffusion model that combines the high fidelity of diffusion models with the scalability of language models. Specifically, it gradually extends a musical sketch from short to long and refines the details from coarse to fine-grained. The interleaved generation paradigm effectively integrates prior semantic and acoustic context to guide the generation process. Experimental results demonstrate that SongBloom outperforms existing methods across both subjective and objective metrics and achieves performance comparable to the state-of-the-art commercial music generation platforms. Audio samples are available on our demo page: https://cypress-yang.github.io/SongBloom\_demo.

ZeroGUI: Automating Online GUI Learning at Zero Human Cost

The rapid advancement of large Vision-Language Models (VLMs) has propelled the development of pure-vision-based GUI Agents, capable of perceiving and operating Graphical User Interfaces (GUI) to autonomously fulfill user instructions. However, existing approaches usually adopt an offline learning framework, which faces two core limitations: (1) heavy reliance on high-quality manual annotations for element grounding and action supervision, and (2) limited adaptability to dynamic and interactive environments. To address these limitations, we propose ZeroGUI, a scalable, online learning framework for automating GUI Agent training at Zero human cost. Specifically, ZeroGUI integrates (i) VLM-based automatic task generation to produce diverse training goals from the current environment state, (ii) VLM-based automatic reward estimation to assess task success without hand-crafted evaluation functions, and (iii) two-stage online reinforcement learning to continuously interact with and learn from GUI environments. Experiments on two advanced GUI Agents (UI-TARS and Aguvis) demonstrate that ZeroGUI significantly boosts performance across OSWorld and AndroidLab environments. The code is available at https://github.com/OpenGVLab/ZeroGUI.

MAPLE: Encoding Dexterous Robotic Manipulation Priors Learned From Egocentric Videos

Large-scale egocentric video datasets capture diverse human activities across a wide range of scenarios, offering rich and detailed insights into how humans interact with objects, especially those that require fine-grained dexterous control. Such complex, dexterous skills with precise controls are crucial for many robotic manipulation tasks, yet are often insufficiently addressed by traditional data-driven approaches to robotic manipulation. To address this gap, we leverage manipulation priors learned from large-scale egocentric video datasets to improve policy learning for dexterous robotic manipulation tasks. We present MAPLE, a novel method for dexterous robotic manipulation that exploits rich manipulation priors to enable efficient policy learning and better performance on diverse, complex manipulation tasks. Specifically, we predict hand-object contact points and detailed hand poses at the moment of hand-object contact and use the learned features to train policies for downstream manipulation tasks. Experimental results demonstrate the effectiveness of MAPLE across existing simulation benchmarks, as well as a newly designed set of challenging simulation tasks, which require fine-grained object control and complex dexterous skills. The benefits of MAPLE are further highlighted in real-world experiments using a dexterous robotic hand, whereas simultaneous evaluation across both simulation and real-world experiments has remained underexplored in prior work.

ORCA: An Open-Source, Reliable, Cost-Effective, Anthropomorphic Robotic Hand for Uninterrupted Dexterous Task Learning

General-purpose robots should possess humanlike dexterity and agility to perform tasks with the same versatility as us. A human-like form factor further enables the use of vast datasets of human-hand interactions. However, the primary bottleneck in dexterous manipulation lies not only in software but arguably even more in hardware. Robotic hands that approach human capabilities are often prohibitively expensive, bulky, or require enterprise-level maintenance, limiting their accessibility for broader research and practical applications. What if the research community could get started with reliable dexterous hands within a day? We present the open-source ORCA hand, a reliable and anthropomorphic 17-DoF tendon-driven robotic hand with integrated tactile sensors, fully assembled in less than eight hours and built for a material cost below 2,000 CHF. We showcase ORCA's key design features such as popping joints, auto-calibration, and tensioning systems that significantly reduce complexity while increasing reliability, accuracy, and robustness. We benchmark the ORCA hand across a variety of tasks, ranging from teleoperation and imitation learning to zero-shot sim-to-real reinforcement learning. Furthermore, we demonstrate its durability, withstanding more than 10,000 continuous operation cycles - equivalent to approximately 20 hours - without hardware failure, the only constraint being the duration of the experiment itself. All design files, source code, and documentation will be available at https://www.orcahand.com/.

SongEditor: Adapting Zero-Shot Song Generation Language Model as a Multi-Task Editor

The emergence of novel generative modeling paradigms, particularly audio language models, has significantly advanced the field of song generation. Although state-of-the-art models are capable of synthesizing both vocals and accompaniment tracks up to several minutes long concurrently, research about partial adjustments or editing of existing songs is still underexplored, which allows for more flexible and effective production. In this paper, we present SongEditor, the first song editing paradigm that introduces the editing capabilities into language-modeling song generation approaches, facilitating both segment-wise and track-wise modifications. SongEditor offers the flexibility to adjust lyrics, vocals, and accompaniments, as well as synthesizing songs from scratch. The core components of SongEditor include a music tokenizer, an autoregressive language model, and a diffusion generator, enabling generating an entire section, masked lyrics, or even separated vocals and background music. Extensive experiments demonstrate that the proposed SongEditor achieves exceptional performance in end-to-end song editing, as evidenced by both objective and subjective metrics. Audio samples are available in \url{https://cypress-yang.github.io/SongEditor_demo/}.

PVC: Progressive Visual Token Compression for Unified Image and Video Processing in Large Vision-Language Models

Large Vision-Language Models (VLMs) have been extended to understand both images and videos. Visual token compression is leveraged to reduce the considerable token length of visual inputs. To meet the needs of different tasks, existing high-performance models usually process images and videos separately with different token compression strategies, limiting the capabilities of combining images and videos. To this end, we extend each image into a "static" video and introduce a unified token compression strategy called Progressive Visual Token Compression (PVC), where the tokens of each frame are progressively encoded and adaptively compressed to supplement the information not extracted from previous frames. Video tokens are efficiently compressed with exploiting the inherent temporal redundancy. Images are repeated as static videos, and the spatial details can be gradually supplemented in multiple frames. PVC unifies the token compressing of images and videos. With a limited number of tokens per frame (64 tokens by default), spatial details and temporal changes can still be preserved. Experiments show that our model achieves state-of-the-art performance across various video understanding benchmarks, including long video tasks and fine-grained short video tasks. Meanwhile, our unified token compression strategy incurs no performance loss on image benchmarks, particularly in detail-sensitive tasks.

VQ-ACE: Efficient Policy Search for Dexterous Robotic Manipulation via Action Chunking Embedding

Dexterous robotic manipulation remains a significant challenge due to the high dimensionality and complexity of hand movements required for tasks like in-hand manipulation and object grasping. This paper addresses this issue by introducing Vector Quantized Action Chunking Embedding (VQ-ACE), a novel framework that compresses human hand motion into a quantized latent space, significantly reducing the action space's dimensionality while preserving key motion characteristics. By integrating VQ-ACE with both Model Predictive Control (MPC) and Reinforcement Learning (RL), we enable more efficient exploration and policy learning in dexterous manipulation tasks using a biomimetic robotic hand. Our results show that latent space sampling with MPC produces more human-like behavior in tasks such as Ball Rolling and Object Picking, leading to higher task success rates and reduced control costs. For RL, action chunking accelerates learning and improves exploration, demonstrated through faster convergence in tasks like cube stacking and in-hand cube reorientation. These findings suggest that VQ-ACE offers a scalable and effective solution for robotic manipulation tasks involving complex, high-dimensional state spaces, contributing to more natural and adaptable robotic systems.

Vision Model Pre-training on Interleaved Image-Text Data via Latent Compression Learning

Recently, vision model pre-training has evolved from relying on manually annotated datasets to leveraging large-scale, web-crawled image-text data. Despite these advances, there is no pre-training method that effectively exploits the interleaved image-text data, which is very prevalent on the Internet. Inspired by the recent success of compression learning in natural language processing, we propose a novel vision model pre-training method called Latent Compression Learning (LCL) for interleaved image-text data. This method performs latent compression learning by maximizing the mutual information between the inputs and outputs of a causal attention model. The training objective can be decomposed into two basic tasks: 1) contrastive learning between visual representation and preceding context, and 2) generating subsequent text based on visual representation. Our experiments demonstrate that our method not only matches the performance of CLIP on paired pre-training datasets (e.g., LAION), but can also leverage interleaved pre-training data (e.g., MMC4) to learn robust visual representation from scratch, showcasing the potential of vision model pre-training with interleaved image-text data. Code is released at https://github.com/OpenGVLab/LCL.

CRAG -- Comprehensive RAG Benchmark

Retrieval-Augmented Generation (RAG) has recently emerged as a promising solution to alleviate Large Language Model (LLM)'s deficiency in lack of knowledge. Existing RAG datasets, however, do not adequately represent the diverse and dynamic nature of real-world Question Answering (QA) tasks. To bridge this gap, we introduce the Comprehensive RAG Benchmark (CRAG), a factual question answering benchmark of 4,409 question-answer pairs and mock APIs to simulate web and Knowledge Graph (KG) search. CRAG is designed to encapsulate a diverse array of questions across five domains and eight question categories, reflecting varied entity popularity from popular to long-tail, and temporal dynamisms ranging from years to seconds. Our evaluation on this benchmark highlights the gap to fully trustworthy QA. Whereas most advanced LLMs achieve <=34% accuracy on CRAG, adding RAG in a straightforward manner improves the accuracy only to 44%. State-of-the-art industry RAG solutions only answer 63% questions without any hallucination. CRAG also reveals much lower accuracy in answering questions regarding facts with higher dynamism, lower popularity, or higher complexity, suggesting future research directions. The CRAG benchmark laid the groundwork for a KDD Cup 2024 challenge, attracting thousands of participants and submissions within the first 50 days of the competition. We commit to maintaining CRAG to serve research communities in advancing RAG solutions and general QA solutions.