Yuchen Song

MiniMax, Aili Chen, Aonian Li et al.

We introduce the MiniMax-M2 series, a family of Mixture-of-Experts language models built around the principle that mini activations can unleash maximum real-world intelligence. The flagship M2 contains 229.9B total parameters with only 9.8B activated per token. Designed end-to-end for agentic deployment, the M2 series rests on three components: (i) agent-driven data pipelines producing large-scale, verifiable trajectories across agentic coding and agentic cowork, each grounded in an executable workspace and an artifact-aligned reward; (ii) Forge, a scalable agent-native RL system that adapts to long-horizon agent trajectories, paired with windowed-FIFO scheduling, prefix-tree merging, inference optimization, and a clean training-inference-agent decoupling that supports both white-box and black-box agents; (iii) the latest M2.7 checkpoint takes an early step toward self-evolution -- autonomously debugging training runs and modifying its own scaffold. Across M2 through M2.7, this combination translates a mini-activation footprint into frontier-tier performance on agentic coding, deep search, office-task, and reasoning benchmarks.

Yao Zhang, Yuchen Song, Xiao Luo et al.

Recent advances in large language models (LLMs) have demonstrated strong capabilities in code generation and text synthesis, yet their potential for symbolic physical reasoning in domain-specific scientific problems remains underexplored. We present a mathematical reasoning enhanced generative AI approach for optical communication formula derivation, focusing on the fiber nonlinear interference modelling. By guiding an LLM with structured prompts, we successfully reconstructed the known closed-form ISRS GN expressions and further derived a novel approximation tailored for multi-span C and C+L band transmissions. Numerical validations show that the LLM-derived model produces central-channel GSNRs nearly identical to baseline models, with mean absolute error across all channels and spans below 0.109 dB, demonstrating both physical consistency and practical accuracy.

Andong Chen, Wenxin Zhu, Qiuyu Ding et al.

Chain-of-Thought reasoning has driven large language models to extend from thinking with text to thinking with images and videos. However, different modalities still have clear limitations: static images struggle to represent temporal structure, while videos introduce substantial redundancy and computational cost. In this work, we propose Thinking with Comics, a visual reasoning paradigm that uses comics as a high information-density medium positioned between images and videos. Comics preserve temporal structure, embedded text, and narrative coherence while requiring significantly lower reasoning cost. We systematically study two reasoning paths based on comics and evaluate them on a range of reasoning tasks and long-context understanding tasks. Experimental results show that Thinking with Comics outperforms Thinking with Images on multi-step temporal and causal reasoning tasks, while remaining substantially more efficient than Thinking with Video. Further analysis indicates that different comic narrative structures and styles consistently affect performance across tasks, suggesting that comics serve as an effective intermediate visual representation for improving multimodal reasoning.

Yuchen Song, Andong Chen, Wenxin Zhu et al.

Cultural awareness capabilities has emerged as a critical capability for Multimodal Large Language Models (MLLMs). However, current benchmarks lack progressed difficulty in their task design and are deficient in cross-lingual tasks. Moreover, current benchmarks often use real-world images. Each real-world image typically contains one culture, making these benchmarks relatively easy for MLLMs. Based on this, we propose C$^3$B ($\textbf{C}$omics $\textbf{C}$ross-$\textbf{C}$ultural $\textbf{B}$enchmark), a novel multicultural, multitask and multilingual cultural awareness capabilities benchmark. C$^3$B comprises over 2000 images and over 18000 QA pairs, constructed on three tasks with progressed difficulties, from basic visual recognition to higher-level cultural conflict understanding, and finally to cultural content generation. We conducted evaluations on 11 open-source MLLMs, revealing a significant performance gap between MLLMs and human performance. The gap demonstrates that C$^3$B poses substantial challenges for current MLLMs, encouraging future research to advance the cultural awareness capabilities of MLLMs.

Shouju Wang, Yuchen Song, Sheng'en Li et al.

Graph anomaly detection (GAD) has become an increasingly important task across various domains. With the rapid development of graph neural networks (GNNs), GAD methods have achieved significant performance improvements. However, fairness considerations in GAD remain largely underexplored. Indeed, GNN-based GAD models can inherit and amplify biases present in training data, potentially leading to unfair outcomes. While existing efforts have focused on developing fair GNNs, most approaches target node classification tasks, where models often rely on simple layer architectures rather than autoencoder-based structures, which are the most widely used architecturs for anomaly detection. To address fairness in autoencoder-based GAD models, we propose \textbf{D}is\textbf{E}ntangled \textbf{C}ounterfactual \textbf{A}dversarial \textbf{F}air (DECAF)-GAD, a framework that alleviates bias while preserving GAD performance. Specifically, we introduce a structural causal model (SCM) to disentangle sensitive attributes from learned representations. Based on this causal framework, we formulate a specialized autoencoder architecture along with a fairness-guided loss function. Through extensive experiments on both synthetic and real-world datasets, we demonstrate that DECAF-GAD not only achieves competitive anomaly detection performance but also significantly enhances fairness metrics compared to baseline GAD methods. Our code is available at https://github.com/Tlhey/decaf_code.

Ri-Zhao Qiu, Yafei Hu, Yuchen Song et al.

An open problem in mobile manipulation is how to represent objects and scenes in a unified manner so that robots can use both for navigation and manipulation. The latter requires capturing intricate geometry while understanding fine-grained semantics, whereas the former involves capturing the complexity inherent at an expansive physical scale. In this work, we present GeFF (Generalizable Feature Fields), a scene-level generalizable neural feature field that acts as a unified representation for both navigation and manipulation that performs in real-time. To do so, we treat generative novel view synthesis as a pre-training task, and then align the resulting rich scene priors with natural language via CLIP feature distillation. We demonstrate the effectiveness of this approach by deploying GeFF on a quadrupedal robot equipped with a manipulator. We quantitatively evaluate GeFF's ability for open-vocabulary object-/part-level manipulation and show that GeFF outperforms point-based baselines in runtime and storage-accuracy trade-offs, with qualitative examples of semantics-aware navigation and articulated object manipulation.

Andong Chen, Yuchen Song, Wenxin Zhu et al.

The o1-Like LLMs are transforming AI by simulating human cognitive processes, but their performance in multilingual machine translation (MMT) remains underexplored. This study examines: (1) how o1-Like LLMs perform in MMT tasks and (2) what factors influence their translation quality. We evaluate multiple o1-Like LLMs and compare them with traditional models like ChatGPT and GPT-4o. Results show that o1-Like LLMs establish new multilingual translation benchmarks, with DeepSeek-R1 surpassing GPT-4o in contextless tasks. They demonstrate strengths in historical and cultural translation but exhibit a tendency for rambling issues in Chinese-centric outputs. Further analysis reveals three key insights: (1) High inference costs and slower processing speeds make complex translation tasks more resource-intensive. (2) Translation quality improves with model size, enhancing commonsense reasoning and cultural translation. (3) The temperature parameter significantly impacts output quality-lower temperatures yield more stable and accurate translations, while higher temperatures reduce coherence and precision.

Ri-Zhao Qiu, Yuchen Song, Xuanbin Peng et al.

'In-the-wild' mobile manipulation aims to deploy robots in diverse real-world environments, which requires the robot to (1) have skills that generalize across object configurations; (2) be capable of long-horizon task execution in diverse environments; and (3) perform complex manipulation beyond pick-and-place. Quadruped robots with manipulators hold promise for extending the workspace and enabling robust locomotion, but existing results do not investigate such a capability. This paper proposes WildLMa with three components to address these issues: (1) adaptation of learned low-level controller for VR-enabled whole-body teleoperation and traversability; (2) WildLMa-Skill -- a library of generalizable visuomotor skills acquired via imitation learning or heuristics and (3) WildLMa-Planner -- an interface of learned skills that allow LLM planners to coordinate skills for long-horizon tasks. We demonstrate the importance of high-quality training data by achieving higher grasping success rate over existing RL baselines using only tens of demonstrations. WildLMa exploits CLIP for language-conditioned imitation learning that empirically generalizes to objects unseen in training demonstrations. Besides extensive quantitative evaluation, we qualitatively demonstrate practical robot applications, such as cleaning up trash in university hallways or outdoor terrains, operating articulated objects, and rearranging items on a bookshelf.

Andong Chen, Yuchen Song, Kehai Chen et al.

Visual information has been introduced for enhancing machine translation (MT), and its effectiveness heavily relies on the availability of large amounts of bilingual parallel sentence pairs with manual image annotations. In this paper, we introduce a stable diffusion-based imagination network into a multimodal large language model (MLLM) to explicitly generate an image for each source sentence, thereby advancing the multimodel MT. Particularly, we build heuristic human feedback with reinforcement learning to ensure the consistency of the generated image with the source sentence without the supervision of image annotation, which breaks the bottleneck of using visual information in MT. Furthermore, the proposed method enables imaginative visual information to be integrated into large-scale text-only MT in addition to multimodal MT. Experimental results show that our model significantly outperforms existing multimodal MT and text-only MT, especially achieving an average improvement of more than 14 BLEU points on Multi30K multimodal MT benchmarks.

Xueyan Zou, Yuchen Song, Ri-Zhao Qiu et al.

We present 3D Spatial MultiModal Memory (M3), a multimodal memory system designed to retain information about medium-sized static scenes through video sources for visual perception. By integrating 3D Gaussian Splatting techniques with foundation models, M3 builds a multimodal memory capable of rendering feature representations across granularities, encompassing a wide range of knowledge. In our exploration, we identify two key challenges in previous works on feature splatting: (1) computational constraints in storing high-dimensional features for each Gaussian primitive, and (2) misalignment or information loss between distilled features and foundation model features. To address these challenges, we propose M3 with key components of principal scene components and Gaussian memory attention, enabling efficient training and inference. To validate M3, we conduct comprehensive quantitative evaluations of feature similarity and downstream tasks, as well as qualitative visualizations to highlight the pixel trace of Gaussian memory attention. Our approach encompasses a diverse range of foundation models, including vision-language models (VLMs), perception models, and large multimodal and language models (LMMs/LLMs). Furthermore, to demonstrate real-world applicability, we deploy M3's feature field in indoor scenes on a quadruped robot. Notably, we claim that M3 is the first work to address the core compression challenges in 3D feature distillation.

Haorui He, Yuchen Song, Yuancheng Wang et al.

The effectiveness of one-shot voice conversion (VC) decreases in real-world scenarios where reference speeches, which are often sourced from the internet, contain various disturbances like background noise. To address this issue, we introduce Noro, a noise-robust one-shot VC system. Noro features innovative components tailored for VC using noisy reference speeches, including a dual-branch reference encoding module and a noise-agnostic contrastive speaker loss. Experimental results demonstrate that Noro outperforms our baseline system in both clean and noisy scenarios, highlighting its efficacy for real-world applications. Additionally, we investigate the hidden speaker representation capabilities of our baseline system by repurposing its reference encoder as a speaker encoder. The results show that it is competitive with several advanced self-supervised learning models for speaker representation under the SUPERB settings, highlighting the potential for advancing speaker representation learning through one-shot VC tasks.

Wenxin Zhu, Andong Chen, Yuchen Song et al.

With the remarkable success of Multimodal Large Language Models (MLLMs) in perception tasks, enhancing their complex reasoning capabilities has emerged as a critical research focus. Existing models still suffer from challenges such as opaque reasoning paths and insufficient generalization ability. Chain-of-Thought (CoT) reasoning, which has demonstrated significant efficacy in language models by enhancing reasoning transparency and output interpretability, holds promise for improving model reasoning capabilities when extended to the multimodal domain. This paper provides a systematic review centered on "Multimodal Chain-of-Thought" (MCoT). First, it analyzes the background and theoretical motivations for its inception from the perspectives of technical evolution and task demands. Then, it introduces mainstream MCoT methods from three aspects: CoT paradigms, the post-training stage, and the inference stage, while also analyzing their underlying mechanisms. Furthermore, the paper summarizes existing evaluation benchmarks and metrics, and discusses the application scenarios of MCoT. Finally, it analyzes the challenges currently facing MCoT and provides an outlook on its future research directions.

Yefeng Wu, Yuchen Song, Ling Wu et al.

Pneumonia remains a leading cause of morbidity and mortality worldwide, necessitating accurate and efficient automated detection systems. While recent transformer-based detectors like RT-DETR have shown promise in object detection tasks, their application to medical imaging, particularly pneumonia detection in chest X-rays, remains underexplored. This paper presents CGF-DETR, an enhanced real-time detection transformer specifically designed for pneumonia detection. We introduce XFABlock in the backbone to improve multi-scale feature extraction through convolutional attention mechanisms integrated with CSP architecture. To achieve efficient feature aggregation, we propose SPGA module that replaces standard multi-head attention with dynamic gating mechanisms and single-head self-attention. Additionally, GCFC3 is designed for the neck to enhance feature representation through multi-path convolution fusion while maintaining real-time performance via structural re-parameterization. Extensive experiments on the RSNA Pneumonia Detection dataset demonstrate that CGF-DETR achieves 82.2% mAP@0.5, outperforming the baseline RT-DETR-l by 3.7% while maintaining comparable inference speed at 48.1 FPS. Our ablation studies confirm that each proposed module contributes meaningfully to the overall performance improvement, with the complete model achieving 50.4% mAP@[0.5:0.95]

Xi Chen, Yuchen Song, Satoshi Nakamura

We propose a stress-aware speech-to-speech translation (S2ST) system that preserves word-level emphasis by leveraging LLMs for cross-lingual emphasis conversion. Our method translates source-language stress into target-language tags that guide a controllable TTS model. To overcome data scarcity, we developed a pipeline to automatically generate aligned training data and introduce the "LLM-as-Judge" for evaluation. Experiments show our approach substantially outperforms baselines in preserving emphasis while maintaining comparable translation quality, speaker intent, and naturalness. Our work highlights the importance of prosody in translation and provides an effective, data-efficient solution for preserving paralinguistic cues in S2ST.

Yao Zhang, Yuchen Song, Shengnan Li et al.

The rapid development of Generative Artificial Intelligence (GenAI) has catalyzed a transformative technological revolution across all walks of life. As the backbone of wideband communication, optical networks are expecting high-level autonomous operation and zero-touch management to accommodate their expanding network scales and escalating transmission bandwidth. The integration of GenAI is deemed as the pivotal solution for realizing zero-touch optical networks. However, the lifecycle management of optical networks involves a multitude of tasks and necessitates seamless collaboration across multiple layers, which poses significant challenges to the existing single-agent GenAI systems. In this paper, we propose a GenAI-driven hierarchical multi-agent framework designed to streamline multi-task autonomous execution for zero-touch optical networks. We present the architecture, implementation, and applications of this framework. A field-deployed mesh network is utilized to demonstrate three typical scenarios throughout the lifecycle of optical network: quality of transmission estimation in the planning stage, dynamic channel adding/dropping in the operation stage, and system capacity increase in the upgrade stage. The case studies, illustrate the capabilities of multi-agent framework in multi-task allocation, coordination, execution, evaluation, and summarization. This work provides a promising approach for the future development of intelligent, efficient, and collaborative network management solutions, paving the way for more specialized and adaptive zero-touch optical networks.

Dazhou Guo, Jia Ge, Xianghua Ye et al.

Accurate organ at risk (OAR) segmentation is critical to reduce the radiotherapy post-treatment complications. Consensus guidelines recommend a set of more than 40 OARs in the head and neck (H&N) region, however, due to the predictable prohibitive labor-cost of this task, most institutions choose a substantially simplified protocol by delineating a smaller subset of OARs and neglecting the dose distributions associated with other OARs. In this work we propose a novel, automated and highly effective stratified OAR segmentation (SOARS) system using deep learning to precisely delineate a comprehensive set of 42 H&N OARs. SOARS stratifies 42 OARs into anchor, mid-level, and small & hard subcategories, with specifically derived neural network architectures for each category by neural architecture search (NAS) principles. We built SOARS models using 176 training patients in an internal institution and independently evaluated on 1327 external patients across six different institutions. It consistently outperformed other state-of-the-art methods by at least 3-5% in Dice score for each institutional evaluation (up to 36% relative error reduction in other metrics). More importantly, extensive multi-user studies evidently demonstrated that 98% of the SOARS predictions need only very minor or no revisions for direct clinical acceptance (saving 90% radiation oncologists workload), and their segmentation and dosimetric accuracy are within or smaller than the inter-user variation. These findings confirmed the strong clinical applicability of SOARS for the OAR delineation process in H&N cancer radiotherapy workflows, with improved efficiency, comprehensiveness, and quality.