arXiv subjects

Bowen Ye

Publications and source records attributed to Bowen Ye.

29 records · Page 2Linked to original sources

arXiv2026

DAP: A Discrete-token Autoregressive Planner for Autonomous Driving

Gaining sustainable performance improvement with scaling data and model budget remains a pivotal yet unresolved challenge in autonomous driving. While autoregressive models exhibited promising data-scaling efficiency in planning tasks, predicting ego trajectories alone suffers sparse supervision and weakly constrains how scene evolution should shape ego motion. Therefore, we introduce DAP, a discrete-token autoregressive planner that jointly forecasts BEV semantics and ego trajectories, thereby enforcing comprehensive representation learning and allowing predicted dynamics to directly condition ego motion. In addition, we incorporate a reinforcement-learning-based fine-tuning, which preserves supervised behavior cloning priors while injecting reward-guided improvements. Despite a compact 160M parameter budget, DAP achieves state-of-the-art performance on open-loop metrics and delivers competitive closed-loop results on the NAVSIM benchmark. Overall, the fully discrete-token autoregressive formulation operating on both rasterized BEV and ego actions provides a compact yet scalable planning paradigm for autonomous driving.

cs.AI↗

arXiv2026

Bridging Perception and Planning: Towards End-to-End Planning for Signal Temporal Logic Tasks

We investigate the task and motion planning problem for Signal Temporal Logic (STL) specifications in robotics. Existing STL methods rely on pre-defined maps or mobility representations, which are ineffective in unstructured real-world environments. We propose the \emph{Structured-MoE STL Planner} (\textbf{S-MSP}), a differentiable framework that maps synchronized multi-view camera observations and an STL specification directly to a feasible trajectory. S-MSP integrates STL constraints within a unified pipeline, trained with a composite loss that combines trajectory reconstruction and STL robustness. A \emph{structure-aware} Mixture-of-Experts (MoE) model enables horizon-aware specialization by projecting sub-tasks into temporally anchored embeddings. We evaluate S-MSP using a high-fidelity simulation of factory-logistics scenarios with temporally constrained tasks. Experiments show that S-MSP outperforms single-expert baselines in STL satisfaction and trajectory feasibility. A rule-based \emph{safety filter} at inference improves physical executability without compromising logical correctness, showcasing the practicality of the approach.

cs.RO↗

arXiv2026

MiMo-V2-Flash Technical Report

We present MiMo-V2-Flash, a Mixture-of-Experts (MoE) model with 309B total parameters and 15B active parameters, designed for fast, strong reasoning and agentic capabilities. MiMo-V2-Flash adopts a hybrid attention architecture that interleaves Sliding Window Attention (SWA) with global attention, with a 128-token sliding window under a 5:1 hybrid ratio. The model is pre-trained on 27 trillion tokens with Multi-Token Prediction (MTP), employing a native 32k context length and subsequently extended to 256k. To efficiently scale post-training compute, MiMo-V2-Flash introduces a novel Multi-Teacher On-Policy Distillation (MOPD) paradigm. In this framework, domain-specialized teachers (e.g., trained via large-scale reinforcement learning) provide dense and token-level reward, enabling the student model to perfectly master teacher expertise. MiMo-V2-Flash rivals top-tier open-weight models such as DeepSeek-V3.2 and Kimi-K2, despite using only 1/2 and 1/3 of their total parameters, respectively. During inference, by repurposing MTP as a draft model for speculative decoding, MiMo-V2-Flash achieves up to 3.6 acceptance length and 2.6x decoding speedup with three MTP layers. We open-source both the model weights and the three-layer MTP weights to foster open research and community collaboration.

cs.CL↗

arXiv2025

MiMo-Audio: Audio Language Models are Few-Shot Learners

Existing audio language models typically rely on task-specific fine-tuning to accomplish particular audio tasks. In contrast, humans are able to generalize to new audio tasks with only a few examples or simple instructions. GPT-3 has shown that scaling next-token prediction pretraining enables strong generalization capabilities in text, and we believe this paradigm is equally applicable to the audio domain. By scaling MiMo-Audio's pretraining data to over one hundred million of hours, we observe the emergence of few-shot learning capabilities across a diverse set of audio tasks. We develop a systematic evaluation of these capabilities and find that MiMo-Audio-7B-Base achieves SOTA performance on both speech intelligence and audio understanding benchmarks among open-source models. Beyond standard metrics, MiMo-Audio-7B-Base generalizes to tasks absent from its training data, such as voice conversion, style transfer, and speech editing. MiMo-Audio-7B-Base also demonstrates powerful speech continuation capabilities, capable of generating highly realistic talk shows, recitations, livestreaming and debates. At the post-training stage, we curate a diverse instruction-tuning corpus and introduce thinking mechanisms into both audio understanding and generation. MiMo-Audio-7B-Instruct achieves open-source SOTA on audio understanding benchmarks (MMSU, MMAU, MMAR, MMAU-Pro), spoken dialogue benchmarks (Big Bench Audio, MultiChallenge Audio) and instruct-TTS evaluations, approaching or surpassing closed-source models. Model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-Audio.

cs.CL↗

arXiv2025

Task and Motion Planning of Dynamic Systems using Hyperproperties for Signal Temporal Logics

We investigate the task and motion planning problem for dynamical systems under signal temporal logic (STL) specifications. Existing works on STL control synthesis mainly focus on generating plans that satisfy properties over a single executed trajectory. In this work, we consider the planning problem for hyperproperties evaluated over a set of possible trajectories, which naturally arise in information-flow control problems. Specifically, we study discrete-time dynamical systems and employ the recently developed temporal logic HyperSTL as the new objective for planning. To solve this problem, we propose a novel recursive counterexample-guided synthesis approach capable of effectively handling HyperSTL specifications with multiple alternating quantifiers. The proposed method is not only applicable to planning but also extends to HyperSTL model checking for discrete-time dynamical systems. Finally, we present case studies on security-preserving planning and ambiguity-free planning to demonstrate the effectiveness of the proposed HyperSTL planning framework.

eess.SY↗

arXiv2025

Bidirectional Prototype-Reward co-Evolution for Test-Time Adaptation of Vision-Language Models

Test-time adaptation (TTA) is crucial in maintaining performance of Vision Language Models (VLMs) when facing distribution shifts, particularly when the source data or target labels are inaccessible. Existing TTA methods predominantly leverage the output probability distribution of CLIP for feature evaluation, resulting in biases under domain shifts, which cause misclassified features due to text priors or incorrect textual associations. To address these issues, we propose \underline{B}idirectional Prototype-Reward co-Evolution (BPRE), a novel VLMs framework with TTA that integrates feature quality assessment with prototype evolution via a synergistic feedback loop. First, the Multi-dimensional Quality-aware Reward Module (MQRM) is designed to evaluate feature quality and guide prototype refinement precisely. The continuous refinement of prototype quality via Prototype-Reward Interactive Evolution (PRIE) enhances the computation more robust. Through this bidirectional interaction, the precision of rewards and prototype evolution mutually reinforce each other, forming a self-evolving feedback cycle. Extensive experiments conducted on 15 diverse recognition datasets demonstrate that our model consistently achieves superior performance compared to other SOTA methods, and advances VLM generalization capabilities through emphasizing comprehensive feature evaluation.

cs.CV↗

arXiv2025

MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining

We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo.

cs.CL↗

arXiv2025

MiMo-VL Technical Report

We open-source MiMo-VL-7B-SFT and MiMo-VL-7B-RL, two powerful vision-language models delivering state-of-the-art performance in both general visual understanding and multimodal reasoning. MiMo-VL-7B-RL outperforms Qwen2.5-VL-7B on 35 out of 40 evaluated tasks, and scores 59.4 on OlympiadBench, surpassing models with up to 78B parameters. For GUI grounding applications, it sets a new standard with 56.1 on OSWorld-G, even outperforming specialized models such as UI-TARS. Our training combines four-stage pre-training (2.4 trillion tokens) with Mixed On-policy Reinforcement Learning (MORL) integrating diverse reward signals. We identify the importance of incorporating high-quality reasoning data with long Chain-of-Thought into pre-training stages, and the benefits of mixed RL despite challenges in simultaneous multi-domain optimization. We also contribute a comprehensive evaluation suite covering 50+ tasks to promote reproducibility and advance the field. The model checkpoints and full evaluation suite are available at https://github.com/XiaomiMiMo/MiMo-VL.

cs.CL↗

arXiv2025

HATA: Trainable and Hardware-Efficient Hash-Aware Top-k Attention for Scalable Large Model Inference

Large Language Models (LLMs) have emerged as a pivotal research area, yet the attention module remains a critical bottleneck in LLM inference, even with techniques like KVCache to mitigate redundant computations. While various top-$k$ attention mechanisms have been proposed to accelerate LLM inference by exploiting the inherent sparsity of attention, they often struggled to strike a balance between efficiency and accuracy. In this paper, we introduce HATA (Hash-Aware Top-$k$ Attention), a novel approach that systematically integrates low-overhead learning-to-hash techniques into the Top-$k$ attention process. Different from the existing top-k attention methods which are devoted to seeking an absolute estimation of qk score, typically with a great cost, HATA maps queries and keys into binary hash codes, and acquires the relative qk score order with a quite low cost, which is sufficient for realizing top-k attention. Extensive experiments demonstrate that HATA achieves up to 7.2$\times$ speedup compared to vanilla full attention while maintaining model accuracy. In addition, HATA outperforms the state-of-the-art top-$k$ attention methods in both accuracy and efficiency across multiple mainstream LLM models and diverse tasks. HATA is open source at https://github.com/gpzlx1/HATA.

cs.LG↗

arXiv2024

Magical Experience with Full-body Action

This paper presents a system that generates a magical experience with full-body motion. The system consists of a locomotion interface and a spatial immersive display. A virtual experience system named the Magical Experience Generator was developed, equipped with a Magical Experience Controller. This system provides a physical movement experience along with magical-like interactions in a virtual space. We developed content inspired by the Japanese story "The Man Who Made Flowers Bloom" using Unity as the system's environment. The locomotion interface records the participant's walking trajectory and hand movements, representing their actions in the virtual space.

cs.HC↗

arXiv2024

Prioritize Team Actions: Multi-Agent Temporal Logic Task Planning with Ordering Constraints

In this paper, we investigate the problem of linear temporal logic (LTL) path planning for multi-agent systems, introducing the new concept of \emph{ordering constraints}. Specifically, we consider a generic objective function that is defined for the path of each individual agent. The primary objective is to find a global plan for the team of agents, ensuring they collectively meet the specified LTL requirements. Simultaneously, we aim to maintain a pre-determined order in the values of the objective function for each agent, which we refer to as the ordering constraints. This new requirement stems from scenarios like security-aware planning, where relative orders outweigh absolute values in importance. We present an efficient algorithm to solve this problem, supported by proofs of correctness that demonstrate the optimality of our solution. Additionally, we provide a case study in security-aware path planning to illustrate the practicality and effectiveness of our proposed approach.

eess.SY↗