arXiv subjects

Tianyun Zhao

Publications and source records attributed to Tianyun Zhao.

9 recordsLinked to original sources

arXiv2026

Thousand-GPU Large-Scale Training and Optimization Recipe for AI-Native Cloud Embodied Intelligence Infrastructure

Embodied intelligence is a key step towards Artificial General Intelligence (AGI), yet its development faces multiple challenges including data, frameworks, infrastructure, and evaluation systems. To address these issues, we have, for the first time in the industry, launched a cloud-based, thousand-GPU distributed training platform for embodied intelligence, built upon the widely adopted LeRobot framework, and have systematically overcome bottlenecks across the entire pipeline. At the data layer, we have restructured the data pipeline to optimize the flow of embodied training data. In terms of training, for the GR00T-N1.5 model, utilizing thousand-GPU clusters and data at the scale of hundreds of millions, the single-round training time has been reduced from 15 hours to just 22 minutes, achieving a 40-fold speedup. At the model layer, by combining variable-length FlashAttention and Data Packing, we have moved from sample redundancy to sequence integration, resulting in a 188% speed increase; {\pi}-0.5 attention optimization has accelerated training by 165%; and FP8 quantization has delivered a 140% speedup. On the infrastructure side, relying on high-performance storage, a 3.2T RDMA network, and a Ray-driven elastic AI data lake, we have achieved deep synergy among data, storage, communication, and computation. We have also built an end-to-end evaluation system, creating a closed loop from training to simulation to assessment. This framework has already been fully validated on thousand-GPU clusters, laying a crucial technical foundation for the development and application of next-generation autonomous intelligent robots, and is expected to accelerate the arrival of the era of human-machine integration.

cs.RO↗

arXiv2026

RL-VLA$^3$: A Flexible and Asynchronous Reinforcement Learning Framework for VLA Training

Reinforcement learning (RL) has emerged as a critical paradigm for post-training Vision-Language-Action (VLA) models, enabling embodied agents to adapt and improve through environmental interaction. However, existing RL frameworks for VLAs inherit synchronous design principles from traditional LLM training, treating entire rollouts as indivisible units and alternating strictly between data collection and policy optimization. This fundamentally mismatches the unique characteristics of VLA training, as physical simulators introduce highly variable, resource-intensive latencies. To address this, we introduce RL-VLA$^3$, a fully asynchronous distributed RL framework that enables fine-grained asynchronous interaction between simulation, inference, and training components through dynamic batching schedulers and flexible environment sharding strategies. Extensive experiments across diverse simulation backends, VLA architectures, and RL algorithms demonstrate that RL-VLA$^3$ achieves throughput improvements of up to 85.2\% over synchronous baselines while maintaining identical sample efficiency, with scalability validated from 8 to 256 GPUs. To our knowledge, RL-VLA$^3$ is the first fully asynchronous RL training framework tailored specifically for the system-level challenges of VLA training.

cs.AI↗

arXiv2025

Seed-Prover: Deep and Broad Reasoning for Automated Theorem Proving

LLMs have demonstrated strong mathematical reasoning abilities by leveraging reinforcement learning with long chain-of-thought, yet they continue to struggle with theorem proving due to the lack of clear supervision signals when solely using natural language. Dedicated domain-specific languages like Lean provide clear supervision via formal verification of proofs, enabling effective training through reinforcement learning. In this work, we propose \textbf{Seed-Prover}, a lemma-style whole-proof reasoning model. Seed-Prover can iteratively refine its proof based on Lean feedback, proved lemmas, and self-summarization. To solve IMO-level contest problems, we design three test-time inference strategies that enable both deep and broad reasoning. Seed-Prover proves $78.1\%$ of formalized past IMO problems, saturates MiniF2F, and achieves over 50\% on PutnamBench, outperforming the previous state-of-the-art by a large margin. To address the lack of geometry support in Lean, we introduce a geometry reasoning engine \textbf{Seed-Geometry}, which outperforms previous formal geometry engines. We use these two systems to participate in IMO 2025 and fully prove 5 out of 6 problems. This work represents a significant advancement in automated mathematical reasoning, demonstrating the effectiveness of formal verification with long chain-of-thought reasoning.

cs.AI↗

arXiv2024

Local-peak scale-invariant feature transform for fast and random image stitching

Image stitching aims to construct a wide field of view with high spatial resolution, which cannot be achieved in a single exposure. Typically, conventional image stitching techniques, other than deep learning, require complex computation and thus computational pricy, especially for stitching large raw images. In this study, inspired by the multiscale feature of fluid turbulence, we developed a fast feature point detection algorithm named local-peak scale-invariant feature transform (LP-SIFT), based on the multiscale local peaks and scale-invariant feature transform method. By combining LP-SIFT and RANSAC in image stitching, the stitching speed can be improved by orders, compared with the original SIFT method. Nine large images (over 2600*1600 pixels), arranged randomly without prior knowledge, can be stitched within 158.94 s. The algorithm is highly practical for applications requiring a wide field of view in diverse application scenes, e.g., terrain mapping, biological analysis, and even criminal investigation.

cs.CV↗

arXiv2023

Flow cytometry with anti-diffraction light sheet (ADLS) by spatial light modulation

Flow cytometry is a widespread and powerful technique, whose resolution is determined by its capacity to accurately distinguish fluorescently positive populations from negative ones. However, most informative results are discarded while performing the measurements of conventional flow cytometry, e.g., the cell size, shape, morphology, and distribution or location of labeled exosomes within the unpurified biological samples. We, herein, propose a novel approach using an anti-diffraction light sheet with anisotroic feature to excite fluorescent tags. Constituted by an anti-diffraction Bessel-Gaussian beam array, the light sheet is 12 $\mu$m wide, 12 $\mu$m high, with a thickness of $~ 0.8 \mu$m. The intensity profile of the excited fluorescent signal can, therefore, reflect the size and allow samples in the range from O(100 nm) to 10 $\mu$m (e.g., blood cells) to be transported via hydrodynamic focusing in a microfluidic chip. The sampling rate is 500 kHz provides a capability of high throughput without sacrificing the spatial resolution. Consequently, the proposed anti-diffraction light-sheet flow cytometry (ADLSFC) can obtain more informative results than the conventional methodologies, and is able to provide multiple characteristics (e.g., the size and distribution of fluorescent signal) helping to distinguish the target samples from the complex backgrounds.

physics.optics↗

arXiv2023

PTA-Det: Point Transformer Associating Point cloud and Image for 3D Object Detection

In autonomous driving, 3D object detection based on multi-modal data has become an indispensable approach when facing complex environments around the vehicle. During multi-modal detection, LiDAR and camera are simultaneously applied for capturing and modeling. However, due to the intrinsic discrepancies between the LiDAR point and camera image, the fusion of the data for object detection encounters a series of problems. Most multi-modal detection methods perform even worse than LiDAR-only methods. In this investigation, we propose a method named PTA-Det to improve the performance of multi-modal detection. Accompanied by PTA-Det, a Pseudo Point Cloud Generation Network is proposed, which can convert image information including texture and semantic features by pseudo points. Thereafter, through a transformer-based Point Fusion Transition (PFT) module, the features of LiDAR points and pseudo points from image can be deeply fused under a unified point-based representation. The combination of these modules can conquer the major obstacle in feature fusion across modalities and realizes a complementary and discriminative representation for proposal generation. Extensive experiments on the KITTI dataset show the PTA-Det achieves a competitive result and support its effectiveness.

cs.CV↗

arXiv2023

Quad-cascade picture of electrokinetic turbulence

Turbulence, ubiquitous in nature and across various systems, exhibits chaotic and intermittent fluctuations in space and time, defying precise prediction. For nearly a century, extensive efforts have been made to uncover the underlying universality and invariant laws from the immense disorder and chaotic nature of turbulence. While the celebrated Kolmogorov -5/3 law stands as a robust cornerstone, it falls short in capturing the diverse scaling behavior exhibited in turbulence influenced by external volume forces, like thermal convection and electrokinetic flows. This study proposes a general framework that couples the fluxes of kinetic energy and scalar variance, culminating in the formulation of a universal conservation law. This framework offers a comprehensive quad-cascade depiction of turbulence, enabling predictions that beyond the limitations of existing models. We illustrate this framework with microfluidic experiments on electrokinetic turbulence, wherein power spectra of concentration and velocity fluctuations exhibit the predicted scaling behaviors, providing remarkable agreement with theory. These findings not only deepen our understanding of the complete cascade process in turbulence driven by external volume forces but also hold promise for insights into other turbulent systems.

physics.flu-dyn↗

arXiv2022

Mixing and flow transition in an optimized electrokinetic turbulent micromixer

Micromixer is a key element in lab on a chip for broad applications in the analysis and measurement of chemistry and engineering. Previous investigations reported electrokinetic (EK) turbulence could be realized in a Y-type micromixer with a cross-sectional dimension of 100 ${\mu}$m order. Although the ultrafast turbulent mixing can be generated at a bulk flow Reynolds number of O(1), the micromixer has not been optimized. In this investigation, we systematically investigated the influence of electric field intensity, AC frequency, electric conductivity ratio, and channel width at the entrance on the mixing effect and transition electric Rayleigh number in the "Y" type electrokinetic micromixer. It is found the optimal mixing is realized in a 350 ${\mu}$m wide micromixer, under 100 kHz and 1.14*10^5 V/m AC electric field, with an electric conductivity ratio of 1:3000. Under the conditions, a maximum degree of mixedness of 0.93 can be achieved at 84 ${\mu}$m from the entrance and 100 ms. A further investigation of the critical electric field and the critical electric Rayleigh number indicates the most unstable condition of EK flow instability is inconsistent with that of the optimal mixing in EK turbulence. To predict the evolution of EK flow under high $Ra_{e}$, it is necessary to apply a computational turbulence model, instead of linear instability analysis.

physics.flu-dyn↗

arXiv2021

Super-resolution reconstruction of cytoskeleton image based on A-net deep learning network

To date, live-cell imaging at the nanometer scale remains challenging. Even though super-resolution microscopy methods have enabled visualization of subcellular structures below the optical resolution limit, the spatial resolution is still far from enough for the structural reconstruction of biomolecules in vivo (i.e. ~24 nm thickness of microtubule fiber). In this study, we proposed an A-net network and showed that the resolution of cytoskeleton images captured by a confocal microscope can be significantly improved by combining the A-net deep learning network with the DWDC algorithm based on degradation model. Utilizing the DWDC algorithm to construct new datasets and taking advantage of A-net neural network's features (i.e., considerably fewer layers), we successfully removed the noise and flocculent structures, which originally interfere with the cellular structure in the raw image, and improved the spatial resolution by 10 times using relatively small dataset. We, therefore, conclude that the proposed algorithm that combines A-net neural network with the DWDC method is a suitable and universal approach for exacting structural details of biomolecules, cells and organs from low-resolution images.

eess.IV↗