--- title: Image Repository-Atlas Community description: Find the Docker images for diverse software keywords: Community,Image Repository Image Repository Find,Docker,Filter Image Type Training Inference R url: https://www.hiascend.com/en/developer/ascendhub section: (其他) --- # Image Repository-Atlas Community URL: https://www.hiascend.com/en/developer/ascendhub 描述: Find the Docker images for diverse software 关键词: Community,Image Repository Image Repository Find,Docker,Filter Image Type Training Inference R Image Repository Image Repository Find the Docker images for diverse software Filter Image Type Training Inference Recommendation AI Framework CANN Other Relevance Downloads Latest mindieInferenceMindIE is Huawei Atlas's inference acceleration suite. It leverages a highly optimized model library and inference engine to boost performance and usability on Atlas hardware. MindIE supports industry-standard model inference, multi-request scheduling, and features like Continuous Batching, PagedAttention, and FlashDecoding to enable high-performance inference.Updated: 2026/08/0774055 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/af85b724a7e5469ebd7ea13c3439d48f) cannCANNCANN (Compute Architecture for Neural Networks) is a heterogeneous computing architecture launched by Huawei for AI scenarios. It provides comprehensive software stack support for Atlas AI processors, covering operator libraries, graph engines, runtime libraries, and compilation toolchains.Updated: 2026/08/0412689 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/17da20d1c2b6493cb38765adeba85884) mis-teiInferenceThe mis-tei image is built based on Text Embedding Inference and adapted for Ascend NPU cards. It provides service interfaces for embedding, reranker and other capabilities, with full compatibility with open-source Text Embedding Inference. This image supports embedding, reranker, Sequence Classification and other models, and runs under the regular user HwHiAiUser.Updated: 2026/05/2611792 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/07a016975cc341f3a5ae131f2b52399d) mindspeed-llmTrainingMindSpeed-LLM is a distributed training suite for large language models tailored to the Huawei Atlas ecosystem. It delivers end-to-end LLM training solutions for ecosystem partners of Huawei Atlas chips. The suite supports distributed pre-training and distributed instruction fine-tuning, and comes with a full development toolchain encompassing data preprocessing, weight conversion, online inference, baseline evaluation and more core capabilities.Updated: 2026/05/124780 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/e26da9266559438b93354792f25b2f4a) ascend-k8sdevicepluginOtherAtlas Device Plugin is one of the core components of the MindCluster cluster scheduling suite, deployed on compute nodes to provide resource discovery and reporting strategies tailored for Atlas devices.Updated: 2026/08/034162 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/a592da7bd2ab4dffa8864abd4eac5068) npu-exporterOtherNPU-Expoter is a Huawei-developed service component that collects monitoring information and indicators of Huawei NPUs and encapsulates the information and indicators into Prometheus-specific data formats.Updated: 2026/08/032998 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/1b1a8c3cc1ff4710bdb0222514a8a7a3) vc-schedulerOtherAtlas for Volcano is built on the open-source Volcano scheduling plugin mechanism, adding features such as Atlas AI processor (NPU) affinity scheduling and virtual device scheduling. It is deployed on management nodes. Volcano is a scheduler component based on the interconnect topology and processing logic of Atlas AI processors, achieving optimal utilization of Atlas AI processors and maximizing their computing performance.Updated: 2026/08/051744 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/54545fa4ff9f446e914bf44b85efdb61) vc-controller-managerOtherAtlas for Volcano is built on the open-source Volcano scheduling plugin mechanism, adding features such as Atlas AI processor (NPU) affinity scheduling and virtual device scheduling. It is deployed on management nodes. Volcano is a scheduler component based on the interconnect topology and processing logic of Atlas AI processors, achieving optimal utilization of Atlas AI processors and maximizing their computing performance.Updated: 2026/08/041516 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/16f17a3c95d54f9da710a9c51bfceaa3) nodedOtherNodeD is a MindCluster cluster scheduling component deployed on compute nodes. It detects node abnormal states, retrieves CPU, memory, and disk fault information from IPMI, and reports it to ClusterD.Updated: 2026/08/03992 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/cc7e6c0a10834f1888d790174fba4bc5) ascend-operatorOtherAtlas Operator is a MindCluster cluster scheduling component deployed on management nodes. It supports distributed training on Kubernetes using two AI frameworks: MindSpore and PyTorch. The CRD (Custom Resource Definition) defines the AtlasJob task type, allowing users to easily implement distributed training by simply configuring YAML files.Updated: 2026/08/03854 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/a066319600634cf6a1e522856a63a1c5) mindspeed-mmTrainingMindSpeed MM: An Atlas multimodal large model suite for large-scale distributed training, supporting mainstream multimodal large model training in the industry. It aims to provide an end-to-end multimodal training solution for Huawei Atlas chips, including features such as pre-built mainstream models, data engineering, distributed training and acceleration, pre-training, fine-tuning, post-training, and online inference tasks.Updated: 2026/05/22634 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/6857f6fc2cfa4a678710a7075426ee5e) clusterdOtherClusterD is a MindCluster cluster scheduling component deployed on management nodes. It collects and aggregates cluster task, resource, and fault information along with their impact scope, performs statistical analysis from task, chip, and fault dimensions, and uniformly determines fault handling levels and strategies.Updated: 2026/08/03480 [了解详情](https://www.hiascend.com/en/developer/ascendhub/detail/b554929b470747448924bc786b5ab95d) 共 34 条 12条/页 24条/页 48条/页 1 2 3 前往 This site uses cookies to improve your browsing experience. By continuing to browse this site, you agree to our use of cookies.View Details [了解详情](https://www.hiascend.com/en/legal/cookies)