From 31c856f05a2cb1c261eb542a45d2e9b74ae1a6a9 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E5=AE=A6=E6=99=93=E7=8E=B2?= <3174348550@qq.com> Date: Thu, 30 Oct 2025 11:02:57 +0800 Subject: [PATCH] modify error links and anchors --- .../source_en/advanced/third_party/npu_info.md | 2 +- .../advanced/third_party/npu_info.md | 3 ++- .../docs/source_en/guide/deployment.md | 18 +++++++++--------- .../training_template_instruction.md | 4 ++-- .../docs/source_zh_cn/guide/deployment.md | 18 +++++++++--------- docs/mindstudio/docs/source_zh_cn/overview.md | 2 +- .../operation/cpp_api_for_custom_ops.md | 18 +++++++++--------- tutorials/source_en/debug/sdc.md | 2 +- .../model_infer/lite_infer/overview.md | 2 +- .../ms_infer/ms_infer_model_serving_infer.md | 8 ++++---- .../operation/cpp_api_for_custom_ops.md | 18 +++++++++--------- tutorials/source_zh_cn/debug/sdc.md | 2 +- .../ms_infer/ms_infer_model_serving_infer.md | 8 ++++---- 13 files changed, 53 insertions(+), 52 deletions(-) diff --git a/docs/lite/docs/source_en/advanced/third_party/npu_info.md b/docs/lite/docs/source_en/advanced/third_party/npu_info.md index b3ea94c93b..c6bd58e1f5 100644 --- a/docs/lite/docs/source_en/advanced/third_party/npu_info.md +++ b/docs/lite/docs/source_en/advanced/third_party/npu_info.md @@ -28,7 +28,7 @@ For more information about compilation, see [Linux Environment Compilation](http When developers need to integrate the use of Kirin NPU features, it is important to note: - - [Configure the Kirin NPU backend](https://www.mindspore.cn/lite/docs/en/r2.7.1/infer/runtime_cpp.html#configuring-the-npu-backend). + - [Configure the Kirin NPU backend](https://www.mindspore.cn/lite/docs/en/r2.7.1/infer/runtime_cpp.html#configuring-the-kirin-npu-backend). For more information about using Runtime to perform inference, see [Using Runtime to Perform Inference (C++)](https://www.mindspore.cn/lite/docs/en/r2.7.1/infer/runtime_cpp.html). - Compile and execute the binary. If you use dynamic linking, refer to [compile output](https://www.mindspore.cn/lite/docs/en/r2.7.1/build/build.html) when the compile option is `-I arm64` or `-I arm32`. diff --git a/docs/lite/docs/source_zh_cn/advanced/third_party/npu_info.md b/docs/lite/docs/source_zh_cn/advanced/third_party/npu_info.md index ba67c70302..e809809bcf 100644 --- a/docs/lite/docs/source_zh_cn/advanced/third_party/npu_info.md +++ b/docs/lite/docs/source_zh_cn/advanced/third_party/npu_info.md @@ -26,7 +26,8 @@ bash build.sh -I arm64 -j8 - 集成说明 开发者需要集成使用Kirin NPU功能时，需要注意： - - 在代码中[配置Kirin NPU后端](https://www.mindspore.cn/lite/docs/zh-CN/r2.7.1/infer/runtime_cpp.html#配置使用npu后端)，有关使用Runtime执行推理详情见[使用Runtime执行推理（C++）](https://www.mindspore.cn/lite/docs/zh-CN/r2.7.1/infer/runtime_cpp.html)。 + + - 在代码中[配置Kirin NPU后端](https://www.mindspore.cn/lite/docs/zh-CN/r2.7.1/infer/runtime_cpp.html#配置使用kirin-npu后端)，有关使用Runtime执行推理详情见[使用Runtime执行推理（C++）](https://www.mindspore.cn/lite/docs/zh-CN/r2.7.1/infer/runtime_cpp.html)。 - 编译执行可执行程序。如采用动态加载方式，参考[编译输出](https://www.mindspore.cn/lite/docs/zh-CN/r2.7.1/build/build.html)中编译选项为`-I arm64`或`-I arm32`时的内容，配置好环境变量，将会动态加载libhiai.so、libhiai_ir.so、libhiai_ir_build.so、libhiai_hcl_model_runtime.so。例如： ```bash diff --git a/docs/mindformers/docs/source_en/guide/deployment.md b/docs/mindformers/docs/source_en/guide/deployment.md index dfd9a2ba1a..e1af0db504 100644 --- a/docs/mindformers/docs/source_en/guide/deployment.md +++ b/docs/mindformers/docs/source_en/guide/deployment.md @@ -6,7 +6,7 @@ ### Overview -The vLLM-MindSpore plugin is designed with the functional goal of integrating MindSpore large models into vLLM and enabling their servitized deployment: [Introduction to the vLLM-MindSpore Plugin](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/index.html#overview). +The vLLM-MindSpore plugin is designed with the functional goal of integrating MindSpore large models into vLLM and enabling their servitized deployment: [Introduction to the vLLM-MindSpore Plugin](https://www.mindspore.cn/vllm_mindspore/docs/en/master/index.html#overview). The MindSpore Transformers suite aims to build a full-cycle development toolkit for large-scale models, covering pre-training, fine-tuning, evaluation, inference, and deployment. It provides mainstream Transformer-based large language models (LLMs) and multimodal understanding models (MMs) in the industry. @@ -14,8 +14,8 @@ The MindSpore Transformers suite aims to build a full-cycle development toolkit The environment installation steps are divided into two methods: -- [Docker Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/installation/installation.html#docker-installation): Suitable for scenarios where users need quick deployment and use. -- [Source Code Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/installation/installation.html#source-code-installation): Suitable for users who require incremental development of the vLLM-MindSpore plugin. +- [Docker Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/installation/installation.html#docker-installation): Suitable for scenarios where users need quick deployment and use. +- [Source Code Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/installation/installation.html#source-code-installation): Suitable for users who require incremental development of the vLLM-MindSpore plugin. ### Quick Start @@ -32,7 +32,7 @@ export MINDFORMERS_MODEL_CONFIG=/path/to/yaml # Required for non-Mcore models Currently, vLLM MindSpore supports different model backends. The environment variables specified above designate MindSpore Transformers as the integrated model suite. For non-MCore models, it is necessary to configure the model's YAML configuration file. -For more environment variables, please refer to: [Environment Variables](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/user_guide/environment_variables/environment_variables.html). +For more environment variables, please refer to: [Environment Variables](https://www.mindspore.cn/vllm_mindspore/docs/en/master/user_guide/environment_variables/environment_variables.html). After preparing the model and environment variables, you can proceed with inference. @@ -40,15 +40,15 @@ After preparing the model and environment variables, you can proceed with infere vLLM online inference is designed for real-time service scenarios, leveraging dynamic batching and the OpenAI API to deliver high concurrency, high throughput, and low latency, making it suitable for enterprise-level applications. -- Please refer to the single-GPU inference process: [Single-Card Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/tutorials/qwen2.5_7b_singleNPU/qwen2.5_7b_singleNPU.html) -- Please refer to the single-node multi-GPU inference process: [Multi-Card Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/tutorials/qwen2.5_32b_multiNPU/qwen2.5_32b_multiNPU.html) -- Please refer to the multi-node parallel inference process: [Multi-machine Parallel Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/tutorials/deepseek_parallel/deepseek_r1_671b_w8a8_dp4_tp4_ep4.html) +- Please refer to the single-GPU inference process: [Single-Card Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/tutorials/qwen2.5_7b_singleNPU/qwen2.5_7b_singleNPU.html) +- Please refer to the single-node multi-GPU inference process: [Multi-Card Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/tutorials/qwen2.5_32b_multiNPU/qwen2.5_32b_multiNPU.html) +- Please refer to the multi-node parallel inference process: [Multi-machine Parallel Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/tutorials/deepseek_parallel/deepseek_r1_671b_w8a8_dp4_tp4_ep4.html) #### Offline Inference vLLM's offline inference is designed for efficiently processing large-scale batch requests, making it particularly suitable for non-real-time, data-intensive model inference scenarios. -For the offline inference process, please refer to: [Offline Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/quick_start/quick_start.html#offline-inference) +For the offline inference process, please refer to: [Offline Inference](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/quick_start/quick_start.html#offline-inference) ### Mcore Model Adaptation @@ -70,7 +70,7 @@ If configuration modifications are required, please refer to the [Configuration] #### Compatible Versions -For supporting information on each component, please refer to: [Compatible Versions](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/installation/installation.html) +For supporting information on each component, please refer to: [Compatible Versions](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/installation/installation.html) #### Supported Models List diff --git a/docs/mindformers/docs/source_zh_cn/advanced_development/training_template_instruction.md b/docs/mindformers/docs/source_zh_cn/advanced_development/training_template_instruction.md index 207365e45d..b43a219bd1 100644 --- a/docs/mindformers/docs/source_zh_cn/advanced_development/training_template_instruction.md +++ b/docs/mindformers/docs/source_zh_cn/advanced_development/training_template_instruction.md @@ -46,7 +46,7 @@ MindSpore Transformers对于不同训练场景提供了对应的配置模板， ### 数据集配置修改 1. 预训练场景使用Megatron数据集，详情请参考[Megatron数据集](https://www.mindspore.cn/mindformers/docs/zh-CN/r1.7.0/feature/dataset.html#megatron%E6%95%B0%E6%8D%AE%E9%9B%86)。 -2. 微调场景使用HuggingFace数据集，详情请参考[HuggingFace数据集](https://www.mindspore.cn/mindformers/docs/zh-CN/r1.7.0/feature/dataset.html#huggingface%E6%95%B0%E6%8D%AE%E9%9B%86)。 +2. 微调场景使用HuggingFace数据集，详情请参考[HuggingFace数据集](https://www.mindspore.cn/mindformers/docs/zh-CN/r1.7.0/feature/dataset.html#hugging-face%E6%95%B0%E6%8D%AE%E9%9B%86)。 ### 模型配置修改 @@ -59,7 +59,7 @@ MindSpore Transformers对于不同训练场景提供了对应的配置模板， | Qwen2_5 | 2. 生成的模型配置优先以yaml配置为准，未配置参数则取值pretrained_model_dir路径下的config.json中的参数。如若要修改定制模型配置，则只需要在model_config中添加相关配置即可。 -3. 通用配置详情请参考[模型配置](https://www.mindspore.cn/mindformers/docs/zh-CN/r1.7.0/feature/configuration.html#%E6%A8%A1%E5%9E%8B%E9%85%8D%E7%BD%AE)。 +3. 通用配置详情请参考[模型配置](https://www.mindspore.cn/mindformers/docs/zh-CN/r1.7.0/feature/configuration.html#legacy-%E6%A8%A1%E5%9E%8B%E9%85%8D%E7%BD%AE)。 ## 进阶配置修改 diff --git a/docs/mindformers/docs/source_zh_cn/guide/deployment.md b/docs/mindformers/docs/source_zh_cn/guide/deployment.md index aa4c5c5582..179e5aa8e1 100644 --- a/docs/mindformers/docs/source_zh_cn/guide/deployment.md +++ b/docs/mindformers/docs/source_zh_cn/guide/deployment.md @@ -6,7 +6,7 @@ ### 概述 -vLLM-MindSpore插件以将MindSpore大模型接入vLLM，并实现服务化部署为功能目标: [vLLM-MindSpore插件简介](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/index.html#vllm-mindspore%E6%8F%92%E4%BB%B6%E7%AE%80%E4%BB%8B)。 +vLLM-MindSpore插件以将MindSpore大模型接入vLLM，并实现服务化部署为功能目标: [vLLM-MindSpore插件简介](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/index.html#vllm-mindspore%E6%8F%92%E4%BB%B6%E7%AE%80%E4%BB%8B)。 MindSpore Transformers 套件的目标是构建一个大模型预训练、微调、评测、推理、部署的全流程开发套件，提供业内主流的 Transformer 类大语言模型（Large Language Models, LLMs）和多模态理解模型（Multimodal Models, MMs）。 @@ -14,8 +14,8 @@ MindSpore Transformers 套件的目标是构建一个大模型预训练、微调 环境安装步骤分为两种安装方式： -- [docker安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/installation/installation.html#docker%E5%AE%89%E8%A3%85)：适合用户快速使用的场景； -- [源码安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/installation/installation.html#%E6%BA%90%E7%A0%81%E5%AE%89%E8%A3%85)：适合用户有增量开发vLLM-MindSpore插件的场景。 +- [docker安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/installation/installation.html#docker%E5%AE%89%E8%A3%85)：适合用户快速使用的场景； +- [源码安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/installation/installation.html#%E6%BA%90%E7%A0%81%E5%AE%89%E8%A3%85)：适合用户有增量开发vLLM-MindSpore插件的场景。 ### 快速开始 @@ -31,7 +31,7 @@ export MINDFORMERS_MODEL_CONFIG=/path/to/yaml # 非MCore模型需要 ``` 目前vLLM MindSpore可支持不同的模型后端，以上环境变量指定MindSpore Tranformers 作为对接模型套件。非MCore模型需要配置模型的yaml配置文件。 -更多环境变量可参考：[环境变量](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/user_guide/environment_variables/environment_variables.html)。 +更多环境变量可参考：[环境变量](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/user_guide/environment_variables/environment_variables.html)。 准备好模型和环境变量后，即可开始推理。 @@ -39,15 +39,15 @@ export MINDFORMERS_MODEL_CONFIG=/path/to/yaml # 非MCore模型需要 vLLM在线推理面向实时服务场景，依托动态批处理和 OpenAI API，具有高并发、高吞吐、低延迟的特点，适用于企业级应用。 -- 单卡推理流程请参照：[单卡推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/tutorials/qwen2.5_7b_singleNPU/qwen2.5_7b_singleNPU.html) -- 单节点多卡推理流程请参照：[多卡推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/tutorials/qwen2.5_32b_multiNPU/qwen2.5_32b_multiNPU.html) -- 多节点的并行推理流程请参照：[多机并行推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/tutorials/deepseek_parallel/deepseek_r1_671b_w8a8_dp4_tp4_ep4.html) +- 单卡推理流程请参照：[单卡推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/tutorials/qwen2.5_7b_singleNPU/qwen2.5_7b_singleNPU.html) +- 单节点多卡推理流程请参照：[多卡推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/tutorials/qwen2.5_32b_multiNPU/qwen2.5_32b_multiNPU.html) +- 多节点的并行推理流程请参照：[多机并行推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/tutorials/deepseek_parallel/deepseek_r1_671b_w8a8_dp4_tp4_ep4.html) #### 离线推理 vLLM的离线推理专为高效处理大规模批量请求而设计，尤其适用于非实时，数据密集型的模型推理场景。 -离线推理流程请参照：[离线推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/quick_start/quick_start.html#%E7%A6%BB%E7%BA%BF%E6%8E%A8%E7%90%86) +离线推理流程请参照：[离线推理](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/quick_start/quick_start.html#%E7%A6%BB%E7%BA%BF%E6%8E%A8%E7%90%86) ### Mcore模型适配 @@ -69,7 +69,7 @@ MindSpore Transformers模型注册表中，注册模型配置类和模型类等 #### 版本配套信息 -各个组件的配套相关信息详见：[版本配套](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/installation/installation.html#%E7%89%88%E6%9C%AC%E9%85%8D%E5%A5%97)。 +各个组件的配套相关信息详见：[版本配套](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/installation/installation.html#%E7%89%88%E6%9C%AC%E9%85%8D%E5%A5%97)。 #### 模型支持列表 diff --git a/docs/mindstudio/docs/source_zh_cn/overview.md b/docs/mindstudio/docs/source_zh_cn/overview.md index f40533c79f..6af5d18c99 100644 --- a/docs/mindstudio/docs/source_zh_cn/overview.md +++ b/docs/mindstudio/docs/source_zh_cn/overview.md @@ -39,6 +39,6 @@ MindStudio Insight 可视化性能调优工具，提供时间线视图、算子耗时、通信瓶颈分析等功能，辅助快速分析模型性能瓶颈。 - 安装MindStudio Insight
查询版本配套关系 + 安装MindStudio Insight
查询版本配套关系 diff --git a/tutorials/source_en/custom_program/operation/cpp_api_for_custom_ops.md b/tutorials/source_en/custom_program/operation/cpp_api_for_custom_ops.md index 38b22aa3b6..f52772c1fc 100644 --- a/tutorials/source_en/custom_program/operation/cpp_api_for_custom_ops.md +++ b/tutorials/source_en/custom_program/operation/cpp_api_for_custom_ops.md @@ -51,7 +51,7 @@ kNumberTypeEnd, // End value for the Number type ### class Tensor -The `Tensor` class is defined in the [tensor.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor.h) header file, representing the tensor object in MindSpore. It provides methods for operating on and querying tensor properties. +The `Tensor` class is defined in the [tensor.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor.h) header file, representing the tensor object in MindSpore. It provides methods for operating on and querying tensor properties. #### Constructors @@ -302,7 +302,7 @@ The following methods are not part of the API and are used only in internal modu ### function tensor -Factory methods for constructing constant tensors, defined in the [tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor_utils.h) header file. +Factory methods for constructing constant tensors, defined in the [tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor_utils.h) header file. ```cpp Tensor tensor(int64_t value, TypeId dtype = TypeId::kNumberTypeInt64) @@ -319,7 +319,7 @@ Tensor tensor(const std::vector &value, TypeId dtype = TypeId::kNumberTy ### function ones -Factory method for constructing a tensor filled with ones, defined in the [tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor_utils.h) header file. +Factory method for constructing a tensor filled with ones, defined in the [tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor_utils.h) header file. ```cpp Tensor ones(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32) @@ -333,7 +333,7 @@ Tensor ones(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32) ### function zeros -Factory method for constructing a tensor filled with zeros, defined in the [tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor_utils.h) header file. +Factory method for constructing a tensor filled with zeros, defined in the [tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor_utils.h) header file. ```cpp Tensor zeros(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32) @@ -349,7 +349,7 @@ Tensor zeros(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32 ### class PyboostRunner -The `PyboostRunner` class for PyNative processes is defined in the [pyboost_extension.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/pynative/pyboost_extension.h) header file. It provides methods for managing execution, memory allocation, and kernel launching. +The `PyboostRunner` class for PyNative processes is defined in the [pyboost_extension.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/pyboost_extension.h) header file. It provides methods for managing execution, memory allocation, and kernel launching. `PyboostRunner` is a subclass of `std::enable_shared_from_this` and requires the use of the smart pointer `std::shared_ptr` to manage its objects. @@ -470,7 +470,7 @@ The `PyboostRunner` class for PyNative processes is defined in the [pyboost_exte ### class AtbOpRunner -The `AtbOpRunner` class is a runner for executing Ascend Transformer Boost (ATB) operators, defined in the [atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/atb/atb_common.h) header file. +The `AtbOpRunner` class is a runner for executing Ascend Transformer Boost (ATB) operators, defined in the [atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/atb/atb_common.h) header file. This class inherits from `PyboostRunner` and encapsulates the process of invoking ATB operators, including initialization, running the ATB operator, managing input/output tensors, memory allocation, and kernel scheduling. @@ -502,7 +502,7 @@ Refer to the tutorial [CustomOpBuilder Using AtbOpRunner to Integrate ATB Operat ### function RunAtbOp -The interface for executing ATB operators in dynamic graphs, defined in the [atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/atb/atb_common.h) header file. +The interface for executing ATB operators in dynamic graphs, defined in the [atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/atb/atb_common.h) header file. ```cpp template @@ -520,7 +520,7 @@ void RunAtbOp(const std::string &op_name, const ParamType ¶m, const std::vec ### class AsdSipFFTOpRunner -The `AsdSipFFTOpRunner` class is a runner for executing Ascend Sip Boost (ASDSIP) operators, defined in the [asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/asdsip/asdsip_common.h) header file. +The `AsdSipFFTOpRunner` class is a runner for executing Ascend Sip Boost (ASDSIP) operators, defined in the [asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/asdsip/asdsip_common.h) header file. This class inherits from `PyboostRunner` and encapsulates the process of invoking ASDSIP FFT operators, including initialization, running the ASDSIP FFT operator, managing input/output tensor, memory allocation, and kernel scheduling. @@ -550,7 +550,7 @@ Refer to the tutorial [CustomOpBuilder Integrates the ASDSIP FFT Operators throu ### function RunAsdSipFFTOp -The interface for executing ASDSIP FFT operators in dynamic graphs, defined in the [asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/asdsip/asdsip_common.h) header file. +The interface for executing ASDSIP FFT operators in dynamic graphs, defined in the [asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/asdsip/asdsip_common.h) header file. ```cpp inline void RunAsdSipFFTOp(const std::string &op_name, const FFTParam &fft_param, const ms::Tensor &input, diff --git a/tutorials/source_en/debug/sdc.md b/tutorials/source_en/debug/sdc.md index d24e210e23..5fb95ae7a9 100644 --- a/tutorials/source_en/debug/sdc.md +++ b/tutorials/source_en/debug/sdc.md @@ -369,7 +369,7 @@ When numerical anomalies are detected, the training task fails and alerts are re * Search application logs for **ERROR** level error logs with the keyword "accuracy sensitivity feature abnormal"; * Monitor the NPU health status: if Health Status displays Warning, Error Code displays 80818C00, and Error Information displays node type=SoC, sensor type=Check Sensor, event state=check fail; -* Check the [Ascend Device Plugin](https://github.com/Ascend/ascend-device-plugin) events, report error code 80818C00, event type is fault event, and the fault level is minor. +* Check the [MindCluster](https://gitcode.com/Ascend/mind-cluster) events, report error code 80818C00, event type is fault event, and the fault level is minor. When using combined detection, if feature value detection anomalies occur and CheckSum detects silent faults, warning logs can be found in the training logs: diff --git a/tutorials/source_en/model_infer/lite_infer/overview.md b/tutorials/source_en/model_infer/lite_infer/overview.md index a72b7047a8..5150712060 100644 --- a/tutorials/source_en/model_infer/lite_infer/overview.md +++ b/tutorials/source_en/model_infer/lite_infer/overview.md @@ -38,7 +38,7 @@ The MindSpore Lite inference framework supports the conversion of MindSpore trai 3. [Quantification after Training](https://www.mindspore.cn/lite/docs/en/r2.7.1/advanced/quantization.html) -4. [Lightweight Micro inference deployment](https://www.mindspore.cn/lite/docs/en/r2.7.1/advanced/micro.html#%20Model%20inference%20code%20generation) +4. [Lightweight Micro inference deployment](https://www.mindspore.cn/lite/docs/en/r2.7.1/advanced/micro.html#generating-model-inference-code) 5. [Benchmark Debugging Tool](https://www.mindspore.cn/lite/docs/en/r2.7.1/tools/benchmark.html) diff --git a/tutorials/source_en/model_infer/ms_infer/ms_infer_model_serving_infer.md b/tutorials/source_en/model_infer/ms_infer/ms_infer_model_serving_infer.md index 17b79d1819..d0e59f44e8 100644 --- a/tutorials/source_en/model_infer/ms_infer/ms_infer_model_serving_infer.md +++ b/tutorials/source_en/model_infer/ms_infer/ms_infer_model_serving_infer.md @@ -41,13 +41,13 @@ As an efficient service-oriented model inference backend, it should provide the ## Inference Tutorial -MindSpore inference works with the vLLM community solution to provide users with full-stack end-to-end inference service capabilities. The vLLM-MindSpore Plugin implements seamless interconnection of the vLLM community service capabilities in the MindSpore framework. For details, see [vLLM-MindSpore Plugin](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/index.html). +MindSpore inference works with the vLLM community solution to provide users with full-stack end-to-end inference service capabilities. The vLLM-MindSpore Plugin implements seamless interconnection of the vLLM community service capabilities in the MindSpore framework. For details, see [vLLM-MindSpore Plugin](https://www.mindspore.cn/vllm_mindspore/docs/en/master/index.html). This section describes the basic usage of vLLM-MindSpore Plugin service-oriented inference. ### Setting Up the Environment -The vLLM-MindSpore Plugin provides [Docker Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/installation/installation.html#docker-installation) and [Source Code Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/getting_started/installation/installation.html#source-code-installation) for users to do installation. The belows are steps for docker installation: +The vLLM-MindSpore Plugin provides [Docker Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/installation/installation.html#docker-installation) and [Source Code Installation](https://www.mindspore.cn/vllm_mindspore/docs/en/master/getting_started/installation/installation.html#source-code-installation) for users to do installation. The belows are steps for docker installation: **Building the Image** User can execute the following commands to clone the vLLM-MindSpore Plugin code repository and build the image: @@ -129,7 +129,7 @@ git lfs install git clone https://huggingface.co/Qwen/Qwen2-7B ``` -If `git lfs install` fails during the pull process, refer to the vLLM-MindSpore Plugin [FAQ](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/faqs/faqs.html) for a solution. +If `git lfs install` fails during the pull process, refer to the vLLM-MindSpore Plugin [FAQ](https://www.mindspore.cn/vllm_mindspore/docs/en/master/faqs/faqs.html) for a solution. ### Starting a Service @@ -141,7 +141,7 @@ export vLLM_MS_MODEL_BACKEND=MindFormers # use MindSpore Transformers as model b Here is an explanation of these environment variables: -- `vLLM_MS_MODEL_BACKEND`: The backend of the model to run. User could find supported models and backends for vLLM-MindSpore Plugin in the [Model Support List](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/user_guide/supported_models/models_list/models_list.html) and [Environment Variable List](https://www.mindspore.cn/vllm_mindspore/docs/en/r0.4.0/user_guide/environment_variables/environment_variables.html). +- `vLLM_MS_MODEL_BACKEND`: The backend of the model to run. User could find supported models and backends for vLLM-MindSpore Plugin in the [Model Support List](https://www.mindspore.cn/vllm_mindspore/docs/en/master/user_guide/supported_models/models_list/models_list.html) and [Environment Variable List](https://www.mindspore.cn/vllm_mindspore/docs/en/master/user_guide/environment_variables/environment_variables.html). Additionally, users need to ensure that MindSpore Transformers is installed. Users can add it by running the following command: diff --git a/tutorials/source_zh_cn/custom_program/operation/cpp_api_for_custom_ops.md b/tutorials/source_zh_cn/custom_program/operation/cpp_api_for_custom_ops.md index 1aea597fb9..227b8dcad1 100644 --- a/tutorials/source_zh_cn/custom_program/operation/cpp_api_for_custom_ops.md +++ b/tutorials/source_zh_cn/custom_program/operation/cpp_api_for_custom_ops.md @@ -51,7 +51,7 @@ kNumberTypeEnd, // Number 类型结束值 ### class Tensor -张量类定义在[tensor.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor.h)头文件中，表示 MindSpore 的张量对象，提供操作和查询张量属性的方法。 +张量类定义在[tensor.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor.h)头文件中，表示 MindSpore 的张量对象，提供操作和查询张量属性的方法。 #### 构造函数 @@ -302,7 +302,7 @@ kNumberTypeEnd, // Number 类型结束值 ### function tensor -构造常量张量的工厂方法，定义在[tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor_utils.h)头文件中。 +构造常量张量的工厂方法，定义在[tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor_utils.h)头文件中。 ```cpp Tensor tensor(int64_t value, TypeId dtype = TypeId::kNumberTypeInt64) @@ -319,7 +319,7 @@ Tensor tensor(const std::vector &value, TypeId dtype = TypeId::kNumberTy ### function ones -构造全1张量的工厂方法，定义在[tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor_utils.h)头文件中。 +构造全1张量的工厂方法，定义在[tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor_utils.h)头文件中。 ```cpp Tensor ones(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32) @@ -333,7 +333,7 @@ Tensor ones(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32) ### function zeros -构造全0张量的工厂方法，定义在[tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/common/tensor_utils.h)头文件中。 +构造全0张量的工厂方法，定义在[tensor_utils.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/tensor_utils.h)头文件中。 ```cpp Tensor zeros(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32) @@ -349,7 +349,7 @@ Tensor zeros(const ShapeVector &shape, TypeId dtype = TypeId::kNumberTypeFloat32 ### class PyboostRunner -PyNative 流程的运行器类，定义在[pyboost_extension.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/pynative/pyboost_extension.h)头文件中，为管理执行、内存分配和内核启动提供方法。 +PyNative 流程的运行器类，定义在[pyboost_extension.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/pyboost/custom/pyboost_extension.h)头文件中，为管理执行、内存分配和内核启动提供方法。 `PyboostRunner` 是 `std::enable_shared_from_this` 的子类，需要使用智能指针 `std::shared_ptr` 管理其对象。 @@ -470,7 +470,7 @@ PyNative 流程的运行器类，定义在[pyboost_extension.h](https://gitee.co ### class AtbOpRunner -用于执行 Ascend Transformer Boost (ATB) 算子的运行器类，定义在[atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/atb/atb_common.h)头文件中。 +用于执行 Ascend Transformer Boost (ATB) 算子的运行器类，定义在[atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/atb/atb_common.h)头文件中。 此类继承自 `PyboostRunner`，并封装了 ATB 算子的调用流程，包括初始化和运行 ATB 算子、管理输入输出 Tensor、内存分配及内核调度。 @@ -502,7 +502,7 @@ PyNative 流程的运行器类，定义在[pyboost_extension.h](https://gitee.co ### function RunAtbOp -动态图执行ATB算子的接口，定义在[atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/atb/atb_common.h)头文件中。 +动态图执行ATB算子的接口，定义在[atb_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/atb/atb_common.h)头文件中。 ```cpp template @@ -520,7 +520,7 @@ void RunAtbOp(const std::string &op_name, const ParamType ¶m, const std::vec ### class AsdSipFFTOpRunner -用于执行 Ascend Sip Boost (ASDSIP) 算子的运行器类，定义在[asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/asdsip/asdsip_common.h)头文件中。 +用于执行 Ascend Sip Boost (ASDSIP) 算子的运行器类，定义在[asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/asdsip/asdsip_common.h)头文件中。 此类继承自 `PyboostRunner`，并封装了 ASDSIP FFT 算子的调用流程，包括初始化和运行 ASDSIP FFT 算子、管理输入输出 Tensor、内存分配及内核调度。 @@ -550,7 +550,7 @@ void RunAtbOp(const std::string &op_name, const ParamType ¶m, const std::vec ### function RunAsdSipFFTOp -动态图执行ASDSIP FFT算子的接口，定义在[asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ccsrc/ms_extension/ascend/asdsip/asdsip_common.h)头文件中。 +动态图执行ASDSIP FFT算子的接口，定义在[asdsip_common.h](https://gitee.com/mindspore/mindspore/blob/v2.7.1/mindspore/ops/kernel/ascend/custom/pyboost_impl/asdsip/asdsip_common.h)头文件中。 ```cpp inline void RunAsdSipFFTOp(const std::string &op_name, const FFTParam &fft_param, const ms::Tensor &input, diff --git a/tutorials/source_zh_cn/debug/sdc.md b/tutorials/source_zh_cn/debug/sdc.md index da414fae1a..c6bd3dd530 100644 --- a/tutorials/source_zh_cn/debug/sdc.md +++ b/tutorials/source_zh_cn/debug/sdc.md @@ -369,7 +369,7 @@ $ grep -m1 'Global CheckSum result is' worker_0.log * 通过搜索应用类日志，查询**ERROR**级别错误日志，关键字"accuracy sensitivity feature abnormal"； * 通过监控NPU健康状态：Health Status显示Warning，Error Code显示80818C00，Error Information显示node type=SoC, sensor type=Check Sensor, event state=check fail； -* 通过查看[Ascend Device Plugin](https://github.com/Ascend/ascend-device-plugin)事件，上报错误码80818C00，事件类型为故障事件，故障级别次要。 +* 通过查看[MindCluster](https://gitcode.com/Ascend/mind-cluster)事件，上报错误码80818C00，事件类型为故障事件，故障级别次要。 当使用联合检测时，若训练中发生特征值异常、CheckSum检测出静默故障，会在业务训练日志中产生告警： diff --git a/tutorials/source_zh_cn/model_infer/ms_infer/ms_infer_model_serving_infer.md b/tutorials/source_zh_cn/model_infer/ms_infer/ms_infer_model_serving_infer.md index d07e17396b..9c30e51b01 100644 --- a/tutorials/source_zh_cn/model_infer/ms_infer/ms_infer_model_serving_infer.md +++ b/tutorials/source_zh_cn/model_infer/ms_infer/ms_infer_model_serving_infer.md @@ -41,13 +41,13 @@ print(generate_text) ## 推理教程 -MindSpore推理结合vLLM社区方案，为用户提供了全栈端到端的推理服务化能力，通过vLLM-MindSpore插件实现vLLM社区的服务化能力在MindSpore框架下的无缝对接，具体可以参考[vLLM-MindSpore插件文档](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/index.html)。 +MindSpore推理结合vLLM社区方案，为用户提供了全栈端到端的推理服务化能力，通过vLLM-MindSpore插件实现vLLM社区的服务化能力在MindSpore框架下的无缝对接，具体可以参考[vLLM-MindSpore插件文档](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/index.html)。 本章主要简单介绍vLLM-MindSpore插件服务化推理的基础使用。 ### 环境准备 -vLLM-MindSpore插件提供了[docker安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/installation/installation.html#docker%E5%AE%89%E8%A3%85)与[源码安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/getting_started/installation/installation.html#%E6%BA%90%E7%A0%81%E5%AE%89%E8%A3%85)的方式，让用户可以便捷地安装使用vLLM-MindSpore插件。以下是部署docker的步骤介绍： +vLLM-MindSpore插件提供了[docker安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/installation/installation.html#docker%E5%AE%89%E8%A3%85)与[源码安装](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/getting_started/installation/installation.html#%E6%BA%90%E7%A0%81%E5%AE%89%E8%A3%85)的方式，让用户可以便捷地安装使用vLLM-MindSpore插件。以下是部署docker的步骤介绍： **构建镜像** @@ -130,7 +130,7 @@ git lfs install git clone https://huggingface.co/Qwen/Qwen2-7B ``` -若在拉取过程中，执行`git lfs install失败`，可以参考vLLM-MindSpore插件 [FAQ](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/faqs/faqs.html) 进行解决。 +若在拉取过程中，执行`git lfs install失败`，可以参考vLLM-MindSpore插件 [FAQ](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/faqs/faqs.html) 进行解决。 ### 启动服务 @@ -142,7 +142,7 @@ export vLLM_MS_MODEL_BACKEND=MindFormers # use MindSpore Transformers as model b 以下是对上述环境变量的解释： -- `vLLM_MS_MODEL_BACKEND`：所运行的模型后端。目前vLLM-MindSpore插件所支持的模型与模型后端，可在[模型支持列表](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/user_guide/supported_models/models_list/models_list.html)与[环境变量清单](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/r0.4.0/user_guide/environment_variables/environment_variables.html)中进行查询。 +- `vLLM_MS_MODEL_BACKEND`：所运行的模型后端。目前vLLM-MindSpore插件所支持的模型与模型后端，可在[模型支持列表](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/user_guide/supported_models/models_list/models_list.html)与[环境变量清单](https://www.mindspore.cn/vllm_mindspore/docs/zh-CN/master/user_guide/environment_variables/environment_variables.html)中进行查询。 另外，用户需要确保MindSpore Transformers已安装。用户可通过以下方式引入MindSpore Transformers： -- Gitee