更新时间：2026-05-07 GMT+08:00
基于自定义镜像创建Tensorboard训练作业场景
本节通过调用一系列API，以训练模型为例介绍ModelArts API的使用流程。
概述
使用PyTorch框架创建训练作业的流程如下：
前提条件
- 已获取IAM的EndPoint和ModelArts的EndPoint。
- 确认服务的部署区域，获取项目ID和名称、获取账号名和ID和获取用户名和用户ID。
- 已准备好Tensorboard日志保存目录，例如OBS并行文件系统的“obs://cnnorth4-job-pfs/summary”目录下
操作步骤
- 调用认证鉴权接口获取用户的Token。
- 请求消息体：
URI格式：POST https://{iam_endpoint}/v3/auth/tokens
请求消息头：Content-Type →application/json
请求Body：{ "auth": { "identity": { "methods": ["password"], "password": { "user": { "name": "user_name", "password": "user_password", "domain": { "name": "domain_name" } } } }, "scope": { "project": { "name": "cn-north-1" } } } }其中，加粗的斜体字段需要根据实际值填写：- iam_endpoint为IAM的终端节点。
- user_name为IAM用户名。
- user_password为用户登录密码。
- domain_name为用户所属的账号名。
- cn-north-1为项目名，代表服务的部署区域。
- 返回状态码“201 Created”，在响应Header中获取“X-Subject-Token”的值即为Token，如下所示：
x-subject-token →MIIZmgYJKoZIhvcNAQcCoIIZizCCGYcCAQExDTALBglghkgBZQMEAgEwgXXXXXX...
- 请求消息体：
- 调用创建训练作业接口使用刚创建的算法返回的uuid创建一个训练作业，记录训练作业id。
- 请求消息体：
URI格式：POST https://{ma_endpoint}/v2/{project_id}/training-jobs
请求消息头：
- X-Auth-Token →MIIZmgYJKoZIhvcNAQcCoIIZizCCGYcCAQExDTALBglghkgBZQMEAgEwgXXXXXX...
- Content-Type →application/json
其中，加粗的斜体字段需要根据实际值填写。
请求Body：
{ "kind": "job", "metadata": { "name": "test-pytorch-cpu01", "description": "test pytorch work cpu", "annotations": { "tensorboard/enable": "true" } }, "algorithm": { "engine": { "image_url": "atelier/pytorch_cuda:pytorch_2.7.0-cuda_12.8-py_3.11.10-ubuntu_22.04-x86_64-20251215163925-4e5422a" }, "command": "sleep 1h", "summary": { "data_sources": [ { "pfs": { "pfs_path": "obs://cnnorth4-job-pfs/summary/" } } ] } }, "spec": { "resource": { "node_count": 1, "pool_id": "pool-maostest-train-06024304be00d5092fbdc0013d201342" } } }其中，加粗的斜体字段需要根据实际值填写：
- “kind”填写训练作业的类型，默认为job。
- “metadata”下的“name”和“description”填写训练作业的名称和描述。
- “metadata”下的“annotations”填写"tensorboard/enable": "true"表示启用tensorboard。
- “algorithm”下的“image_url”填写训练作业镜像的地址。
- “algorithm”下的“command”填写训练作业启动命令。
- “algorithm”下的“data_sources”填写Tensorboard日志的保存目录。
- “spec”字段下的“pool_id”表示训练作业所依赖的资源池ID。“node_count”表示训练是否需要多机训练（分布式训练），此处为单机情况使用默认值“1”。“log_export_path”用于指定用户需要上传日志的obs目录。
- 返回状态码“201 Created”，表示训练作业创建成功，响应Body如下所示：
{ "kind": "job", "metadata": { "id": "dc0e330d-c6a8-4f1d-9cd3-1f108c3bbaf9", "name": "test-pytorch-cpu01", "description": "test pytorch work cpu", "create_time": 1777545590838, "workspace_id": "0", "ai_project": "default-ai-project", "labels": { "training-job": "modelarts-os" }, "user_name": "", "annotations": { "job_template": "Template DL", "key_task": "worker", "tensorboard/enable": "true" }, "training_experiment_reference": {}, "tags": [] }, "status": { "phase": "Pending", "secondary_phase": "Creating", "pending_time": 2000, "duration": 0, "is_hanged": false, "retry_count": 0, "start_time": 0, "node_count_metrics": null, "tasks": [ "worker-0" ], "metrics_statistics": { "cpu_usage": { "average": -1, "max": -1, "min": -1 }, "mem_usage": { "average": -1, "max": -1, "min": -1 } } }, "algorithm": { "command": "sleep 1h", "engine": { "engine_id": "", "engine_name": "", "engine_version": "", "v1_compatible": false, "image_url": "atelier/pytorch_cuda:pytorch_2.7.0-cuda_12.8-py_3.11.10-ubuntu_22.04-x86_64-20251215163925-4e5422a", "non_swr_image": false, "run_user": "", "image_source": true, "image_repo_id": "", "image_id": "" }, "summary": { "data_sources": [ { "pfs": { "pfs_path": "obs://cnnorth4-job-pfs/summary/" } } ] } }, "spec": { "resource": { "pool_id": "pool-maostest-train-06024304be00d5092fbdc0013d201342", "pool_resource_flavor": "", "node_count": 1, "pool_info": { "cpu_arch": "x86", "core_num": 5, "mem_size": 22, "cache_size": 0, "accelerator": "", "accelerator_num": 0, "accelerator_type": "", "accelerator_size": 0, "variant": "", "huge_pages": 0, "x_parameter_plane": "", "use_privileged": false, "use_host_network": false, "use_ib_network": false, "project_id": "", "pool_resource_flavor": "liumuqi-eni-test", "pool_id": "pool-maostest-train-06024304be00d5092fbdc0013d201342", "cluster_id": "", "maos_pool": true, "quota_id": "", "maos_migrated": false, "detect_all_in_int": false, "pool_type": "", "enable_cabinet": false, "enable_memarts": false, "enable_ems": false, "empty_dir_size": 0 }, "main_container_allocated_resources": { "cpu_arch": "x86", "cpu_core_num": 4, "mem_size": 20, "accelerator_num": 0, "accelerator_type": "" } }, "is_hosted_log": true, "runtime_type": "production" }, "endpoints": { "tensorboard": {} }, "ftjob_config": { "checkpoint_config": { "save_checkpoints_max": 0, "checkpoint_id": "", "skipped_steps": 0, "restore_training": 0 }, "task_env": { "envs": null } } }- 记录“metadata”下的“id”（训练作业的任务ID）字段的值便于后续步骤使用。
- “Status”下的“phase”和“secondary_phase”为表示训练作业的状态和下一步状态。示例中“Creating”表示训练作业正在创建中。
- 请求消息体：
- 调用查询训练作业详情接口使用刚创建的训练作业返回的uuid查询训练作业状态。
- 请求消息体：
URI格式：GET https://{ma_endpoint}/v2/{project_id}/training-jobs/{training_job_id}
请求消息头：X-Auth-Token →MIIZmgYJKoZIhvcNAQcCoIIZizCCGYcCAQExDTALBglghkgBZQMEAgEwgXXXXXX...
其中，加粗的斜体字段需要根据实际值填写：
“training_job_id”为2记录的训练作业的任务ID。
- 返回状态码“200 OK”，响应Body如下所示：
{ "kind": "job", "metadata": { "id": "dc0e330d-c6a8-4f1d-9cd3-1f108c3bbaf9", "name": "test-pytorch-cpu01", "description": "test pytorch work cpu", "create_time": 1777545590838, "workspace_id": "0", "ai_project": "default-ai-project", "labels": { "training-job": "modelarts-os" }, "user_name": "ei_modelarts_y00218826_05", "annotations": { "job_template": "Template DL", "key_task": "worker", "tensorboard/enable": "true" }, "training_experiment_reference": {}, "tags": [] }, "status": { "phase": "Running", "secondary_phase": "Running", "pending_time": 77162, "duration": 181000, "is_hanged": false, "retry_count": 0, "task_ips": [ { "task": "worker-0", "ip": "172.16.0.79", "host_ip": "192.168.209.44", "schedule_count": 1 } ], "start_time": 1777545668000, "node_count_metrics": [ [ 1777545658000, 0 ], [ 1777545667000, 0 ], [ 1777545668000, 1 ], [ 1777545848000, 1 ], [ 1777545849000, 1 ] ], "tasks": [ "worker-0" ], "metrics_statistics": { "cpu_usage": { "average": -1, "max": -1, "min": -1 }, "mem_usage": { "average": -1, "max": -1, "min": -1 } }, "running_records": [ { "start_at": 1777545675, "start_type": "init_or_rescheduled" } ] }, "algorithm": { "command": "sleep 1h", "engine": { "engine_id": "", "engine_name": "", "engine_version": "", "v1_compatible": false, "image_url": "atelier/pytorch_cuda:pytorch_2.7.0-cuda_12.8-py_3.11.10-ubuntu_22.04-x86_64-20251215163925-4e5422a", "non_swr_image": false, "run_user": "", "image_source": true, "image_repo_id": "", "image_id": "" }, "summary": { "data_sources": [ { "pfs": { "pfs_path": "obs://cnnorth4-job-pfs/summary/" } } ] } }, "spec": { "resource": { "pool_id": "pool-maostest-train-06024304be00d5092fbdc0013d201342", "pool_resource_flavor": "", "node_count": 1, "pool_info": { "cpu_arch": "x86", "core_num": 5, "mem_size": 22, "cache_size": 0, "accelerator": "", "accelerator_num": 0, "accelerator_type": "", "accelerator_size": 0, "variant": "", "huge_pages": 0, "x_parameter_plane": "", "use_privileged": false, "use_host_network": false, "use_ib_network": false, "project_id": "", "pool_resource_flavor": "liumuqi-eni-test", "pool_id": "pool-maostest-train-06024304be00d5092fbdc0013d201342", "cluster_id": "", "maos_pool": true, "quota_id": "", "maos_migrated": false, "detect_all_in_int": false, "pool_type": "", "enable_cabinet": false, "enable_memarts": false, "enable_ems": false, "empty_dir_size": 0 }, "main_container_allocated_resources": { "cpu_arch": "x86", "cpu_core_num": 4, "mem_size": 20, "accelerator_num": 0, "accelerator_type": "" } }, "is_hosted_log": true, "runtime_type": "production" }, "endpoints": { "tensorboard": { "url": "https://authoring-modelarts-cnnorth7.ulanqab.huawei.com/3efaddb9-dc0e330d-c6a8-4f1d-9cd3-1f108c3bbaf9/proxy/6006/", "token": "fa8e321f772xxxxxxxxxxx3f1844d06" } }, "ftjob_config": { "checkpoint_config": { "save_checkpoints_max": 0, "checkpoint_id": "", "skipped_steps": 0, "restore_training": 0 }, "task_env": { "envs": null } } }根据响应可以了解训练作业的版本详情，其中“status”为“Running”表示训练作业正在运行。当训练作业运行后，开始启动Tensorboard，Tensorboard启动成功后作业详情中的endpoints会返回tensorboard的打开地址和token。
- 请求消息体：
- 打开Tensorboard Endpoint查看可视化指标数据。
URI格式：GET {tensorboard_endpoint}?token={tensorboard_token}
其中，加粗的斜体字段需要根据实际值填写:
- “tensorboard_endpoint”为训练作业的Tensorboard打开地址。
- “tensorboard_token”训练作业的Tensorboard访问凭证。
- 当训练作业使用完成或不再需要时，调用删除训练作业接口删除训练作业。
父主题： 应用示例
