腾讯云TI-ONE训练平台提交训练任务_AI解决方案_同尘科技

describe_postpaid_training_price(self)    查询每种配置的每小时的价格，价格单位：元        :rtype:     tencentcloud.tione.v20211111.models.DescribeBillingSpecsResponse
describe_system_reasoning_images(self)    获取平台内置的推理镜像
    :return:    推理镜像信息    :rtype:     :class:`tikit.tencentcloud.tione.v20211111.models.DescribeInferTemplatesResponse`    数据格式如：    {        "FrameworkTemplates": [        {            "Framework": "TENSORFLOW",            "FrameworkVersion": "2.4",            "Groups": [            "TENSORFLOW",            "LIGHT"            ],            "InferTemplates": [            {                "InferTemplateId": "tf2.4-py38-cpu",                "InferTemplateImage": "ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-tensorflow-cpu:py38-tensorflow2.4-cpu-20211206"            },            {                "InferTemplateId": "tf2.4-py38-gpu",                "InferTemplateImage": "ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-tensorflow-gpu:py38-tensorflow2.4-cu110-20211206"            }            ]        }        ],        "RequestId": "3654e19b-c2ba-4953-b131-d66495723008"    }
    print的返回结果，输出如下：（镜像标识用来配置新模型的运行环境    +------------+----------------+---------------------------------------------+----------------------------------+-------------------------------------------------------------------------------------------------------------+    |  算法框架  | 算法框架版本号 |              支持的训练框架集合             |             镜像标识             |                                                   镜像url                                                   |    +------------+----------------+---------------------------------------------+----------------------------------+-------------------------------------------------------------------------------------------------------------+    | TENSORFLOW |      2.4       |           ['TENSORFLOW', 'LIGHT']           |          tf2.4-py38-cpu          |         ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-tensorflow-cpu:py38-tf2.4-cpu-1.0.0        |    | TENSORFLOW |      2.4       |           ['TENSORFLOW', 'LIGHT']           |          tf2.4-py38-gpu          |       ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-tensorflow-cpu:py38-tf2.4-cu11.0-1.0.0       |    | TENSORFLOW |      1.15      |      ['TENSORFLOW', 'LIGHT', 'TI_ACC']      |         tf1.15-py37-cpu          |        ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-tensorflow-cpu:py37-tf1.15-cpu-1.0.0        |    | TENSORFLOW |      1.15      |      ['TENSORFLOW', 'LIGHT', 'TI_ACC']      |         tf1.15-py37-gpu          |       ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-tensorflow-cpu:py37-tf1.15-cu10.0-1.0.0      |    |  PYTORCH   |      1.9       | ['PYTORCH', 'LIGHT', 'TI_ACC', 'AUTOML_CV'] |       torch1.9.0-py38-cpu        |        ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-pytorch-cpu:py38-torch1.9.0-cpu-1.0.0       |    |  PYTORCH   |      1.9       | ['PYTORCH', 'LIGHT', 'TI_ACC', 'AUTOML_CV'] |      torch1.9.0-py38-cu111       |       ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-pytorch-gpu:py38-torch1.9.0-cu111-1.0.0      |    |    PMML    |     0.9.12     |             ['SPARK', 'PYSPARK']            |           pypmml-py38            |             ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-pmml:py38-pmml0.9.16-1.0.0             |    |    OCR     |      112       |                ['AUTOML_OCR']               |           ocr-serving            |                ccr.ccs.tencentyun.com/qcloud-ti-platform/ocr_serving_gpu:serving-v121-335584d               |    | DETECTRON2 |      112       |                ['DETECTRON2']               | detectron2-torch1.9.0-py38-cu111 | ccr.ccs.tencentyun.com/qcloud-ti-platform/ti-cloud-infer-pytorch-gpu:py38-torch1.9.0-detectron2-cu111-1.0.0 |    +------------+----------------+---------------------------------------------+----------------------------------+-------------------------------------------------------------------------------------------------------------+
def create_training_task(self,                            name,                            framework,                            cos_output,                            worker_resource,                            code_package_path,                            ps_resource=None,                            input_data_config=None,                            worker_start_cmd=None,                            ps_start_cmd=None,                            tags=None,                            tuning_parameters_dict={},                            resource_group_id="",                            remark=None,                            log_enable=False,                            log_logset_id=None,                            log_topic_id=None,                            vpc_id=None,                            sub_net_id=None):    """创建训练任务
    :param name:        任务名称    :type name:         str    :param framework:   运行的框架环境    :type framework:    :class:`tikit.models.FrameworkInfo`    :param cos_output:          输出的cos信息    :type cos_output:           str    :param worker_resource:     worker节点的配置    :type worker_resource:      :class:`tikit.models.ResourceConfigInfo`    :param code_package_path:   代码的cos信息    :type code_package_path:    str    :param ps_resource:         ps节点的配置    :type ps_resource:          :class:`tikit.models.ResourceConfigInfo`    :param input_data_config:   输入的数据信息    :type input_data_config:    list or :class:`tikit.models.TrainingDataConfig`    :param worker_start_cmd:    worker的启动命令    :type worker_start_cmd:     str    :param ps_start_cmd:        ps节点的启动命令    :type ps_start_cmd:         str    :param tags:                标签    :type tags:                 list of :class:`tikit.tencentcloud.tione.v20211111.models.Tag`    :param tuning_parameters_dict:  调参字典    :type tuning_parameters_dict:   dict    :param resource_group_id:   预付费的资源组id    :type resource_group_id:    str    :param remark:              描述    :type remark:               str    :param log_enable:          日志开关    :type log_enable:           bool    :param log_logset_id:       日志集id    :type log_logset_id:        str    :param log_topic_id:        日志的topic id    :type log_topic_id:         str    :param vpc_id:              vpc的id    :type vpc_id:               str    :param sub_net_id:          子网id    :type sub_net_id:           str    :return:    :rtype: :class:`tikit.tencentcloud.tione.v20211111.models.CreateTrainingTaskResponse`    """
def describe_training_frameworks(self):    """查看训练框架
    :rtype: :class:`tikit.tencentcloud.tione.v20211111.models.DescribeTrainingFrameworksResponse`    print返回的结果，输出如下：    +------------+----------------------------------------+-------------------------+    | 框架名称   | 版本                                   | 训练模式                |    +------------+----------------------------------------+-------------------------+    | TENSORFLOW | ti-acc1.0-tf1.15-py3.6-cuda10.0-gpu    | PS_WORKER               |    | TENSORFLOW | light3.1.3-tf2.4-py3.8-cuda11.1-gpu    | MPI, HOROVOD            |    | TENSORFLOW | tf1.15-py3.7-cpu                       | PS_WORKER, MPI, HOROVOD |    | TENSORFLOW | tf1.15-py3.7-cuda10.0-gpu              | PS_WORKER, MPI, HOROVOD |    | TENSORFLOW | tf2.4-py3.8-cpu                        | PS_WORKER, MPI, HOROVOD |    | TENSORFLOW | tf2.4-py3.8-cuda11.1-gpu               | PS_WORKER, MPI, HOROVOD |    | PYTORCH    | ti-acc1.0-torch1.9-py3.8-cuda11.1-gpu  | DDP                     |    | PYTORCH    | light3.1.3-torch1.9-py3.8-cuda11.1-gpu | DDP, MPI, HOROVOD       |    | PYTORCH    | torch1.9-py3.8-cuda11.1-gpu            | DDP, MPI, HOROVOD       |    | SPARK      | spark2.4.5-cpu                         | SPARK                   |    | PYSPARK    | spark2.4.5-py3.6-cpu                   | SPARK                   |    +------------+----------------------------------------+-------------------------+    """
def describe_training_tasks(self, filters=None, tag_filters=None, offset=0, limit=50, order="DESC",                            order_field="UpdateTime"):    """获取训练任务列表
    :param filters:     过滤器，eg：[{ "Name": "TaskStatus", "Values": ["Running"] }]    :type filters:      list of Filter    :param tag_filters: 标签过滤器，eg：[{ "TagKey": "TagKeyA", "TagValue": ["TagValueA"] }]    :type tag_filters:  list of TagFilter    :param offset:      偏移量，默认为0    :type offset:       int    :param limit:       返回数量，默认为50    :type limit:        int    :param order:       输出列表的排列顺序。取值范围：ASC：升序排列 DESC：降序排列    :type order:        str    :param order_field: 排序的依据字段， 取值范围 "CreateTime" "UpdateTime"    :type order_field:  str    :return:    :rtype: :class:`tikit.tencentcloud.tione.v20211111.models.DescribeTrainingTasksResponse`    """
describe_training_task(self, task_id)    获取单个训练任务信息        :param task_id: 训练任务ID    :type task_id: str    :rtype: :class:`tencentcloud.tione.v20211111.models.DescribeTrainingTaskResponse`
describe_training_task_pods(self, task_id)    获取训练任务的pod列表        :param task_id: 训练任务ID    :type task_id: str    :rtype: :class:`tencentcloud.tione.v20211111.models.DescribeTrainingTaskPodsResponse`
def describe_train_logs(self, pod_name, start_time=None, end_time=None, limit=None, order=None,                        context=None, filters=None):    """查看训练任务的日志
    :param pod_name: 查询哪个Pod的日志，支持通配符。查看某个训练任务的全部pod的日志可以填： "*"，如：train-51cd6bf7ec1000*    :type pod_name: str    :param start_time: 日志查询开始时间。RFC3339格式的时间字符串，比如2021-12-16T13:20:24+08:00，默认值为当前时间的前一个小时    :type start_time: str    :param end_time: 日志查询结束时间。RFC3339格式的时间字符串，比如2021-12-16T13:20:24+08:00，默认值为当前时间    :type end_time: str    :param limit: 日志查询条数，默认值100，最大值100    :type limit: int    :param order: 排序方向。(ASC | DESC) 默认值为DESC    :type order: str    :param context: 分页的游标    :type context: str    :param filters: 过滤Filters    :type filters: list of tikit.tencentcloud.tione.v20211111.models.Filter    :rtype: :class:`tikit.tencentcloud.tione.v20211111.models.DescribeLogsResponse`
        返回的对象如果非空，就会有 next() 方法，能不断地获取下一页的日志（如果有多页的话），如下：        now_time = datetime.datetime.now(datetime.timezone.utc)        now_time_str = now_time.isoformat()        result = client.describe_train_logs("train-51cd6bf7ec1000-37c5p5nlr01s-launcher",                                            "2021-12-10T09:32:03.823509+00:00",                                            now_time_str,                                            limit=30)        print(result)        print(result.next())        print(result.next())        print(result.next())    """
def push_training_metrics(self, timestamp, value_map, task_id=None, epoch=None, total_steps=None, step=None):    """上报训练自定义指标（单条）。单个子账号每秒可以调用20次，请在您的训练代码中注意控制上报频率，避免超限报错。或者使用push_training_metrics_list
    :param timestamp:   时间戳    :type timestamp:    int    :param value_map:   指标映射。 指标名称 -> 指标值    :type value_map:    map: str -> float    :param task_id:     任务ID。若为空，就当前取任务节点环境的 TI_TASK_ID 环境变量的值    :type task_id:      str    :param epoch:       epoch值    :type epoch:        int    :param total_steps: 总步数    :type total_steps:  int    :param step:        第几步    :type step:         int    :return:    :rtype:             :class:`tikit.tencentcloud.tione.v20211111.models.PushTrainingMetricsResponse`
    client.push_training_metrics(int(time.time()), {"field1": 11, "field2": 12}, "task-id-00001", 3, 1000, 66)    """
push_training_metrics_list(self, metric_list)    上报训练自定义指标（列表）        :param metric_list: MetricData 数组。 若任务ID为空，就当前取任务节点环境的 TI_TASK_ID 环境变量的值    :type metric_list:  list of :class:`tencentcloud.tione.v20211111.models.MetricData`    :return:    :rtype:             :class:`tencentcloud.tione.v20211111.models.PushTrainingMetricsResponse`
describe_training_metrics(self, task_id)    查询训练自定义指标        :param task_id: 任务ID    :type task_id: str    :rtype: :class:`tencentcloud.tione.v20211111.models.DescribeTrainingMetricsResponse`
stop_training_task(self, task_id)    停止某个训练任务        :param task_id: 训练任务ID    :type task_id: str    :rtype: :class:`tencentcloud.tione.v20211111.models.StopTrainingTaskResponse`
delete_training_task(self, task_id)    删除某个训练任务        :param task_id: 训练任务ID    :type task_id: str    :rtype: :class:`tencentcloud.tione.v20211111.models.DeleteTrainingTaskResponse`
def describe_train_resource_groups(self, offset=0, limit=20, search_word="", tag_filters=None):    """获取训练资源组列表
    :param offset: 偏移量，默认为0；分页查询起始位置，如：Limit为100，第一页Offset为0，第二页OffSet为100....即每页左边为开区间    :type offset: int    :param limit: 返回数量，默认为20，最大值为30；分页查询每页大小，最大30    :type limit: int    :param search_word: 支持模糊查找资源组id和资源组名    :type search_word: str    :param tag_filters: 标签过滤    :type tag_filters: list of tikit.tencentcloud.tione.v20211111.models.Tag    :return:    :rtype:    :class:`tikit.tencentcloud.tione.v20211111.models.DescribeBillingResourceGroupsResponse`    """

中间配置：

tikit.models.FrameworkInfo
def new_custom(training_mode, image_type, image_url, registry_region=None, registry_id=None):    """自定义训练框架的配置
    :param training_mode:   训练模式。 通过describe_training_frameworks()查看列表    :type training_mode:    str    :param image_type:      腾讯云容器镜像服务的镜像类型，如"CCR"    :type image_type:       str    :param image_url:       腾讯云容器镜像服务的镜像地址    :type image_url:        str    :param registry_region: 腾讯云容器镜像服务的镜像仓库的域    :type registry_region:  str    :param registry_id:     腾讯云容器镜像服务的镜像仓库ID    :type registry_id:      str    :return:    :rtype:    """
def new_custom_image(image_type, image_url, registry_region=None, registry_id=None):    """自定义镜像的配置
    :param image_type:      腾讯云容器镜像服务的镜像类型，如"CCR"    :type image_type:       str    :param image_url:       腾讯云容器镜像服务的镜像地址    :type image_url:        str    :param registry_region: 腾讯云容器镜像服务的镜像仓库的域    :type registry_region:  str    :param registry_id:     腾讯云容器镜像服务的镜像仓库ID    :type registry_id:      str    :return:    :rtype:    """
def new_system_framework(framework_name, framework_environment, training_mode):    """系统内置的训练框架
    :param framework_name:      框架名称。 通过describe_training_frameworks()查看列表    :type framework_name:       str    :param framework_environment:   框架环境。 通过describe_training_frameworks()查看列表    :type framework_environment:    str    :param training_mode:       训练模式。 通过describe_training_frameworks()查看列表    :type training_mode:        str    :return:    :rtype:    """
----------------------------------------------------tikit.models.ResourceConfigInfo
def new_postpaid(instance_type, instance_num):    """获取后付费模式下的资源配置
    :param instance_type:   实例类型。通过 describe_postpaid_training_price() 查看实例列表    :type instance_type:    str    :param instance_num:    实例数量    :type instance_num:     int    :return:    :rtype:    """|  def new_prepaid(cpu, memory, gpu=0, gpu_type=None, instance_num=1):    """获取预付费模式下的资源配置
    :param cpu:     CPU个数，单位是核    :type cpu:      float    :param memory:  内存大小，单位是GB    :type memory:   float    :param gpu_type: gpu类型    :type gpu_type: str    :param gpu:     gpu个数    :type gpu:      float    :param instance_num:    实例数量    :type instance_num:     int    :return:    :rtype:    """
----------------------------------------------------tikit.models.TrainingDataConfig
def new_mount_cos(cos_str, target_path):    """一个cos类型的训练数据
    :param cos_str:      cos存储，格式： //    :type cos_str: str    :param  target_path:  目标挂载路径    :type target_path: str    :return:    :rtype:    """
def new_dataset_mount(dataset_id, target_path):    """一个dataset类型的训练数据
    :param dataset_id:  数据集ID    :type dataset_id: str    :param  target_path:  目标挂载路径    :type target_path: str    :return:    :rtype:    """
def new_mount_cfs(cfs_id, source_path, target_path):    """新建一个cfs类型的训练数据集配置
    :param cfs_id:      CFS的ID    :type cfs_id: str    :param  source_path: CFS的路径    :type source_path: str    :param  target_path: 目标挂载路径    :type target_path: str    :return:    :rtype:    """
def new_mount_hdfs(hdfs_id, source_path, target_path):    """新建一个hdfs类型的训练数据集配置
    :param hdfs_id:      EMR上HDFS的ID    :type hdfs_id: str    :param  source_path: HDFS的路径    :type source_path: str    :param  target_path: 目标挂载路径    :type target_path: str    :return:    :rtype:    """
def new_mount_wedata_hdfs(wedata_id, source_path):    """新建一个wedata hdfs类型的训练数据集配置
    :param wedata_id:     wedata数据源id    :type wedata_id: int    :param  source_path: HDFS的路径    :type source_path: str    :return:    :rtype:    """
def new_dataset(id_target_dict):    """ Deprecated !    新建一个dataset类型的训练数据集配置
    :param id_target_dict:  数据集信息。 dataset id -> 下载的目标路径    :type id_target_dict:   dict    :return:    :rtype:    """
def new_cos_data(cos_str_target_dict):    """Deprecated !    新建一个cos类型的训练数据集配置
    :param cos_str_target_dict:     数据集信息。  // -> 下载的目标路径    :type cos_str_target_dict:      dict    :return:    :rtype:    """




对解决方案有疑惑？想了解解决方案收费？ 联系解决方案专家

腾讯云限时活动1折起，即将结束： 马上收藏

同尘科技为腾讯云授权服务中心，购买腾讯云享受折上折，更有现金返利:同意关联，立享优惠

阿里云解决方案也看看？： 点击对比阿里云的解决方案