书栈网 · BookStack 本次搜索耗时 0.231 秒,为您找到 836 个相关结果.
  • torch.nn

    torch.nn Parameters(参数) Containers(容器) Module(模块) Sequential ModuleList (模块列表) ModuleDict (模块词典) ParameterList (参数列表) ParameterDict (参数词典) Convolution layers (卷积层) Conv1d ...
  • torch

    7370 2018-04-07 《PyTorch中文文档》
    张量 Tensors 创建操作 Creation Ops 索引,切片,连接,换位Indexing, Slicing, Joining, Mutating Ops torch.cat torch.gather torch.index_select torch.masked_select torch.nonzero torch.split tor...
  • torch

    张量 Tensors 创建操作 Creation Ops 索引,切片,连接,换位Indexing, Slicing, Joining, Mutating Ops torch.cat torch.gather torch.index_select torch.masked_select torch.nonzero torch.split tor...
  • Job

    Job 运行示例 Job 编写 Job 规约 Job 标签 Pod 模板 Pod 选择算符 Job 的并行执行 控制并行性 完成模式 说明: 处理 Pod 和容器失效 Pod 回退失效策略 说明: 逐索引的回退限制 说明: Pod 失效策略 说明: 说明: 说明: 说明: 成功策略 说明: 说明: Job 终止与...
  • Torch

    Torch 张量 Tensors 创建操作 Creation Ops 索引,切片,连接,变异操作 torch.stack [source] torch.unbind torch.unsqueeze 随机抽样 Random sampling torch.manual_seed torch.initial_seed torch.get_rng_...
  • Improving Deep Neural Networks: Hyperparameter tun

    Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization This is the second course of the deep learning specialization at Coursera which is moderat...
  • OneFlow 系统设计

    OneFlow系统设计 一、OneFlow 的设计目标 二、Actor:一套简洁的机制解决几乎所有技术难题 1. Actor 机制实现去中心化调度 2. Actor 机制实现流水线 3. 数据搬运是一等公民 4. 尽可能并行 三、OneFlow 如何做到分布式最易用 Placement SBP 最易用的分布式并行框架 四、总结 O...
  • basic_train

    basic_train Basic training functionality class Learner [source] [test] Model fitting methods lr_find [source] [test] fit [source] [test] fit_one_cycle [source] [test] ...
  • vision.image

    vision.image The fastai Image classes The Image classes class Image [source] [test] open_image [source] [test] Image.show [source] [test] class ImageSegment [source] ...
  • Jobs

    Jobs Running an example Job Writing a Job spec Job Labels Pod Template Pod selector Parallel execution for Jobs Controlling parallelism Completion mode Handling Pod and con...