提交 bd753a8d 编写于 作者: C caifubi

change dump doc

上级 7a83a0e6
......@@ -242,20 +242,15 @@ When the training result deviates from the expectation on Ascend, the input and
"op_debug_mode": 0,
"iteration": 0,
"kernels": ["Default/Conv2D-op2", "Default/TensorAdd-op10"]
},
"DumpSettingsSpec": {
"net_name": "net name eg:ResNet50",
"dump_mode": "0: dump all kernels, 1: dump kernels in kernels list",
"op_debug_mode": "0: close debug, 1: debug ai-core overflow, 2: debug atomic overflow, 3: debug all overflow",
"iteration": "specified iteration",
"kernels": "op's full scope name which need to be dump"
}
}
```
> - Iteration should be set to 0 when `dataset_sink_mode` is False and data of every iteration will be dumped.
> - Iteration should increase by 1 when `dataset_sink_mode` is True. For example, data of GetNext will be dumped in iteration 0 and data of compute graph will be dumped in iteration 1.
> - `net_name`: net name eg:ResNet50.
> - `dump_mode`: 0: dump all kernels, 1: dump kernels in kernels list.
> - `op_debug_mode`: please set to 0.
> - `iteration`: specified iteration to dump. `iteration` should be set to 0 when `dataset_sink_mode` is False and data of every iteration will be dumped.
> - `kernels`: `fullname_with_scope` of kernel which need to dump.
5. Set environment variables.
......
......@@ -247,20 +247,15 @@ val:[[1 1]
"op_debug_mode": 0,
"iteration": 0,
"kernels": ["Default/Conv2D-op2", "Default/TensorAdd-op10"]
},
"DumpSettingsSpec": {
"net_name": "net name eg:ResNet50",
"dump_mode": "0: dump all kernels, 1: dump kernels in kernels list",
"op_debug_mode": "0: close debug, 1: debug ai-core overflow, 2: debug atomic overflow, 3: debug all overflow",
"iteration": "specified iteration",
"kernels": "op's full scope name which need to be dump"
}
}
```
> - 非数据下沉模式下,iteration需要设置成0,并且会Dump出每个iteration的数据。
> - 数据下沉模式iteration需要增加1。例如json中"iteration":0会Dump出GetNext算子的数据,而"iteration":1才会去Dump真正的计算图的第0个iteration数据。
> - `net_name`:自定义的网络名称,例如:"Resnet50"。
> - `dump_mode`:设置成0,表示Dump所有的算子;设置成1,表示Dump`"kernel"`里面制定的算子。
> - `op_debug_mode`:该属性用于算子溢出调试,在使用Dump功能的时候,请设置成0。
> - `iteration`:指定需要Dump的迭代。非数据下沉模式下,`iteration`需要设置成0,并且会Dump出每个迭代的数据。
> - `kernels`:指定需要Dump的算子名称(`fullname_with_scope`)。
5. 设置数据Dump的环境变量。
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册