Eric Shaw

Update keras_optimization_helper.py

3

origin helper requires tensorflow cuda9.0, however, the hunter will output nothing and stop when working on cuda10. so add some information

[Bug] build wrong runtime tree relationship

1

 图中的runtime名称做了脱敏 A compute 调用了两个cudaruntime，分别是cudaRT_P和cudaLaunchKernel，其中cudaRT_P调用了cudaRT_C，但是 https://github.com/PaddlePaddle/VisualDL/blob/e420b8cace3001bcd77ae627d8b6ac6a23448526/visualdl/component/profiler/parser/event_node.py#L457 将三个cudaruntime 一起作为A compute的runtime子节点，即len(Acompute.runtime_node) == 3 对此，已经提交了pr，见 https://github.com/PaddlePaddle/VisualDL/pull/1256，希望参与问题的后续讨论，期待回复

[Bug] fix _build_tree_relationship in ProfilerResult

1

### bug描述： 当OP调用runtimeP，而runtime P调用runtime C时，两个runtime会成为兄弟节点 存放在 `OP.runtime_node` 中 ### 期望： runtime C应该成为runtime P的子runtime节点 ``` OP .runtime_node=[ runtime P .runtime_node=[ runtime C ] ] ``` ### 附件 附件中是一个脱敏的样例，`cudaRT_P` 调用了 `cudaRT_C`

Trying to backward through the graph a second time, but the buffers have already been freed. Specify retain_graph=True when calling backward the first time.

3

when updating the sub network, is there any need to retain graph like `loss.backward(retain_graph=True)` because when i reproduce the procedure, the code runs wrong, but i dont know if retaining...