未验证 提交 81fe02c3 编写于 作者: C Chen Weihang 提交者: GitHub

Fix config description error in cuda_profiler function document (#18750)

* fix profiler doc error, test=develop

* update API.spec, test=develop
上级 311f90f1
...@@ -1015,7 +1015,7 @@ paddle.fluid.dygraph_grad_clip.GradClipByNorm ('paddle.fluid.dygraph_grad_clip.G ...@@ -1015,7 +1015,7 @@ paddle.fluid.dygraph_grad_clip.GradClipByNorm ('paddle.fluid.dygraph_grad_clip.G
paddle.fluid.dygraph_grad_clip.GradClipByNorm.__init__ (ArgSpec(args=['self', 'clip_norm'], varargs=None, keywords=None, defaults=None), ('document', '6adf97f83acf6453d4a6a4b1070f3754')) paddle.fluid.dygraph_grad_clip.GradClipByNorm.__init__ (ArgSpec(args=['self', 'clip_norm'], varargs=None, keywords=None, defaults=None), ('document', '6adf97f83acf6453d4a6a4b1070f3754'))
paddle.fluid.dygraph_grad_clip.GradClipByGlobalNorm ('paddle.fluid.dygraph_grad_clip.GradClipByGlobalNorm', ('document', 'd1872377e7d7a5fe0dd2e8c42e4c9656')) paddle.fluid.dygraph_grad_clip.GradClipByGlobalNorm ('paddle.fluid.dygraph_grad_clip.GradClipByGlobalNorm', ('document', 'd1872377e7d7a5fe0dd2e8c42e4c9656'))
paddle.fluid.dygraph_grad_clip.GradClipByGlobalNorm.__init__ (ArgSpec(args=['self', 'max_global_norm'], varargs=None, keywords=None, defaults=None), ('document', '6adf97f83acf6453d4a6a4b1070f3754')) paddle.fluid.dygraph_grad_clip.GradClipByGlobalNorm.__init__ (ArgSpec(args=['self', 'max_global_norm'], varargs=None, keywords=None, defaults=None), ('document', '6adf97f83acf6453d4a6a4b1070f3754'))
paddle.fluid.profiler.cuda_profiler (ArgSpec(args=['output_file', 'output_mode', 'config'], varargs=None, keywords=None, defaults=(None, None)), ('document', '49f5db5da13cfd8c069754dd11be3901')) paddle.fluid.profiler.cuda_profiler (ArgSpec(args=['output_file', 'output_mode', 'config'], varargs=None, keywords=None, defaults=(None, None)), ('document', '4053b45953807a24e28027dc86829d6c'))
paddle.fluid.profiler.reset_profiler (ArgSpec(args=[], varargs=None, keywords=None, defaults=None), ('document', 'fd1f25a7a06516ca9a1f4ab0783a4d70')) paddle.fluid.profiler.reset_profiler (ArgSpec(args=[], varargs=None, keywords=None, defaults=None), ('document', 'fd1f25a7a06516ca9a1f4ab0783a4d70'))
paddle.fluid.profiler.profiler (ArgSpec(args=['state', 'sorted_key', 'profile_path'], varargs=None, keywords=None, defaults=(None, '/tmp/profile')), ('document', 'a2be24e028dffa06ab28cc55a27c59e4')) paddle.fluid.profiler.profiler (ArgSpec(args=['state', 'sorted_key', 'profile_path'], varargs=None, keywords=None, defaults=(None, '/tmp/profile')), ('document', 'a2be24e028dffa06ab28cc55a27c59e4'))
paddle.fluid.profiler.start_profiler (ArgSpec(args=['state'], varargs=None, keywords=None, defaults=None), ('document', '4c192ea399e6e80b1ab47a8265b022a5')) paddle.fluid.profiler.start_profiler (ArgSpec(args=['state'], varargs=None, keywords=None, defaults=None), ('document', '4c192ea399e6e80b1ab47a8265b022a5'))
......
...@@ -43,7 +43,7 @@ def cuda_profiler(output_file, output_mode=None, config=None): ...@@ -43,7 +43,7 @@ def cuda_profiler(output_file, output_mode=None, config=None):
`output_file` with Key-Value pair format or Comma separated values format. `output_file` with Key-Value pair format or Comma separated values format.
The user can set the output mode by `output_mode` argument and set the The user can set the output mode by `output_mode` argument and set the
counters/options for profiling by `config` argument. The default config counters/options for profiling by `config` argument. The default config
is ['gpustarttimestamp', 'gpustarttimestamp', 'gridsize3d', is ['gpustarttimestamp', 'gpuendtimestamp', 'gridsize3d',
'threadblocksize', 'streamid', 'enableonstart 0', 'conckerneltrace']. 'threadblocksize', 'streamid', 'enableonstart 0', 'conckerneltrace'].
Then users can use NVIDIA Visual Profiler Then users can use NVIDIA Visual Profiler
(https://developer.nvidia.com/nvidia-visual-profiler) tools to load this (https://developer.nvidia.com/nvidia-visual-profiler) tools to load this
......
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