Created by: xiegegege

1. human_pose_estimation在paddle1.4分支下运行val和test时会报如下的错误： Traceback (most recent call last): File "val.py", line 233, in valid(args) File "val.py", line 127, in valid loss_name=loss.name) File "/opt/_internal/cpython-3.6.0/lib/python3.6/site-packages/paddle/fluid/parallel_executor.py", line 134, in init self._compiled_program._compile(place=self._place, scope=self._scope) File "/opt/_internal/cpython-3.6.0/lib/python3.6/site-packages/paddle/fluid/compiler.py", line 282, in _compile scope=self._scope) File "/opt/_internal/cpython-3.6.0/lib/python3.6/site-packages/paddle/fluid/compiler.py", line 253, in _compile_data_parallel self._exec_strategy, self._build_strategy, self._graph) paddle.fluid.core.EnforceNotMet: Illegal Pass. Generated graph shouldn't has cycle. at [/ssd1/xiege/paddle_ce/Paddle/paddle/fluid/framework/ir/pass.cc:35] 在paddle 1.3运行正常

2.在cpu下运行val.py会在运行22个epoch后报如下的错误： Traceback (most recent call last): File "val.py", line 233, in valid(args) File "val.py", line 160, in valid feed=feeder.feed(data)) File "/opt/internal/cpython-3.6.0/lib/python3.6/site-packages/paddle/fluid/parallel_executor.py", line 278, in run feed_tensor_dict) paddle.fluid.core.EnforceNotMet: Enforce failed. Expected member->places_.size() == lod_tensors.size(), but received member_->places_.size():26 != lod_tensors.size():14. The number of samples of current batch is less than the count of devices, currently, it is not allowed. (26 vs 14) at [/paddle/paddle/fluid/framework/parallel_executor.cc:452]