Paddle

Commit Graph

Author	SHA1	Message	Date
emailweixu	28bc05b126	Fix sparse training for trainer_count=1 (#204 ) * Fix sparse training for trainer_count=1 For trainer_count=1, the gradient machine is NeuralNetwork, which does not create parameter buf for PARAMETER_GRADIENT for sparse update in Parameter::enableType. But gradient parameter buf is still used in SgdThreadUpdater. * Minor update to comment	8 years ago
emailweixu	6decbdf4f5	Fix potential dead lock in PyDataProvider2 (#140 ) This bug occasionally causes dead lock in test_RecurrentGradientMachine In general, conditional_variable::notify should be used together with mutex for changing condition.	8 years ago
zhangjinchao01	b72beee4f0	fix dash and space bug, ISSUE=4586495 git-svn-id: https://svn.baidu.com/idl/trunk/paddle@1408 1ad973e4-5ce8-4261-8a94-b56d1f490c56	9 years ago

Author

SHA1

Message

Date

emailweixu

28bc05b126

Fix sparse training for trainer_count=1 (#204 )

* Fix sparse training for trainer_count=1

For trainer_count=1, the gradient machine is NeuralNetwork, which does not create parameter buf for PARAMETER_GRADIENT for sparse update in Parameter::enableType. But gradient parameter buf is still used in SgdThreadUpdater.

* Minor update to comment

emailweixu

6decbdf4f5

Fix potential dead lock in PyDataProvider2 (#140 )

This bug occasionally causes dead lock in test_RecurrentGradientMachine
In general, conditional_variable::notify should be used together with mutex for changing condition.

zhangjinchao01

b72beee4f0

fix dash and space bug,

ISSUE=4586495

git-svn-id: https://svn.baidu.com/idl/trunk/paddle@1408 1ad973e4-5ce8-4261-8a94-b56d1f490c56

3 Commits (e05f4ff26700dd34aa6d3c6da7061c62c5fa39c9)