10 Commits (2b259bf14d396a1aaed9cf3ea519a1c07989060f)

Author	SHA1	Message	Date
typhoonzero	903d5609c6	follow comment1	7 years ago
typhoonzero	641b4c0fe6	wip	7 years ago
typhoonzero	5361911c68	adam support sparse	7 years ago
typhoonzero	6cc4bd536f	wip	7 years ago
Yang Yu	1fdf8853d4	Optimize adam_op	7 years ago
Yang Yu	39ef5736c2	Use tranform to rewrite adam	7 years ago
QI JUN	61ec0b9516	Refine device context (#6433) ... There are mainly following fixes: - take `DeviceContext` as the template parameter of math functors and OpKernel instead of `Place` - remove `eigen_device` interface in base class `DeviceContext` - remove `GetEigenDevice` interface in `ExecutionContext` and base class `DeviceContext` - remove unused `platform::EigenDeviceConverter` - rename `REGISTER_OP_GPU_KERNEL` to `REGISTER_OP_CUDA_KERNEL` - rename `USE_GPU_ONLY_OP` to `USE_CUDA_ONLY_OP`	7 years ago
Yu Yang	a5e73f9eaf	Support many data types of several operators (#5731) ... * Support many data types of several operators * SeqConv only support float/double * Revert adagrad	7 years ago
Abhinav Arora	11bebeb2dc	Removing updates of Beta1 and Beta2 power accumulators outside the op (#4925)	7 years ago
Abhinav Arora	116800378a	Adding the Adam Optimizer operator (#4733) ... * add adam op moment1_out = beta1 * moment1 + (1 − beta1) * grad moment2_out = beta2 * moment2 + (1 − beta2) * grad * grad moment1_hat = moment1_out / (1 - beta1^t) moment2_hat = moment2_out / (1 - beta2^t) param_out = param - learning_rate * moment1_hat / (sqrt(moment2_hat) + epsilon) * fix moment 2 * Adding the Adam optimization operator * Adding more tests for Adam op	7 years ago