Xiaoda Zhang
|
9f4b8a3cd1
|
changing the succive edges order in GetAliveSuccEdges() so that Triangle and Star Elimination can be merged into particular node; adding some check information
|
5 years ago |
leopz
|
40e15996b0
|
move default_param out of parameter and remove pybind11 in anf define
|
5 years ago |
yao_yf
|
f0bf438a55
|
reshape strategy search
|
5 years ago |
|
Xiaoda Zhang
|
def8573275
|
implementing-searching-strategy-for-inference
|
5 years ago |
ch-l
|
f806b72447
|
use DeviceMemory for memory control
|
5 years ago |
yangzhenzhang
|
6d522f0a4f
|
add parallel op for layernorm
|
5 years ago |
|
Xiaoda Zhang
|
0ac50a19f5
|
Model the memory cost in auto-parallel. It is calculated by the output of operators, plus the parameters. Additionally, modify the graph-operations in auto_parallel to include memory_cost.
|
5 years ago |
c00425699
|
d62f560b50
|
add_bool_type_check_in_comm_op
|
5 years ago |
buxue
|
5841fe010e
|
Support pow's second input could be tensor and fix bug in bprop of pow
|
5 years ago |
|
yangzhenzhang
|
b34c0e7a17
|
add parallel op for dropoutdomask
|
5 years ago |
|
c00425699
|
b413638f23
|
refactor OperatorCostPtr in OperatorInfo
|
5 years ago |
mindspore-ci-bot
|
2e6e94b2b6
|
!177 prelu operator support parallel on the channel
Merge pull request !177 from yao_yf/fix_auto_parallel_prelu
|
5 years ago |
|
yao_yf
|
b5e3fa9593
|
fix auto parallel prelu
|
5 years ago |
|
Xiaoda Zhang
|
a153fad874
|
This commit is to separate the computation cost and memory cost in auto_parallel. Some related memory correction is removed.
|
5 years ago |
|
c00425699
|
3bb48ffee1
|
use std::vector instead of std::list to promote performance for parallel module
|
5 years ago |
zhunaipan
|
930a1fb0a8
|
initial version
Signed-off-by: leonwanghui <leon.wanghui@huawei.com>
|
5 years ago |
