forked from huawei/mindspore2022
add realease note for flclient
This commit is contained in:
parent
101abbb969
commit
1bc7c0521f
|
|
@ -51,6 +51,8 @@
|
|||
|
||||
#### Federated Learning
|
||||
|
||||
- [STABLE] Change the deployment environment of FL-Client.
|
||||
|
||||
#### Running Data Recorder
|
||||
|
||||
- [STABLE] RDR saves collected data files within directories named by Rank ID on distributed training on Ascend, GPU and CPU.
|
||||
|
|
@ -59,9 +61,9 @@
|
|||
|
||||
#### Boost
|
||||
|
||||
* [BETA] Add LessBN algorithm, achieves 1.14x faster training throughput while maintaining negligible or no impact on the accuracy in our benchmark.
|
||||
* [BETA] Add gradient frozen algorithm, achieves 1.1x faster training throughput while maintaining negligible on the accuracy in our benchmark.
|
||||
* [BETA] Add "boost_level" input parameter in the Model interface to control boost level, where you can choose O1/O2 for 1.15x/1.2x faster.
|
||||
- [BETA] Add LessBN algorithm, achieves 1.14x faster training throughput while maintaining negligible or no impact on the accuracy in our benchmark.
|
||||
- [BETA] Add gradient frozen algorithm, achieves 1.1x faster training throughput while maintaining negligible on the accuracy in our benchmark.
|
||||
- [BETA] Add "boost_level" input parameter in the Model interface to control boost level, where you can choose O1/O2 for 1.15x/1.2x faster.
|
||||
|
||||
### API Change
|
||||
|
||||
|
|
|
|||
Loading…
Reference in New Issue