Document feedback

Question document fragment

When a question document fragment contains a formula, it is displayed as a space.

Submission type

issue

It's a little complicated...

I'd like to ask someone.

Please select the submission type

Problem type

Specifications and Common Mistakes

- Specifications and Common Mistakes:

- Misspellings or punctuation mistakes,incorrect formulas, abnormal display.

- Incorrect links, empty cells, or wrong formats.

- Chinese characters in English context.

- Minor inconsistencies between the UI and descriptions.

- Low writing fluency that does not affect understanding.

- Incorrect version numbers, including software package names and version numbers on the UI.

Usability

- Usability:

- Incorrect or missing key steps.

- Missing main function descriptions, keyword explanation, necessary prerequisites, or precautions.

- Ambiguous descriptions, unclear reference, or contradictory context.

- Unclear logic, such as missing classifications, items, and steps.

Correctness

- Correctness:

- Technical principles, function descriptions, supported platforms, parameter types, or exceptions inconsistent with that of software implementation.

- Incorrect schematic or architecture diagrams.

- Incorrect commands or command parameters.

- Incorrect code.

- Commands inconsistent with the functions.

- Wrong screenshots.

- Sample code running error, or running results inconsistent with the expectation.

Risk Warnings

- Risk Warnings:

- Lack of risk warnings for operations that may damage the system or important data.

Content Compliance

- Content Compliance:

- Contents that may violate applicable laws and regulations or geo-cultural context-sensitive words and expressions.

- Copyright infringement.

Please select the type of question

Problem description

Describe the bug so that we can quickly locate the problem.

Submit

Document feedback

mindspore.nn.AdaSumByDeltaWeightWrapCell

class mindspore.nn.AdaSumByDeltaWeightWrapCell(optimizer)

Enable the adasum in "auto_parallel/semi_auto_parallel" mode. The implementation of the Adaptive Summation (AdaSum) algorithm is calculated based on the difference of weights before and after the updating of optimizer. See the paper AdaSum: Scaling Distributed Training with Adaptive Summation.

$$\begin{array}{r}\begin{array}{l}{w}_{t+1}=w_t-\alpha \cdot Adasum(g_1,g_2)\\ w_{t+1}=w_t-\alpha \cdot [(1-\frac{g_2^T\cdot g_1}{2\cdot {\Vert g_1\Vert }^2})\cdot g_1+(1-\frac{g_1^T\cdot g_2}{2\cdot {\Vert g_2\Vert }^2})\cdot g_2]\end{array}\end{array}$$

In this implementation, $g$ represents the weight difference before and after the updating of optimizer, and the subscripts represent different devices in the data parallel dimension.

Note

When using AdaSum, the number of traning cards needs to be a power of 2 and at least 16 cards are required. Currently, the optimizer sharding and pipeline parallel is not supported when using AdaSum. It is recommended to using AdaSumByDeltaWeightWrapCell in semi auto parallel/auto parallel mode. In data parallel mode, we recommend to using mindspore.boost to applying AdaSum.

Parameters

optimizer (Union[Cell]) – Optimizer for updating the weights. The construct function of the optimizer requires only one input.

Inputs:

• grads (Tuple(Tensor)) - Tuple of gradients, same with the input of passed optimizer.

Raises

• RuntimeError – If parallel_mode uses stand_alone mode, AdaSum only supports use in distributed scenarios.

• RuntimeError – If the optimizer parallel is used when using AdaSum.

• RuntimeError – If the pipeline parallel is used when using AdaSum.

• RuntimeError – If device_num is not a power of 2, or less than 16.

Supported Platforms:

Ascend GPU

Examples

>>> import mindspore as ms
>>> from mindspore import nn
>>> # Define the network structure of LeNet5. Refer to
>>> # https://gitee.com/mindspore/docs/blob/master/docs/mindspore/code/lenet.py
>>> net = LeNet5()
>>> optim = nn.AdaSumByDeltaWeightWrapCell(nn.Momentum(params=net.trainable_params(),
...                                                 learning_rate=0.1, momentum=0.9))
>>> loss = nn.SoftmaxCrossEntropyWithLogits()
>>> model = ms.train.Model(net, loss_fn=loss, optimizer=optim, metrics=None)