mindspore.experimental.optim.AdamW

class mindspore.experimental.optim.AdamW(params, lr=1e-3, betas=(0.9, 0.999), eps=1e-8, weight_decay=1e-2, amsgrad=False, *, maximize=False)

Implements Adam Weight Decay algorithm.

$$\begin{array}{r}\begin{array}{rl}& \mathbf{\text{input}}:\gamma \text{(lr)},{\beta }_1,{\beta }_2\text{(betas)},{\theta }_0\text{(params)},f(\theta )\text{(objective)},ϵ\text{ (epsilon)}\\ & \lambda \text{(weight decay)},\mathit{\text{amsgrad}},\mathit{\text{maximize}}\\ & \mathbf{\text{initialize}}:m_0\leftarrow 0\text{ (first moment)},v_0\leftarrow 0\text{ ( second moment)},{\widehat{v_0}}^{max}\leftarrow 0\\ & \mathbf{\text{for}}t=1\mathbf{\text{to}}\dots \mathbf{\text{do}}\\ & \mathbf{\text{if}}\mathit{\text{maximize}}:\\ & g_t\leftarrow -{\mathrm{\nabla }}_{\theta }{f}_t({\theta }_{t-1})\\ & \mathbf{\text{else}}\\ & g_t\leftarrow {\mathrm{\nabla }}_{\theta }{f}_t({\theta }_{t-1})\\ & {\theta }_t\leftarrow {\theta }_{t-1}-\gamma \lambda {\theta }_{t-1}\\ & m_t\leftarrow {\beta }_1{m}_{t-1}+(1-{\beta }_1)g_t\\ & v_t\leftarrow {\beta }_2{v}_{t-1}+(1-{\beta }_2)g_t^2\\ & \widehat{m_t}\leftarrow m_t/(1-{\beta }_1^t)\\ & \widehat{v_t}\leftarrow v_t/(1-{\beta }_2^t)\\ & \mathbf{\text{if}}amsgrad\\ & {\widehat{v_t}}^{max}\leftarrow \max ({\widehat{v_t}}^{max},\widehat{v_t})\\ & {\theta }_t\leftarrow {\theta }_t-\gamma \widehat{m_t}/(\sqrt{{\widehat{v_t}}^{max}}+ϵ)\\ & \mathbf{\text{else}}\\ & {\theta }_t\leftarrow {\theta }_t-\gamma \widehat{m_t}/(\sqrt{\widehat{v_t}}+ϵ)\\ & \mathbf{r}\mathbf{e}\mathbf{t}\mathbf{u}\mathbf{r}\mathbf{n}{\theta }_{\mathbf{t}}\end{array}\end{array}$$

Warning

This is an experimental optimizer API that is subject to change. This module must be used with lr scheduler module in LRScheduler Class .

Parameters

• params (Union[list(Parameter), list(dict)]) – list of parameters to optimize or dicts defining parameter groups

• lr (Union[int, float, Tensor], optional) – learning rate. Default: 1e-3.

• betas (Tuple[float, float], optional) – The exponential decay rate for the moment estimations. Default: (0.9, 0.999).

• eps (float, optional) – term added to the denominator to improve numerical stability. Default: 1e-8.

• weight_decay (float, optional) – weight decay (L2 penalty). Default: 0..

• amsgrad (bool, optional) – whether to use the AMSGrad algorithm. Default: False.

Keyword Arguments

maximize (bool, optional) – maximize the params based on the objective, instead of minimizing. Default: False.

Inputs:

• gradients (tuple[Tensor]) - The gradients of params.

Raises

• ValueError – If the learning rate is not int, float or Tensor.

• ValueError – If the learning rate is less than 0.

• ValueError – If the eps is less than 0.0.

• ValueError – If the betas not in the range of [0, 1).

• ValueError – If the weight_decay is less than 0.

Supported Platforms:

Ascend GPU CPU

Examples

>>> import mindspore
>>> from mindspore import nn
>>> from mindspore.experimental import optim
>>> # Define the network structure of LeNet5. Refer to
>>> # https://gitee.com/mindspore/docs/blob/r2.5.0/docs/mindspore/code/lenet.py
>>> net = LeNet5()
>>> loss_fn = nn.SoftmaxCrossEntropyWithLogits(sparse=True)
>>> optimizer = optim.AdamW(net.trainable_params(), lr=0.1)
>>> def forward_fn(data, label):
...     logits = net(data)
...     loss = loss_fn(logits, label)
...     return loss, logits
>>> grad_fn = mindspore.value_and_grad(forward_fn, None, optimizer.parameters, has_aux=True)
>>> def train_step(data, label):
...     (loss, _), grads = grad_fn(data, label)
...     optimizer(grads)
...     return loss