Implement the composition of pow_grad

[YibinLiu666]

PR types

Others

PR changes

Others

Description

实现pow的一阶微分组合，测试代码为：

import paddle
import numpy as np

def test_pow_grad():
    paddle.framework.core.set_prim_eager_enabled(False)
    shape = [1024, 512]
    x = paddle.to_tensor(np.random.randn(*shape), dtype="float32")
    x.stop_gradient=False
    y_fused_res = paddle.pow(x, 2.7)
    dx_fused_res = paddle.grad(y_fused_res, x, create_graph=True, retain_graph=False)[0]
    paddle.framework.core.set_prim_eager_enabled(True)
    y_composite_res = paddle.pow(x, 2.7)
    dx_composite_res = paddle.grad(y_composite_res, x, create_graph=True, retain_graph=False)[0]
    np.testing.assert_allclose(dx_fused_res.numpy(), dx_composite_res.numpy(), 1e-6, 1e-6)

@HydrogenSulfate

[paddle-bot]

你的PR提交成功，感谢你对开源项目的贡献!
请关注后续CI自动化测试结果，详情请参考Paddle-CI手册。
Your PR has been submitted. Thanks for your contribution!
Please wait for the result of CI firstly. See Paddle CI Manual for details.

[HydrogenSulfate]

LGTM

@cyber-pioneer 麻烦卓哥也review一下

[Charles-hit]

因为这个拆解规则是动静统一的，一阶拆解的话加一个静态图测试吧

[YibinLiu666]

因为这个拆解规则是动静统一的，一阶拆解的话加一个静态图测试吧

已添加

[paddle-ci-bot]

Sorry to inform you that 3519a8e's CIs have passed for more than 7 days. To prevent PR conflicts, you need to re-run all CIs manually.

[HydrogenSulfate]

这行删除

[heavyrain-lzy]

LGTM for YAML