【Sparse】add new API/OP(csr->csr) of SparseTensor softmax

[zhwesky2010]

PR types

New features

PR changes

OPs

Describe

新增稀疏Tensor激活函数paddle.incubate.sparse.nn.Softmax 与 paddle.incubate.sparse.nn.functional.softmax

paddle.incubate.sparse.nn.Softmax

import paddle
import numpy as np
from paddle.fluid.framework import _test_eager_guard

paddle.seed(100)

with _test_eager_guard():
    mask = np.random.rand(3, 4) < 0.5
    np_x = np.random.rand(3, 4) * mask
    # [[0.         0.         0.96823406 0.19722934]
    #  [0.94373937 0.         0.02060066 0.71456372]
    #  [0.         0.         0.         0.98275049]]

    csr = paddle.to_tensor(np_x).to_sparse_csr()
    # Tensor(shape=[3, 4], dtype=paddle.float64, place=Place(gpu:0), stop_gradient=True, 
    #        crows=[0, 2, 5, 6], 
    #        cols=[2, 3, 0, 2, 3, 3], 
    #        values=[0.96823406, 0.19722934, 0.94373937, 0.02060066, 0.71456372,
    #                0.98275049])

    m = paddle.incubate.sparse.nn.Softmax()
    out = m(csr)
    # Tensor(shape=[3, 4], dtype=paddle.float64, place=Place(gpu:0), stop_gradient=True, 
    #        crows=[0, 2, 5, 6], 
    #        cols=[2, 3, 0, 2, 3, 3], 
    #        values=[0.68373820, 0.31626180, 0.45610887, 0.18119845, 0.36269269,
    #                1.        ])

paddle.incubate.sparse.nn.functional.softmax

import paddle
import numpy as np
from paddle.fluid.framework import _test_eager_guard

paddle.seed(100)

with _test_eager_guard():
    mask = np.random.rand(3, 4) < 0.5
    np_x = np.random.rand(3, 4) * mask
    # [[0.         0.         0.96823406 0.19722934]
    #  [0.94373937 0.         0.02060066 0.71456372]
    #  [0.         0.         0.         0.98275049]]

    csr = paddle.to_tensor(np_x).to_sparse_csr()
    # Tensor(shape=[3, 4], dtype=paddle.float64, place=Place(gpu:0), stop_gradient=True, 
    #        crows=[0, 2, 5, 6], 
    #        cols=[2, 3, 0, 2, 3, 3], 
    #        values=[0.96823406, 0.19722934, 0.94373937, 0.02060066, 0.71456372,
    #                0.98275049])

    out = paddle.incubate.sparse.nn.functional.softmax(csr)
    # Tensor(shape=[3, 4], dtype=paddle.float64, place=Place(gpu:0), stop_gradient=True, 
    #        crows=[0, 2, 5, 6], 
    #        cols=[2, 3, 0, 2, 3, 3], 
    #        values=[0.68373820, 0.31626180, 0.45610887, 0.18119845, 0.36269269,
    #                1.        ])

[paddle-bot-old]

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.

[zkh2016]

T exp = functor(x_values[row_first + idx] - row_max_val);
out_values[row_first + idx] = exp;
exp_sum += exp;

[zhwesky2010]

Done, thx~

[dingjiaweiww]

only support axis=-1 ?

[zhwesky2010]

supported写错了，应该是support，下个PR改下哈