Created by: Lggggggx

您好，请问现在X2Paddle是暂时不支持onehot和LSTM操作吗？ 例如torch中的gather用到了onehot操作

class test(torch.nn.Module):
    def forward(self, batch_input, batch_slice):
        '''batch_input:(batch_size, time_step, hidden_dim)
        batch_slice:(batch_size, 2)
        '''
        shape_input = batch_input.size()
        batch_slice = batch_slice.long().unsqueeze(2).expand(-1, -1, shape_input[2])

        tensor_slice = torch.gather(batch_input, 1, batch_slice)

        return tensor_slice