CANN/ops-nn KL散度目标反向计算

aclnnKlDivTargetBackward【免费下载链接】ops-nn本项目是CANN提供的神经网络类计算算子库实现网络在NPU上加速计算。项目地址: https://gitcode.com/cann/ops-nn 查看源码产品支持情况产品是否支持Ascend 950PR/Ascend 950DT√Atlas A3 训练系列产品/Atlas A3 推理系列产品√Atlas A2 训练系列产品/Atlas A2 推理系列产品√Atlas 200I/500 A2 推理产品×Atlas 推理系列产品×Atlas 训练系列产品×功能说明算子功能进行aclnnKlDiv api的结果的target反向计算。计算公式$$ gradTarget \begin{cases} gradOutput * (log(target) - self 1), ~~~~~~~~~~~~~~~~~~~~~~~~~logTargetFalse \ gradOutput * exp(target) * (target - self 1), ~~~~~~~~~~logTargetTrue \end{cases} $$函数原型每个算子分为两段式接口必须先调用“aclnnKlDivTargetBackwardGetWorkspaceSize”接口获取计算所需workspace大小以及包含了算子计算流程的执行器再调用“aclnnKlDivTargetBackward”接口执行计算。aclnnStatus aclnnKlDivTargetBackwardGetWorkspaceSize( const aclTensor* gradOutput, const aclTensor* self, const aclTensor* target, int64_t reduction, bool logTarget, aclTensor* gradTarget, uint64_t* workspaceSize, aclOpExecutor** executor)aclnnStatus aclnnKlDivTargetBackward( void* workspace, uint64_t workspaceSize, aclOpExecutor* executor, aclrtStream stream)aclnnKlDivTargetBackwardGetWorkspaceSize参数说明参数名输入/输出描述使用说明数据类型数据格式维度(shape)非连续TensorgradOutputaclTensor*输入梯度反向输入。shape需要与self满足broadcast关系。FLOAT、FLOAT16、BFLOAT16ND1-8√selfaclTensor*输入输入张量。-FLOAT、FLOAT16、BFLOAT16ND1-8√targetaclTensor*输入真实的标签。shape需要与self满足broadcast关系。FLOAT、FLOAT16、BFLOAT16ND1-8√reductionint64_t输入指定要应用到输出的缩减。支持0(none)|1(mean)|2(sum)|3(batchmean)。none表示不应用缩减。mean表示输出的总和将除以输出中的元素数。sum表示输出将被求和。batchmean表示输出的总和将除以batch的个数。INT64---logTargetbool输入是否对target进行log空间转换。-BOOL--√gradTargetaclTensor*输出输出的损失。-与self保持一致ND1-8√workspaceSizeuint64_t*输出返回需要在Device侧申请的workspace大小。-----executoraclOpExecutor**输出返回op执行器包含了算子计算流程。-----返回值aclnnStatus返回状态码具体参见aclnn返回码。第一段接口完成入参校验出现以下场景时报错返回值错误码描述ACLNN_ERR_PARAM_NULLPTR161001传入的gradOutput、self、target和gradTarget是空指针。ACLNN_ERR_PARAM_INVALID161002gradOutput、self、target和gradTarget的数据类型不在支持的范围内时。target、gradTarget的数据类型不一致。gradOutput的shape不能向self或者target做broadcast。target的shape和self的shape不满足broadcast关系。target的shape与gradTarget的shape不相同。gradOutput、self、target或者gradTarget维度大于8。aclnnKlDivTargetBackward参数说明参数名输入/输出描述workspace输入在Device侧申请的workspace内存地址。workspaceSize输入在Device侧申请的workspace大小由第一段接口aclnnKlDivTargetBackwardGetWorkspaceSize获取。executor输入op执行器包含了算子计算流程。stream输入指定执行任务的Stream。返回值aclnnStatus返回状态码具体参见aclnn返回码。约束说明确定性计算aclnnKlDivTargetBackward默认确定性实现。调用示例示例代码如下仅供参考具体编译和执行过程请参考编译与运行样例。#include iostream #include vector #include acl/acl.h #include aclnnop/aclnn_kl_div_target_backward.h #define CHECK_RET(cond, return_expr) \ do { \ if (!(cond)) { \ return_expr; \ } \ } while (0) #define LOG_PRINT(message, ...) \ do { \ printf(message, ##__VA_ARGS__); \ } while (0) int64_t GetShapeSize(const std::vectorint64_t shape) { int64_t shapeSize 1; for (auto i : shape) { shapeSize * i; } return shapeSize; } int Init(int32_t deviceId, aclrtStream* stream) { // 固定写法资源初始化 auto ret aclInit(nullptr); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclInit failed. ERROR: %d\n, ret); return ret); ret aclrtSetDevice(deviceId); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclrtSetDevice failed. ERROR: %d\n, ret); return ret); ret aclrtCreateStream(stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclrtCreateStream failed. ERROR: %d\n, ret); return ret); return 0; } template typename T int CreateAclTensor(const std::vectorT hostData, const std::vectorint64_t shape, void** deviceAddr, aclDataType dataType, aclTensor** tensor) { auto size GetShapeSize(shape) * sizeof(T); // 调用aclrtMalloc申请device侧内存 auto ret aclrtMalloc(deviceAddr, size, ACL_MEM_MALLOC_HUGE_FIRST); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclrtMalloc failed. ERROR: %d\n, ret); return ret); // 调用aclrtMemcpy将host侧数据拷贝到device侧内存上 ret aclrtMemcpy(*deviceAddr, size, hostData.data(), size, ACL_MEMCPY_HOST_TO_DEVICE); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclrtMemcpy failed. ERROR: %d\n, ret); return ret); // 计算连续tensor的strides std::vectorint64_t strides(shape.size(), 1); for (int64_t i shape.size() - 2; i 0; i--) { strides[i] shape[i 1] * strides[i 1]; } // 调用aclCreateTensor接口创建aclTensor *tensor aclCreateTensor(shape.data(), shape.size(), dataType, strides.data(), 0, aclFormat::ACL_FORMAT_ND, shape.data(), shape.size(), *deviceAddr); return 0; } int main() { // 1. 固定写法device/stream初始化参考acl API手册 // 根据自己的实际device填写deviceId int32_t deviceId 0; aclrtStream stream; auto ret Init(deviceId, stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(Init acl failed. ERROR: %d\n, ret); return ret); // 2. 构造输入与输出需要根据API的接口自定义构造 std::vectorint64_t gradOutputShape {2, 2}; std::vectorint64_t selfShape {2, 2}; std::vectorint64_t targetShape {2, 2}; std::vectorint64_t gradTargetShape {2, 2}; void* gradOutputDeviceAddr nullptr; void* selfDeviceAddr nullptr; void* targetDeviceAddr nullptr; void* gradTargetDeviceAddr nullptr; aclTensor* gradOutput nullptr; aclTensor* self nullptr; aclTensor* target nullptr; aclTensor* gradTarget nullptr; std::vectorfloat gradOutputHostData {2, 3, 5, 8}; std::vectorfloat selfHostData {2, 3, 5, 8}; std::vectorfloat targetHostData {2, 3, 5, 8}; std::vectorfloat gradTargetHostData {2, 3, 5, 8}; int64_t reduction 0; bool logTarget false; // 创建gradOutput aclTensor ret CreateAclTensor(gradOutputHostData, gradOutputShape, gradOutputDeviceAddr, aclDataType::ACL_FLOAT, gradOutput); CHECK_RET(ret ACL_SUCCESS, return ret); // 创建self aclTensor ret CreateAclTensor(selfHostData, selfShape, selfDeviceAddr, aclDataType::ACL_FLOAT, self); CHECK_RET(ret ACL_SUCCESS, return ret); // 创建target aclTensor ret CreateAclTensor(targetHostData, targetShape, targetDeviceAddr, aclDataType::ACL_FLOAT, target); CHECK_RET(ret ACL_SUCCESS, return ret); // 创建gradTarget aclTensor ret CreateAclTensor(gradTargetHostData, gradTargetShape, gradTargetDeviceAddr, aclDataType::ACL_FLOAT, gradTarget); CHECK_RET(ret ACL_SUCCESS, return ret); // 3. 调用CANN算子库API uint64_t workspaceSize 0; aclOpExecutor* executor; // 调用aclnnKlDivTargetBackward第一段接口 ret aclnnKlDivTargetBackwardGetWorkspaceSize(gradOutput, self, target, reduction, logTarget, gradTarget, workspaceSize, executor); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnKlDivTargetBackwardGetWorkspaceSize failed. ERROR: %d\n, ret); return ret); // 根据第一段接口计算出的workspaceSize申请device内存 void* workspaceAddr nullptr; if (workspaceSize 0) { ret aclrtMalloc(workspaceAddr, workspaceSize, ACL_MEM_MALLOC_HUGE_FIRST); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(allocate workspace failed. ERROR: %d\n, ret); return ret); } // 调用aclnnKlDivTargetBackward第二段接口 ret aclnnKlDivTargetBackward(workspaceAddr, workspaceSize, executor, stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnKlDivTargetBackward failed. ERROR: %d\n, ret); return ret); // 4. 固定写法同步等待任务执行结束 ret aclrtSynchronizeStream(stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclrtSynchronizeStream failed. ERROR: %d\n, ret); return ret); // 5. 获取输出的值将device侧内存上的结果拷贝至host侧需要根据具体API的接口定义修改 auto size GetShapeSize(gradTargetShape); std::vectorfloat resultData(size, 0); ret aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), gradTargetDeviceAddr, size * sizeof(resultData[0]), ACL_MEMCPY_DEVICE_TO_HOST); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(copy resultData from device to host failed. ERROR: %d\n, ret); return ret); for (int64_t i 0; i size; i) { LOG_PRINT(resultData[%ld] is: %f\n, i, resultData[i]); } // 6. 释放aclTensor需要根据具体API的接口定义修改 aclDestroyTensor(gradOutput); aclDestroyTensor(self); aclDestroyTensor(target); aclDestroyTensor(gradTarget); // 7. 释放device资源需要根据具体API的接口定义修改 aclrtFree(gradOutputDeviceAddr); aclrtFree(selfDeviceAddr); aclrtFree(targetDeviceAddr); aclrtFree(gradTargetDeviceAddr); if (workspaceSize 0) { aclrtFree(workspaceAddr); } aclrtDestroyStream(stream); aclrtResetDevice(deviceId); aclFinalize(); return 0; }【免费下载链接】ops-nn本项目是CANN提供的神经网络类计算算子库实现网络在NPU上加速计算。项目地址: https://gitcode.com/cann/ops-nn创作声明：本文部分内容由AI辅助生成（AIGC），仅供参考