CANN/ops-nn LeakyReLU激活函数

aclnnLeakyReluaclnnInplaceLeakyRelu【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√功能说明接口功能激活函数用于解决Relu函数在输入小于0时输出为0的问题避免神经元无法更新参数。计算公式$$ out max(0,self) negativeSlope * min(0,self) $$函数原型aclnnLeakyRelu和aclnnInplaceLeakyRelu实现相同的功能使用区别如下请根据自身实际场景选择合适的算子。aclnnLeakyRelu需新建一个输出张量对象存储计算结果。aclnnInplaceLeakyRelu无需新建输出张量对象直接在输入张量的内存中存储计算结果。每个算子分为两段式接口必须先调用“aclnnLeakyReluGetWorkspaceSize”或者“aclnnInplaceLeakyReluGetWorkspaceSize”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“aclnnLeakyRelu”或者“aclnnInplaceLeakyRelu”接口执行计算。aclnnStatus aclnnLeakyReluGetWorkspaceSize( const aclTensor *self, const aclScalar *negativeSlope, aclTensor *out, uint64_t *workspaceSize, aclOpExecutor **executor)aclnnStatus aclnnLeakyRelu( void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)aclnnStatus aclnnInplaceLeakyReluGetWorkspaceSize( aclTensor *selfRef, const aclScalar *negativeSlope, uint64_t *workspaceSize, aclOpExecutor **executor)aclnnStatus aclnnInplaceLeakyRelu( void *workspace, uint64_t workspaceSize, aclOpExecutor *executor, aclrtStream stream)aclnnLeakyReluGetWorkspaceSize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TensorselfaclTensor*输入待进行LeakyRelu激活函数的入参公式中的self。shape支持0到8维shape需要与out一致。支持空Tensor。FLOAT、FLOAT16、BFLOAT16ND0-8√negativeSlopeaclScalar*输入表示self 0时的斜率公式中的negativeSlope。-FLOAT---outaclTensor*输出待进行LeakyRelu激活函数的出参。out的数据类型需要是self可转换的数据类型参见互转换关系。FLOAT、FLOAT16、BFLOAT16ND0-8√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。-----Atlas 训练系列产品 FLOAT、FLOAT16、DOUBLE。返回值aclnnStatus返回状态码具体参见aclnn返回码。第一段接口会完成入参校验出现以下场景时报错返回码错误码描述ACLNN_ERR_PARAM_NULLPTR161001传入的self、negativeSlope或out是空指针。ACLNN_ERR_PARAM_INVALID161002self的数据类型不在支持的范围之内。self的shape超过8维。out的数据类型不是self可转换的。out的shape与self不一致。aclnnLeakyRelu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aclnnLeakyReluGetWorkspaceSize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。aclnnInplaceLeakyReluGetWorkspaceSize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TensorselfRefaclTensor*输入输出即公式中的self与out。支持空Tensor。FLOAT、FLOAT16、BFLOAT16、DOUBLEND0-8√negativeSlopeaclScalar*输入表示self 0时的斜率公式中的negativeSlope。-FLOAT---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。-----Atlas 训练系列产品 数据类型支持FLOAT、FLOAT16、DOUBLE。返回值aclnnStatus返回状态码具体参见aclnn返回码。第一段接口会完成入参校验出现以下场景时报错返回码错误码描述ACLNN_ERR_PARAM_NULLPTR161001传入的selfRef、negativeSlope是空指针。ACLNN_ERR_PARAM_INVALID161002selfRef的数据类型不在支持的范围之内。aclnnInplaceLeakyRelu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aclnnInplaceLeakyReluGetWorkspaceSize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aclnnLeakyReluaclnnInplaceLeakyRelu默认确定性实现。negativeSlope使用整型类型作为属性输入而输入self是FLOAT类型那么如果negativeSlope大于2^24或小于-2^24可能存在精度损失。同理如果输入self是FLOAT16类型那么negativeSlope大于2^11或小于-2^11可能存在精度损失。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。aclnnLeakyRelu示例代码#include iostream #include vector #include acl/acl.h #include aclnnop/aclnn_leaky_relu.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 // 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 selfShape {4}; std::vectorint64_t outShape {4}; void* selfDeviceAddr nullptr; void* outDeviceAddr nullptr; aclTensor* self nullptr; aclScalar* negativeSlope nullptr; aclTensor* out nullptr; std::vectorfloat selfHostData {1, 2, 3, 4}; std::vectorfloat outHostData {0, 0, 0, 0}; float negativeSlopeValue 0.01f; // 创建self aclTensor ret CreateAclTensor(selfHostData, selfShape, selfDeviceAddr, aclDataType::ACL_FLOAT, self); CHECK_RET(ret ACL_SUCCESS, return ret); // 创建negativeSlope aclScalar negativeSlope aclCreateScalar(negativeSlopeValue, aclDataType::ACL_FLOAT); CHECK_RET(negativeSlope ! nullptr, LOG_PRINT(negativeSlope is null!\n); return false); // 创建out aclTensor ret CreateAclTensor(outHostData, outShape, outDeviceAddr, aclDataType::ACL_FLOAT, out); CHECK_RET(ret ACL_SUCCESS, return ret); // 3. 调用CANN算子库API需要修改为具体的API名称 uint64_t workspaceSize 0; aclOpExecutor* executor; // 调用aclnnLeakyRelu第一段接口 ret aclnnLeakyReluGetWorkspaceSize(self, negativeSlope, out, workspaceSize, executor); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnLeakyReluGetWorkspaceSize 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aclnnLeakyRelu第二段接口 ret aclnnLeakyRelu(workspaceAddr, workspaceSize, executor, stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnLeakyRelu 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(outShape); std::vectorfloat resultData(size, 0); ret aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), outDeviceAddr, size * sizeof(resultData[0]), ACL_MEMCPY_DEVICE_TO_HOST); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(copy result from device to host failed. ERROR: %d\n, ret); return ret); for (int64_t i 0; i size; i) { LOG_PRINT(result[%ld] is: %f\n, i, resultData[i]); } // 6. 释放aclTensor和aclScalar需要根据具体API的接口定义修改 aclDestroyTensor(self); aclDestroyScalar(negativeSlope); aclDestroyTensor(out); // 7. 释放device资源需要根据具体API的接口定义修改 aclrtFree(selfDeviceAddr); aclrtFree(outDeviceAddr); if (workspaceSize 0) { aclrtFree(workspaceAddr); } aclrtDestroyStream(stream); aclrtResetDevice(deviceId); aclFinalize(); return 0; }aclnnInplaceLeakyRelu示例代码#include iostream #include vector #include acl/acl.h #include aclnnop/aclnn_leaky_relu.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 // 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 selfRefShape {4}; void* selfRefDeviceAddr nullptr; aclTensor* selfRef nullptr; aclScalar* negativeSlope nullptr; std::vectorfloat selfRefHostData {1, 2, 3, 4}; float negativeSlopeValue 0.01f; // 创建selfRef aclTensor ret CreateAclTensor(selfRefHostData, selfRefShape, selfRefDeviceAddr, aclDataType::ACL_FLOAT, selfRef); CHECK_RET(ret ACL_SUCCESS, return ret); // 创建negativeSlope aclScalar negativeSlope aclCreateScalar(negativeSlopeValue, aclDataType::ACL_FLOAT); CHECK_RET(negativeSlope ! nullptr, LOG_PRINT(negativeSlope is null!\n); return false); // 3. 调用CANN算子库API需要修改为具体的API名称 uint64_t workspaceSize 0; aclOpExecutor* executor; // 调用aclnnInplaceLeakyRelu第一段接口 ret aclnnInplaceLeakyReluGetWorkspaceSize(selfRef, negativeSlope, workspaceSize, executor); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnInplaceLeakyReluGetWorkspaceSize 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aclnnInplaceLeakyRelu第二段接口 ret aclnnInplaceLeakyRelu(workspaceAddr, workspaceSize, executor, stream); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(aclnnInplaceLeakyRelu 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(selfRefShape); std::vectorfloat resultData(size, 0); ret aclrtMemcpy(resultData.data(), resultData.size() * sizeof(resultData[0]), selfRefDeviceAddr, size * sizeof(resultData[0]), ACL_MEMCPY_DEVICE_TO_HOST); CHECK_RET(ret ACL_SUCCESS, LOG_PRINT(copy result from device to host failed. ERROR: %d\n, ret); return ret); for (int64_t i 0; i size; i) { LOG_PRINT(result[%ld] is: %f\n, i, resultData[i]); } // 6. 释放aclTensor和aclScalar需要根据具体API的接口定义修改 aclDestroyTensor(selfRef); aclDestroyScalar(negativeSlope); // 7. 释放device资源需要根据具体API的接口定义修改 aclrtFree(selfRefDeviceAddr); 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