You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

boosted_trees_ops.h 2.1 kB

5 years ago
4 years ago
5 years ago
5 years ago
5 years ago
5 years ago
5 years ago
5 years ago
5 years ago
5 years ago
12345678910111213141516171819202122232425262728293031323334353637383940414243444546474849505152535455565758596061626364
  1. /**
  2. * Copyright 2019 Huawei Technologies Co., Ltd
  3. *
  4. * Licensed under the Apache License, Version 2.0 (the "License");
  5. * you may not use this file except in compliance with the License.
  6. * You may obtain a copy of the License at
  7. *
  8. * http://www.apache.org/licenses/LICENSE-2.0
  9. *
  10. * Unless required by applicable law or agreed to in writing, software
  11. * distributed under the License is distributed on an "AS IS" BASIS,
  12. * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
  13. * See the License for the specific language governing permissions and
  14. * limitations under the License.
  15. */
  16. /*!
  17. * \file boosted_trees_ops.h
  18. * \brief
  19. */
  20. #ifndef OPS_BUILT_IN_OP_PROTO_INC_BOOSTED_TREES_OPS_H_
  21. #define OPS_BUILT_IN_OP_PROTO_INC_BOOSTED_TREES_OPS_H_
  22. #include "graph/operator_reg.h"
  23. namespace ge {
  24. /**
  25. *@brief Bucketizes each feature based on bucket boundaries . \n
  26. *@par Inputs:
  27. *Input "float_values" is a 1D tensor. Input "bucket_boundaries" is
  28. a list of 1D tensors. It's a dynamic input.
  29. * @li float_values: A list of rank 1 tensors each containing float
  30. values for a single feature.
  31. * @li bucket_boundaries: A list of rank 1 tensors each containing
  32. the bucket boundaries for a single feature . It's a dynamic input. \n
  33. *@par Attributes:
  34. *@li num_features: Number of features
  35. *@par Outputs:
  36. *@li y: A list of rank 1 tensors each containing the bucketized values for
  37. a single feature . \n
  38. *@attention Constraints:
  39. *BoostedTreesBucketize runs on the Ascend AI CPU, which delivers poor performance. \n
  40. *@par Third-party framework compatibility
  41. *Compatible with the TensorFlow operator BoostedTreesBucketize . \n
  42. *@par Restrictions:
  43. *Warning: THIS FUNCTION IS EXPERIMENTAL. Please do not use.
  44. */
  45. REG_OP(BoostedTreesBucketize)
  46. .DYNAMIC_INPUT(float_values, TensorType({DT_FLOAT}))
  47. .DYNAMIC_INPUT(bucket_boundaries, TensorType({DT_FLOAT}))
  48. .DYNAMIC_OUTPUT(y, TensorType({DT_INT32}))
  49. .REQUIRED_ATTR(num_features, Int)
  50. .OP_END_FACTORY_REG(BoostedTreesBucketize)
  51. } // namespace ge
  52. #endif // OPS_BUILT_IN_OP_PROTO_INC_BOOSTED_TREES_OPS_H_

图引擎模块(GE)是MindSpore的一个子模块,其代码由C++实现,位于前端模块ME和底层硬件之间,起到承接作用。图引擎模块以ME下发的图作为输入,然后进行一系列的深度图优化操作,最后输出一张可以在底层硬件上高效运行的图。GE针对昇腾AI处理器的硬件结构特点,做了特定的优化工作,以此来充分发挥出昇腾AI处理器的强大算力。在进行模型训练/推理时,GE会被自动调用而用户并不感知。GE主要由GE API和GE Core两部分组成,详细的架构图如下所示