InnerProductAlso known as a “fully-connected” layer, computes an inner product with a set of learned weights, and (optionally) adds biases.Parameters:int M_ int K_ int N_ bool bias_term_ Blob Dtype bias_multiplier_ bool transpose_ if true, assume transposed weightsTileParameters:unsigned int axis_ unsigned int tiles_ unsigned int outer_dim_ unsigned int inner_dim_Reshapevector int copy_axes_ vector of axes indices whose dimensions well copy from the bottom int inferred_axis_ the index of the axis whose dimension we infer, or -1 if none int constant_count_ the product of the constant output dimensions参考地址1.http://caffe.berkeleyvision.org/doxygen/annotated.html
 手动分配内存和拷贝（NVIDIA 课程 Part five）)
