Medial Code Documentation
|
a block of data, containing several rows in sparse matrix This is useful for (streaming-sxtyle) algorithms that scans through rows of data examples include: SGD, GD, L-BFGS, kmeans More...
#include <data.h>
Public Member Functions | |
Row< IndexType, DType > | operator[] (size_t rowid) const |
get specific rows in the batch | |
size_t | MemCostBytes (void) const |
RowBlock | Slice (size_t begin, size_t end) const |
slice a RowBlock to get rows in [begin, end) | |
Data Fields | |
size_t | size |
batch size | |
const size_t * | offset |
array[size+1], row pointer to beginning of each rows | |
const DType * | label |
array[size] label of each instance | |
const real_t * | weight |
With weight: array[size] label of each instance, otherwise nullptr. | |
const uint64_t * | qid |
With qid: array[size] session id of each instance, otherwise nullptr. | |
const IndexType * | field |
field id | |
const IndexType * | index |
feature index | |
const DType * | value |
feature value, can be NULL, indicating all values are 1 | |
a block of data, containing several rows in sparse matrix This is useful for (streaming-sxtyle) algorithms that scans through rows of data examples include: SGD, GD, L-BFGS, kmeans
The size of batch is usually large enough so that parallelizing over the rows can give significant speedup
IndexType | type to store the index used in row batch |
DType | type to store the label and value used in row batch |
|
inline |
|
inline |
get specific rows in the batch
rowid | the rowid in that row |
|
inline |