private ParallelMultiCountTableBuilder(IHostEnvironment env, MultiCountTable table, DataViewSchema.Column[] inputCols, long labelCardinality)
{
Contracts.AssertValue(env, nameof(env));
env.AssertValue(table, nameof(table));
_host = env.Register(RegistrationName);
var n = table.ColCount;
_host.Check(Utils.Size(inputCols) == n, "Inconsistent number of columns");
_countTableBuilders = new InternalCountTableBuilderBase[n][];
var slotCounts = table.SlotCount;
for (int i = 0; i < _countTableBuilders.Length; i++)
{
var size = slotCounts[i];
_host.Assert(size > 0);
_host.Check(size == inputCols[i].Type.GetValueCount(), "Inconsistent number of slots");
_countTableBuilders[i] = new InternalCountTableBuilderBase[size];
for (int j = 0; j < size; j++)
{
_countTableBuilders[i][j] = ((CountTableBase)table[i, j]).ToBuilder(labelCardinality);
}
}
}
public ParallelMultiCountTableBuilder(IHostEnvironment env,
DataViewSchema.Column[] inputColumns,
CountTableBuilderBase[] builders,
long labelCardinality)
{
Contracts.CheckValue(env, nameof(env));
env.CheckValue(inputColumns, nameof(inputColumns));
_host = env.Register(RegistrationName);
var n = inputColumns.Length;
_countTableBuilders = new InternalCountTableBuilderBase[n][];
for (int i = 0; i < _countTableBuilders.Length; i++)
{
var col = inputColumns[i];
var size = col.Type.GetValueCount();
_host.Check(size > 0, "vectors of unknown length are not supported");
_countTableBuilders[i] = new InternalCountTableBuilderBase[size];
for (int j = 0; j < size; j++)
{
_countTableBuilders[i][j] = builders[i].GetInternalBuilder(labelCardinality);
}
}
}
public BagMultiCountTableBuilder(IHostEnvironment env, MultiCountTable table, DataViewSchema.Column[] inputCols, long labelCardinality)
{
Contracts.CheckValue(env, nameof(env));
env.CheckValue(table, nameof(table));
_host = env.Register(LoaderSignature);
_host.Check(Utils.Size(inputCols) == table.ColCount, "Inconsistent number of columns");
_host.Check(table.SlotCount.Zip(inputCols, (count, col) => (count, col)).
All(pair => pair.col.Type.GetValueCount() == pair.count), "Inconsistent number of slots");
_builder = table.BaseTable.ToBuilder(labelCardinality);
_colCount = table.ColCount;
_slotCount = new int[_colCount];
table.SlotCount.CopyTo(_slotCount, 0);
}
public BagMultiCountTableBuilder(IHostEnvironment env, DataViewSchema.Column[] inputColumns, CountTableBuilderBase builder, long labelCardinality)
{
Contracts.CheckValue(env, nameof(env));
env.CheckNonEmpty(inputColumns, nameof(inputColumns));
_host = env.Register(LoaderSignature);
// REVIEW: how to disallow non-zero garbage bin for bag dict count table? Or maybe just ignore?
_builder = builder.GetInternalBuilder(labelCardinality);
_colCount = inputColumns.Length;
_slotCount = new int[_colCount];
for (int i = 0; i < _colCount; i++)
{
_slotCount[i] = inputColumns[i].Type.GetValueCount();
}
}
