@ExportPackage @PublicApi
Package com.yahoo.tensor.functions
Tensor function evaluation
- Author:
- bratseth
-
Interface Summary Interface Description ScalarFunction<NAMETYPE extends Name> A function which returns a scalarToStringContext A context which is passed down to all nested functions when returning a string representation. -
Class Summary Class Description Argmax<NAMETYPE extends Name> Argmin<NAMETYPE extends Name> CompositeTensorFunction<NAMETYPE extends Name> A composite tensor function is a tensor function which can be expressed (less tersely) as a tree of primitive tensor functions.Concat<NAMETYPE extends Name> Concatenation of two tensors along an (indexed) dimensionConstantTensor<NAMETYPE extends Name> A function which returns a constant tensor.Diag<NAMETYPE extends Name> A tensor generator which returns a tensor of any dimension filled with 1 in the diagonal and 0 elsewhere.DynamicTensor<NAMETYPE extends Name> A function which is a tensor whose values are computed by individual lambda functions on evaluation.Generate<NAMETYPE extends Name> An indexed tensor whose values are generated by a functionJoin<NAMETYPE extends Name> The join tensor operation produces a tensor from the argument tensors containing the set of cells given by the cross product of the cells of the given tensors, having as values the value produced by applying the given combinator function on the values from the two source cells.L1Normalize<NAMETYPE extends Name> L2Normalize<NAMETYPE extends Name> Map<NAMETYPE extends Name> The map tensor function produces a tensor where the given function is applied on each cell value.Matmul<NAMETYPE extends Name> Merge<NAMETYPE extends Name> The merge tensor operation produces from two argument tensors having equal types a tensor having the same type where the values are the union of the values of both tensors.PrimitiveTensorFunction<NAMETYPE extends Name> A primitive tensor function is a tensor function which cannot be expressed in terms of other tensor functions.Random<NAMETYPE extends Name> A tensor generator which returns a tensor of any dimension filled with random numbers between 0 and 1.Range<NAMETYPE extends Name> A tensor generator which returns a tensor of any dimension filled with the sum of the tensor indexes of each position.Reduce<NAMETYPE extends Name> The reduce tensor operation returns a tensor produced from the argument tensor where some dimensions are collapsed to a single value using an aggregator function.ReduceJoin<NAMETYPE extends Name> An optimization for tensor expressions where a join immediately follows a reduce.Rename<NAMETYPE extends Name> The rename tensor function returns a tensor where some dimensions are assigned new names.ScalarFunctions Factory of scalar Java functions.ScalarFunctions.Abs ScalarFunctions.Acos ScalarFunctions.Add ScalarFunctions.Asin ScalarFunctions.Atan ScalarFunctions.Ceil ScalarFunctions.Cos ScalarFunctions.Divide ScalarFunctions.Elu ScalarFunctions.Equal ScalarFunctions.EqualElements ScalarFunctions.Exp ScalarFunctions.Floor ScalarFunctions.Greater ScalarFunctions.LeakyRelu ScalarFunctions.Less ScalarFunctions.Log ScalarFunctions.Max ScalarFunctions.Mean ScalarFunctions.Min ScalarFunctions.Multiply ScalarFunctions.Neg ScalarFunctions.Pow ScalarFunctions.Random ScalarFunctions.Reciprocal ScalarFunctions.Relu ScalarFunctions.Rsqrt ScalarFunctions.Selu ScalarFunctions.Sigmoid ScalarFunctions.Sin ScalarFunctions.Sqrt ScalarFunctions.Square ScalarFunctions.SquaredDifference ScalarFunctions.Subtract ScalarFunctions.SumElements ScalarFunctions.Tan ScalarFunctions.Tanh Slice<NAMETYPE extends Name> Returns a subspace of a tensorSlice.DimensionValue<NAMETYPE extends Name> Softmax<NAMETYPE extends Name> TensorFunction<NAMETYPE extends Name> A representation of a tensor function which is able to be translated to a set of primitive tensor functions if necessary.ToStringContext.EmptyStringContext XwPlusB<NAMETYPE extends Name> -
Enum Summary Enum Description Reduce.Aggregator