Skip to content

Add foreachWithName to TensorTree #87

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
benikm91 wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Add foreachWithName to TensorTree #87

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
41 changes: 40 additions & 1 deletion core/src/main/scala/dimwit/autodiff/TensorTree.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,17 +3,31 @@ package dimwit.autodiff
import dimwit.tensor.*
import scala.deriving.*
import scala.compiletime.*
import scala.quoted.*

// TODO hot fix with retag and context parameter... maybe this can be improved?

trait TensorTree[P]:
def foreach(p: P, consumer: TensorTree.TensorConsumerWithPath, path: String = "", pathSeparator: String = TensorTree.DEFAULT_PATH_SEPARATOR): Unit
def fill(provider: TensorTree.TensorProvider, path: String = "", pathSeparator: String = TensorTree.DEFAULT_PATH_SEPARATOR): P
def map(p: P, f: [T <: Tuple, V] => (Labels[T]) ?=> (Tensor[T, V] => Tensor[T, V])): P
def zipMap(p1: P, p2: P, f: [T <: Tuple, V] => (Labels[T]) ?=> ((Tensor[T, V], Tensor[T, V]) => Tensor[T, V])): P

object TensorTree extends TensorTreeLowPriority:

val DEFAULT_PATH_SEPARATOR = "."

type TensorConsumerWithPath = [T <: Tuple, V] => (Labels[T]) ?=> (String, Tensor[T, V]) => Unit
type TensorProvider = [T <: Tuple, V] => (Labels[T]) ?=> String => Tensor[T, V]

def apply[P](using pt: TensorTree[P]): TensorTree[P] = pt

given [Q <: Tuple, V](using n: Labels[Q]): TensorTree[Tensor[Q, V]] with
def foreach(t: Tensor[Q, V], f: [T <: Tuple, V2] => (Labels[T]) ?=> (String, Tensor[T, V2]) => Unit, path: String, pathSeparator: String): Unit =
f[Q, V](using n)(path, t)

def fill(provider: TensorProvider, path: String, pathSeparator: String): Tensor[Q, V] = provider[Q, V](using n)(path)

def map(t: Tensor[Q, V], f: [T <: Tuple, V2] => (Labels[T]) ?=> (Tensor[T, V2] => Tensor[T, V2])): Tensor[Q, V] =
import TensorOps.retag
f[Q, V](using n)(t.retag[Q](using n))
Expand All @@ -25,12 +39,29 @@ object TensorTree extends TensorTreeLowPriority:
inline given derived[P <: Product](using m: Mirror.ProductOf[P]): TensorTree[P] =
val elemInstances = summonAll[Tuple.Map[m.MirroredElemTypes, TensorTree]]
val instances = elemInstances.toList.asInstanceOf[List[TensorTree[Any]]]
derivedImpl(instances, m)
val fieldNames = constValueTuple[m.MirroredElemLabels].toList.map(_.toString)
derivedImpl(instances, fieldNames, m)

private def derivedImpl[P <: Product](
instances: List[TensorTree[Any]],
fieldNames: List[String],
m: Mirror.ProductOf[P]
): TensorTree[P] = new TensorTree[P]:
def foreach(p: P, consumer: TensorConsumerWithPath, path: String, pathSeparator: String): Unit =
val inputs = p.productIterator.toList
inputs.zip(instances).zip(fieldNames).foreach:
case ((elem, inst), fieldName) =>
// Construct the nested path
val nextName = if path.isEmpty then fieldName else s"$path$pathSeparator$fieldName"
inst.foreach(elem, consumer, nextName, pathSeparator)

def fill(provider: TensorProvider, path: String, pathSeparator: String): P =
val generatedElems = instances.zip(fieldNames).map:
case (inst, fieldName) =>
val newPath = if path.isEmpty then fieldName else s"$path$pathSeparator$fieldName"
inst.fill(provider, newPath, pathSeparator)
m.fromProduct(Tuple.fromArray(generatedElems.map(_.asInstanceOf[Object]).toArray))

def map(p: P, f: [T <: Tuple, V] => (Labels[T]) ?=> (Tensor[T, V] => Tensor[T, V])): P =
val inputs = p.productIterator.toList
val mappedElems = inputs
Expand All @@ -49,7 +80,15 @@ object TensorTree extends TensorTreeLowPriority:
case ((e1, e2), inst) => inst.zipMap(e1, e2, f)
m.fromProduct(Tuple.fromArray(mappedElems.map(_.asInstanceOf[Object]).toArray))

def foreach[P](p: P, consumer: TensorConsumerWithPath, pathSeparator: String = DEFAULT_PATH_SEPARATOR)(using tt: TensorTree[P]): Unit =
tt.foreach(p, consumer, "", pathSeparator)

def fill[P](provider: TensorProvider, pathSeparator: String = DEFAULT_PATH_SEPARATOR)(using tt: TensorTree[P]): P =
tt.fill(provider, "", pathSeparator)

trait TensorTreeLowPriority:
given identity[A]: TensorTree[A] = new TensorTree[A]:
def foreach(p: A, consumer: TensorTree.TensorConsumerWithPath, path: String, pathSeparator: String): Unit = ()
def fill(provider: [T <: Tuple, V] => (x: Labels[T]) ?=> String => Tensor[T, V], path: String, pathSeparator: String): A = ???
def map(p: A, f: [T <: Tuple, V] => (Labels[T]) ?=> (Tensor[T, V] => Tensor[T, V])): A = p
def zipMap(p1: A, p2: A, f: [T <: Tuple, V] => (Labels[T]) ?=> ((Tensor[T, V], Tensor[T, V]) => Tensor[T, V])): A = p1
57 changes: 57 additions & 0 deletions core/src/test/scala/dimwit/autodiff/TensorTreeSuite.scala
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,6 +8,63 @@ import org.scalatest.matchers.should.Matchers

class TensorTreeSuite extends AnyFunSpec with Matchers:

describe("foreach"):
it("2-level case class"):
case class Child(
val numbers: Tensor1[A, Float],
val counts: Tensor1[A, Int]
)
case class Parent(
val child: Child,
val flags: Tensor1[A, Boolean]
)
val params = Parent(
Child(
Tensor1(Axis[A]).fromArray(Array(0.1f, 0.2f, 0.3f)),
Tensor1(Axis[A]).fromArray(Array(1, 2, 3))
),
Tensor1(Axis[A]).fromArray(Array(true, false, true))
)
TensorTree.foreach(
params,
[T <: Tuple, V] =>
(labels: Labels[T]) ?=>
(name: String, tensor: Tensor[T, V]) =>
name match
case "child-numbers" => tensor should equal(params.child.numbers)
case "child-counts" => tensor should equal(params.child.counts)
case "flags" => tensor should equal(params.flags)
case _ => fail(s"Unexpected name: $name"),
pathSeparator = "-"
)

describe("fill"):
it("2-level case class"):
case class Child(
val numbers: Tensor1[A, Float],
val counts: Tensor1[A, Int]
)
case class Parent(
val child: Child,
val flags: Tensor1[A, Boolean]
)
val expected = Parent(
Child(
Tensor1(Axis[A]).fromArray(Array(0.1f, 0.2f, 0.3f)),
Tensor1(Axis[A]).fromArray(Array(1, 2, 3))
),
Tensor1(Axis[A]).fromArray(Array(true, false, true))
)
val provider: TensorTree.TensorProvider = [T <: Tuple, V] =>
(labels: Labels[T]) ?=>
(name: String) =>
name match
case "child-numbers" => expected.child.numbers.asInstanceOf[Tensor[T, V]]
case "child-counts" => expected.child.counts.asInstanceOf[Tensor[T, V]]
case "flags" => expected.flags.asInstanceOf[Tensor[T, V]]
case _ => fail(s"Unexpected name: $name")
TensorTree.fill[Parent](provider, pathSeparator = "-") should equal(expected)

describe("map"):
it("1-level case class"):
case class Data(
Expand Down
Loading