TorchLean API

Docs HomeInstallationGuideExamplesGraphsTools

NN.Runtime.Autograd.Engine.LibTorch.Ops.Shape

CUDA Tape Operations: Shape and Reduction Nodes #

Reductions / views #

source
@[inline]
def Runtime.Autograd.LibTorch.Tape.sum {s : Spec.Shape} (t : Tape SomeBuffer) (xId : ℕ) :
Result (Tape SomeBuffer × ℕ)

Reduce-sum of all entries, producing a scalar.

Instances For
    source
    @[inline]
    def Runtime.Autograd.LibTorch.Tape.flatten {s : Spec.Shape} (t : Tape SomeBuffer) (xId : ℕ) :
    Result (Tape SomeBuffer × ℕ)

    Flatten s into a 1D vector of length Spec.Shape.size s.

    Instances For
      source
      @[inline]
      def Runtime.Autograd.LibTorch.Tape.reshape {s₁ s₂ : Spec.Shape} (t : Tape SomeBuffer) (xId : ℕ) (_h : s₁.size = s₂.size) :
      Result (Tape SomeBuffer × ℕ)

      Reshape a buffer while preserving number of elements.

      This is a no-copy view operation: it reuses the same contiguous buffer.

      Instances For
        source
        @[inline]
        def Runtime.Autograd.LibTorch.Tape.swapAdjacentAtDepth {s : Spec.Shape} (t : Tape SomeBuffer) (depth xId : ℕ) :
        Result (Tape SomeBuffer × ℕ)

        Swap adjacent axes at a given depth in an N-D buffer.

        If depth is out of range, this is treated as the identity (matches the spec-layer helper).

        Instances For
          source
          @[inline]
          def Runtime.Autograd.LibTorch.Tape.broadcastTo {s₁ s₂ : Spec.Shape} (t : Tape SomeBuffer) (cb : s₁.CanBroadcastTo s₂) (xId : ℕ) :
          Result (Tape SomeBuffer × ℕ)

          Broadcast x : s₁ to s₂.

          Forward: broadcastTo. Backward: sum-reduce broadcasted axes (reduceFromBroadcastTo).

          Instances For
            source
            @[inline]
            def Runtime.Autograd.LibTorch.Tape.reduceSum {s : Spec.Shape} (axis : ℕ) [_valid : Spec.Shape.HasNonemptyAxis axis s] [_wf : s.WellFormed] (t : Tape SomeBuffer) (xId : ℕ) :
            Result (Tape SomeBuffer × ℕ)

            Reduce-sum along axis.

            Instances For
              source
              @[inline]
              def Runtime.Autograd.LibTorch.Tape.reduceMean {s : Spec.Shape} (axis : ℕ) [valid : Spec.Shape.HasNonemptyAxis axis s] [_wf : s.WellFormed] (t : Tape SomeBuffer) (xId : ℕ) :
              Result (Tape SomeBuffer × ℕ)

              Reduce-mean along axis.

              Instances For