Loss.cross_entropy: def of bendtorch-core@0.1.0.0
Import: import bendtorch-core@0.1.0.0/src/autograd/loss.bend as Loss
Lines 415-426 of bendtorch-core@0.1.0.0/src/autograd/loss.bend: https://jend.leeeet.dev/read.txt?path=bendtorch-core@0.1.0.0/src/autograd/loss.bend&start_line=415&end_line=426

import Base
import bendtorch-core@0.1.0.0/src/tensor/core.bend as Core

# PyTorch's cross_entropy(logits, targets) with mean reduction, for logits
# of [rows, columns] and one class index below `columns` per row. Per row,
# the stable log_softmax: m = the row maximum (max_step), shifted = x - m,
# lse = log of the sum of exp(shifted) in ascending order from +0.0, and
# lp = shifted - lse, PyTorch's (x - max) - log(sum). The loss is the sum of
# -lp at each row's target, in row order from +0.0, divided by the rows.
# The logits come back unchanged, next to the loss, lp and the one-hot rows
# of the targets, which the backward reads. Logits that are not of rank 2
# fail with NotLogits (107); a target count other than the rows or a target
# at or past `columns` with TargetOutOfRange (108).
def cross_entropy(logits: Core.Dynamic, +targets: List<&2, U32>) -> Core.Dynamic & Result<&2, &1, Error, Recorded>:
  ce_shaped(targets, Core.shape(logits))

Uses (9):
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=List
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=Pair
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=Result
  https://jend.leeeet.dev/definition.txt?path=Base/base.bend&name=U32
  https://jend.leeeet.dev/definition.txt?path=bendtorch-core@0.1.0.0/src/autograd/loss.bend&name=Error
  https://jend.leeeet.dev/definition.txt?path=bendtorch-core@0.1.0.0/src/autograd/loss.bend&name=Recorded
  https://jend.leeeet.dev/definition.txt?path=bendtorch-core@0.1.0.0/src/autograd/loss.bend&name=ce_shaped
  https://jend.leeeet.dev/definition.txt?path=bendtorch-core@0.1.0.0/src/tensor/core.bend&name=Dynamic
  https://jend.leeeet.dev/definition.txt?path=bendtorch-core@0.1.0.0/src/tensor/core.bend&name=shape