NuTorch

Loss ops

Generated from the binaries by scripts/gen-ops-reference.ts — do not edit by hand. Every op also documents itself: torch <op> --help.

mse_loss

mean squared error (—reduction mean|sum|none)

torch mse_loss <t1> <t2> [--reduction <Str>]

l1_loss

mean absolute error (—reduction)

torch l1_loss <t1> <t2> [--reduction <Str>]

smooth_l1_loss

smooth L1 loss (—beta, default 1.0)

torch smooth_l1_loss <t1> <t2> [--reduction <Str>] [--beta <Float>]

huber_loss

Huber loss (—delta, default 1.0)

torch huber_loss <t1> <t2> [--reduction <Str>] [--delta <Float>]

cross_entropy

cross entropy over logits vs int64 class indices

torch cross_entropy <t1> <t2> [--reduction <Str>]

nll_loss

negative log likelihood (log-prob inputs, int64 targets)

torch nll_loss <t1> <t2> [--reduction <Str>]

binary_cross_entropy

BCE over probabilities in [0,1]

torch binary_cross_entropy <t1> <t2> [--reduction <Str>]

binary_cross_entropy_with_logits

BCE over logits (the stable form)

torch binary_cross_entropy_with_logits <t1> <t2> [--reduction <Str>]

kl_div

KL divergence (—log_target if target is log-space)

torch kl_div <t1> <t2> [--reduction <Str>] [--log_target]