Function Differences with torch.nn.L1Loss

torch.nn.L1Loss

torch.nn.L1Loss(size_average=None, reduce=None, reduction='mean')(input, target) -> Tensor

For more information, see torch.nn.L1Loss.

mindspore.nn.L1Loss

mindspore.nn.L1Loss(reduction='mean')(logits, labels) -> Tensor

For more information, see mindspore.nn.L1Loss.

Differences

PyTorch: L1Loss is used to calculate the average absolute error between the predicted and target values.

MindSpore: Includes PyTorch function, which can still run when logits and labels have different shapes but can broadcast to each other, while PyTorch cannot.

Categories	Subcategories	PyTorch	MindSpore	Difference
Parameters	Parameter 1	size_average	-	Deprecated, function taken over by reduction
	Parameter 2	reduce	-	Deprecated, function taken over by reduction
	Parameter 3	reduction	reduction	-
Input	Input 1	input	logits	Same function, different parameter names
	Input 2	target	labels	Same function, different parameter names

Code Example

The two APIs achieve the same function and have the same usage.

# PyTorch
import torch
import torch.nn as nn

loss = nn.L1Loss()
input = torch.tensor([2, 2, 3], dtype=torch.float32)
target = torch.tensor([1, 2, 2], dtype=torch.float32)
output = loss(input, target)
output = output.detach().numpy()
print(output)
# 0.6666667

# MindSpore
import mindspore
from mindspore import Tensor, nn
import numpy as np

loss = nn.L1Loss()
logits = Tensor(np.array([2, 2, 3]), mindspore.float32)
labels = Tensor(np.array([1, 2, 2]), mindspore.float32)
output = loss(logits, labels)
print(output)
# 0.6666667