PyTorch torch.unique Function
Pytorch torch Reference Manual
torch.uniqueIt is a function in PyTorch used to get unique values. It returns the unique values in the tensor and can optionally return the corresponding indices.
Function Definition
torch.unique(input, sorted=True, return_inverse=False, return_counts=False, dim=None)
Usage Examples
Examples
import torch
# Create a tensor
x = torch.tensor([1, 2, 2, 3, 1, 4, 3])
# Get unique values
unique_vals = torch.unique(x)
print(f"Unique values: {unique_vals}")
# Also return indices
unique_vals, inverse_idx = torch.unique(x, return_inverse=True)
print(f"Inverse indices: {inverse_idx}")
# Also return counts
unique_vals, counts = torch.unique(x, return_counts=True)
print(f"Counts: {counts}")
# Create a tensor
x = torch.tensor([1, 2, 2, 3, 1, 4, 3])
# Get unique values
unique_vals = torch.unique(x)
print(f"Unique values: {unique_vals}")
# Also return indices
unique_vals, inverse_idx = torch.unique(x, return_inverse=True)
print(f"Inverse indices: {inverse_idx}")
# Also return counts
unique_vals, counts = torch.unique(x, return_counts=True)
print(f"Counts: {counts}")
Other Extensions