VSATensor

class torchhd.VSATensor

Base class

Each model must implement the methods specified on this base class.

bind(other: VSATensor) → VSATensor

Bind the hypervector with other

bundle(other: VSATensor) → VSATensor

Bundle the hypervector with other

cosine_similarity(others: VSATensor) → Tensor

Cosine similarity with other hypervectors

dot_similarity(others: VSATensor) → Tensor

Inner product with other hypervectors

classmethod empty(num_vectors: int, dimensions: int, *, dtype=None, device=None) → VSATensor

Creates hypervectors representing empty sets

classmethod identity(num_vectors: int, dimensions: int, *, dtype=None, device=None) → VSATensor

Creates identity hypervectors for binding

inverse() → VSATensor

Inverse the hypervector for binding

multibind() → VSATensor

Bind multiple hypervectors

multibundle() → VSATensor

Bundle multiple hypervectors

negative() → VSATensor

Negate the hypervector for the bundling inverse

normalize() → VSATensor

Normalize the hypervector

permute(shifts: int = 1) → VSATensor

Permute the hypervector

classmethod random(num_vectors: int, dimensions: int, *, dtype=None, device=None, generator=None) → VSATensor

Creates random or uncorrelated hypervectors