torch.js has not been released yet.

torch.IsAtLeast1D

export type IsAtLeast1D<S extends Shape> = S['length'] extends 0
  ? false
  : number extends S['length']
    ? true
    : true;

Sextends Shape

Type guard to check if a shape is at least 1-dimensional (not a scalar).

Returns true if the shape has at least one dimension, false for scalars (rank 0). Useful for constraining functions to operate on non-scalar tensors only.

Shape Constraint:

Scalars have shape [] (empty tuple) with rank 0 → returns false
Vectors, matrices, and higher-rank tensors → return true
Dynamic shapes (rank unknown) → return true (optimistic)

Use Cases:

Constraining generic functions to reject scalar inputs
Type-safe error messages when scalar operations fail
Guarding operations that require at least one dimension

Examples

// Scalar shape: rank 0
type scalarShape = readonly [];
type check1 = IsAtLeast1D<scalarShape>;  // false

// Vector shape: rank 1
type vectorShape = readonly [5];
type check2 = IsAtLeast1D<vectorShape>;  // true

// Matrix shape: rank 2
type matrixShape = readonly [3, 4];
type check3 = IsAtLeast1D<matrixShape>;  // true

// Dynamic shape (unknown rank)
type dynamicShape = readonly number[];
type check4 = IsAtLeast1D<dynamicShape>;  // true (optimistic)

// Use in function constraints
function flattenFirst<S extends Shape>(
  tensor: Tensor<S>,
  // Only accept if S is at least 1D
  ...args: IsAtLeast1D<S> extends true ? [options?: Options] : never
): Tensor {
  // Implementation
}