torchsparse.nn.functional

relu(input: SparseTensor, inplace: bool = True) → SparseTensor

leaky_relu(input: SparseTensor, negative_slope: float = 0.1, inplace: bool = True) → SparseTensor

build_kernel_map(_coords: Tensor, kernel_size: Union[int, Tuple[int, ...]] = 2, stride: Union[int, Tuple[int, ...]] = 2, tensor_stride: Union[int, Tuple[int, ...]] = 1, mode='hashmap') → Tensor

conv3d(input: SparseTensor, weight: Tensor, kernel_size: Union[int, List[int], Tuple[int, ...]], bias: Optional[Tensor] = None, stride: Union[int, List[int], Tuple[int, ...]] = 1, dilation: Union[int, Tuple[int, ...]] = 1, transposed: bool = False, epsilon: float = 0.0, mm_thresh: int = 0, kmap_mode: str = 'hashmap') → SparseTensor

spcount(coords: Tensor, num: Tensor) → Tensor

spcrop(input: SparseTensor, coords_min: Optional[Tuple[int, ...]] = None, coords_max: Optional[Tuple[int, ...]] = None) → SparseTensor

spdevoxelize(feats: Tensor, coords: Tensor, weights: Tensor) → Tensor

calc_ti_weights(coords: Tensor, idx_query: Tensor, scale: float = 1) → Tensor

spdownsample(coords: Tensor, stride: Union[int, Tuple[int, ...]] = 2, kernel_size: Union[int, Tuple[int, ...]] = 2, tensor_stride: Union[int, Tuple[int, ...]] = 1) → Tensor

sphash(coords: Tensor, offsets: Optional[Tensor] = None) → Tensor

global_avg_pool(inputs: SparseTensor) → Tensor

global_max_pool(inputs: SparseTensor) → Tensor

sphashquery(queries: Tensor, references: Tensor) → Tensor

spvoxelize(feats: Tensor, coords: Tensor, counts: Tensor) → Tensor