For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /get-started.md).

Mojo function

tpool_patch_merger

def tpool_patch_merger[dtype: DType, output_layout: TensorLayout, x_layout: TensorLayout, bounds_layout: TensorLayout, OutputStorage: TensorStorage = PointerStorage, XStorage: TensorStorage = PointerStorage, BoundsStorage: TensorStorage = PointerStorage](output: TileTensor[dtype, output_layout, MutAnyOrigin, Storage=OutputStorage], x: TileTensor[dtype, x_layout, ImmutAnyOrigin, Storage=XStorage], bounds: TileTensor[DType.int64, bounds_layout, ImmutAnyOrigin, Storage=BoundsStorage], kH: Int, kW: Int, max_h: Int, max_w: Int, ctx: DeviceContext)

Temporal pooling patch merger entry point.

Parameters:

• ​dtype (DType): Element type of the input and output tensors.
• ​output_layout (TensorLayout): Memory layout of the output tensor.
• ​x_layout (TensorLayout): Memory layout of the input tensor.
• ​bounds_layout (TensorLayout): Memory layout of the bounds tensor.
• ​OutputStorage (TensorStorage): Storage policy of the output tensor.
• ​XStorage (TensorStorage): Storage policy of the input tensor.
• ​BoundsStorage (TensorStorage): Storage policy of the bounds tensor.

Args:

• ​output (TileTensor[dtype, output_layout, MutAnyOrigin, Storage=OutputStorage]): Contiguous output tensor [total_output_patches, D].
• ​x (TileTensor[dtype, x_layout, ImmutAnyOrigin, Storage=XStorage]): Input tensor [n_tokens, D].
• ​bounds (TileTensor[DType.int64, bounds_layout, ImmutAnyOrigin, Storage=BoundsStorage]): Grid dimensions tensor [n_vids, 3] with (T, H, W) per video.
• ​kH (Int): Merge kernel height.
• ​kW (Int): Merge kernel width.
• ​max_h (Int): Maximum H across all videos (for grid sizing).
• ​max_w (Int): Maximum W across all videos (for grid sizing).
• ​ctx (DeviceContext): Device context.