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
moe_cast_window_bf16
def moe_cast_window_bf16[in_dtype: DType, num_cols: Int, cast_cols: Int, //, WindowEngine: TensorEngine, InputEngine: TensorEngine](ctx: DeviceContext, window: TileTensor[.bfloat16, Engine=WindowEngine, linear_idx_type=window.linear_idx_type], input: TileTensor[in_dtype, Engine=InputEngine, linear_idx_type=input.linear_idx_type])
Launches _moe_cast_window_bf16_kernel over input's rows.
Args:
- โctx (
DeviceContext): Device context. - โwindow (
TileTensor[.bfloat16, Engine=WindowEngine, linear_idx_type=window.linear_idx_type]): The[num_rows, cast_cols]bfloat16 output. - โinput (
TileTensor[in_dtype, Engine=InputEngine, linear_idx_type=input.linear_idx_type]): The producer row,num_cols + cast_cols(or more) wide.